You are on page 1of 77

How Firms Make Friends:

Communities in Private-Collective Innovation

Matthias Stuermer
ETH Zürich
Doctoral Dissertation No. 18630
Diss. ETH No 18630

How Firms Make Friends:


Communities in Private-Collective Innovation

A dissertation submitted to

ETH ZURICH

for the degree of

Doctor of Sciences

presented by

Matthias Emmanuel Stuermer

lic.rer.pol. University of Bern

Date of birth February 20th 1980 in Wattenwil, Bern

Citizen of Pfaeffikon ZH, Switzerland and Germany

Accepted on the recommendation of

Prof. Dr. Georg von Krogh

Prof. Dr. Sonali Shah

2009
Acknowledgements
Writing a doctoral dissertation is on the one hand a challenge which has to be accomplished as a
lone fighter with endurance and discipline. On the other hand, I have to state clearly that this thesis
would not have been possible without the support and inspiration of many friends and relatives.
First of all I want to thank Prof. Georg von Krogh for giving me the opportunity to get introduced to
the fascinating profession of scientific research. It was his advice, his guidance, and his skills which
have enabled me to conduct many great studies during the last three years. Not only did I learn
tremendously from his strong academic experience and his creativity of tackling challenging re-
search issues. In addition, his authentic way of motivating and inspiring people has taught me im-
portant social skills. As a role model in many areas he has shaped my way of thinking fundamental-
ly leading me to hope that our friendship will be long-lasting.
My visit at the University of Washington in Seattle was short but very intense. Through the kind-
ness and hospitality of Prof. Sonali Shah I received insight into a culturally different research envi-
ronment. Our extensive conversations on various aspects of open source communities have ad-
vanced my understanding significantly. I am particularly grateful for her expertise in assessing my
doctoral dissertation as co-referee.
Another key source of scientific know-how was provided by my team colleagues Dr. Sebastian
Spaeth and Dr. Stefan Haefliger, both co-authors of all papers I was involved in. Their in-depth
comprehension of theory and empiricism was crucial in getting acquainted with the day-to-day life
of a social scientist. They integrated me open-heartedly into their well attuned team and answered
even my most basic questions with patience. This was of great value for my academic experience
and the efficient progress of my doctoral thesis.
Especially the great team culture at the Chair of Strategic Management and Innovation has been a
unique motivator for my work. Through many talks in the office, at conferences, and team retreats I
became inspired with new ideas and research questions and learned essential elements of scientific
research. I would like to thank all of its members not yet mentioned: Dr. Martin Wallin, Peter
Jaeger, Zeynep Erden, Jan Henrik Sieg, Renato Sydler, Alban Fischer, and Lise Rechsteiner. Also
thanks to our very supportive secretary Hildgard Brune.
Many more friends have accompanied me during my time as a research assistant substantially influ-
encing the results of my thesis. Regular meetings with Marcus Dapp of ETH Zurich have chal-
lenged my work in a highly constructive way thus helping me to advance in the right direction. In
addition I received important inspiration from the practitioner's side through the interaction with
various members of the Swiss Open Systems User Group /ch/open, particularly with Dr. Matthias
Guenter of the Institute of Intellectual Property, Hannes Gassert of Liip AG, and Sven Leser of
SyGroup GmbH. Staying in close contact with experienced open source experts has enabled me to
understand the dynamics of this industry and filter out relevant research questions. Much of this
knowledge I gained from the roughly 40 in-depth interviews with community members and man-
agers whom I would like to thank for their valuable time and openness.
Also I would like to thank my father, Dr. Ekkehard Stuermer, for his great interest and support of
my research. And last but not least I am deeply grateful to my family, my wife Anita, and my boys
Lionel and Kai, who supported me despite my sometimes erratic working style during the last three
years.
Matthias Stuermer, Bern, June 2009
Abstract
When firms contribute to open source projects, they in fact invest into public goods which may be
used by everyone, even by their competitors. This seemingly paradoxical behavior can be explained
by the model of private-collective innovation where private investors participate in collective ac-
tion. Previous literature has shown that companies benefit through the production process providing
them with unique incentives such as learning and reputation effects. By contributing to open source
projects firms are able to build a network of external individuals and organizations participating in
the creation and development of the software. As will be shown in this doctoral dissertation firm-
sponsored communities involve the formation of interorganizational relationships which eventually
may lead to a source of sustained competitive advantage. However, managing a largely independent
open source community is a challenging balancing act between exertion of control to appropriate
value creation, and openness in order to gain and preserve credibility and motivate external contri-
butions. Therefore, this dissertation consisting of an introductory chapter and three separate re-
search papers analyzes characteristics of firm-driven open source communities, finds reasons why
and mechanisms by which companies facilitate the creation of such networks, and shows how firms
can benefit most from their communities.

Zusammenfassung
Wenn Unternehmen Beiträge an Open Source Projekte leisten, dann investieren sie in öffentliche
Güter, die alle nutzen können – selbst die Konkurrenten. Dieses auf den ersten Blick paradoxe
Verhalten wird durch das Private-Collective Model of Innovation beschrieben, das die Teilnahme
von Privaten an kollektiven Innovationsprozessen erklärt. Die bisherige Forschung hat gezeigt, dass
Unternehmen Vorteile aus dem Prozess der Software-Entwicklung ziehen und von Anreizen wie
Lerneffekten und verbessertem Ansehen profitieren. Firmen können mittels Beiträgen an Open
Source Projekte ein Netzwerk von Personen und Organisationen aufbauen, die bei der Schaffung
und Weiterentwicklung mitwirken. In dieser Doktorarbeit wird gezeigt, wie von Unternehmen fi-
nanzierte Communities ein interorganisationales Beziehungsnetz darstellen, das zu einem dau-
erhaften Wettbewerbsvorteil führen kann. Die Leitung einer weitgehend unabhängigen Open Source
Community bleibt jedoch eine herausfordernde Gratwanderung zwischen der notwendigen Kon-
trolle, um die Wertschöpfung sicherstellen zu können und der unverzichtbaren Offenheit, um die
Glaubwürdigkeit der Firma sowie die Motivation der extern Beitragenden zu bewahren. Die
Dissertation besteht aus einem Einführungskapitel und drei unabhängigen Forschungspublikationen.
Unter anderem werden die Eigenschaften von Firmen-eigenen Open Source Communities darge-
stellt, Ursachen und Mechanismen der effizienten Schaffung solcher Netzwerke durch Firmen auf-
gezeigt sowie erläutert, wie Firmen den grösstmöglichen Nutzen aus ihren Communities ziehen
können.
Table of Contents

1. The Balancing Act of Community Management.........................................................................1

2. Communities of Open Source Projects.........................................................................................2


2.1. Motivation
2.2. Governance
2.3. Competitive Dynamics
2.4. Characterizing Firm-Sponsored Open Source Projects
2.5. The Private-Collective Model of Innovation

3. Research Framework.....................................................................................................................6
3.1. Filling Gaps in the Literature
3.2. Extending the Open Innovation Paradigm
3.3. Nokia's Entrance Into the Open Source Industry
3.4. Antecedents of Intrinsic and Extrinsic Motivation

4. Three Cases of Firm-Sponsored Open Source Communities.....................................................7


4.1. Eclipse by IBM
4.2. Maemo by Nokia
4.3. Openmoko by Openmoko Inc.

5. Conclusions and Implications......................................................................................................10


5.1. The Concept of Interorganizational Competitive Advantage:
Towards a Related Research Agenda
5.2. Impact on Theory and Practice
5.3. Future Research Agenda

References..........................................................................................................................................15

Appendix 1
Enabling Knowledge Creation through Outsiders:
Towards a Push Model of Open Innovation...................................................................................18

Appendix 2
Extending Private-Collective Innovation:
A Case Study.....................................................................................................................................32

Appendix 3
The Credible Sponsor: Participants’ Motivation
and Organization Attributes in Collaborative Digital Innovation...............................................50

CV Matthias Stürmer.......................................................................................................................67
How Firms Make Friends:
Communities in Private-Collective Innovation

Introductory chapter of the doctoral dissertation by

Matthias Stuermer

Chair of Strategic Management and Innovation


Department of Management, Technology, and Economics
ETH Zurich, Switzerland
June 2009

When firms contribute to open source projects, they in fact invest into public goods which may be
used by everyone, even by their competitors. This seemingly paradoxical behavior is explained by
the model of private-collective innovation where private investors participate in collective action.
Previous literature explains that companies benefit through the production process providing them
with unique incentives such as learning and reputation effects. By contributing to such open source
projects firms are able to build a network of external individuals and organizations, who may partic-
ipate in the creation and development of the software. As will be shown in this doctoral dissertation
firm-sponsored communities involve the formation of interorganizational relationships which even-
tually may lead to a source of sustained competitive advantage. However, managing a largely inde-
pendent open source community is a challenging balancing act between exertion of control to ap-
propriate value creation, and openness in order to gain and preserve credibility and motivate exter-
nal contributions. Therefore, this dissertation consisting of an introductory chapter and three sepa-
rate research papers analyzes characteristics of firm-driven open source communities, finds reasons
why and mechanisms by which companies facilitate the creation of such networks, and shows how
firms can benefit most from their communities.

1. The Balancing Act of Community Management 2006). Small as well as large software companies re-
lease previously proprietary software under open
Creation and appropriation of value are the two source licenses, they employ engineers who program
core elements of every firm's business model. There- code for publicly available open source software, and
fore a company usually protects its assets in order to they sponsor the creation and maintenance of volun-
appropriate their value exclusively. Since knowledge teer and partner firm communities. Obviously, mak-
represents one of the key sources of competitive ad- ing friends seems to be an attractive option for many
vantage (Grant, 1996), there exist various ways for a firms to support their innovation activities.
firm to protect its intellectual property (Liebeskind, It is known that community building is one of the
1996). Surprisingly, empirical research on open major goals in all open source projects (Stuermer,
source software has shown that in certain cases firms 2005). However, if a profit-oriented organization in-
do not conceal their knowledge but actively reveal it tends to create an innovating community, this task be-
publicly (Lerner and Tirole, 2002; Lee and Cole, comes particularly challenging. On the one hand, vol-
2003; Henkel, 2006). The behavior of investing in unteers may become suspicious if a firm invests in a
public goods for the purpose of increasing private public good because the main objective of a for-profit
profits has been coined as the private-collective mod- company is to maximize corporate profits (Bae and
el of innovation (von Hippel and von Krogh, 2003; Cameron, 2006). On the other hand, while engineers

How Firms Make Friends: Communities in Private-Collective Innovation 1


within a company follow hierarchical control, com- findings positioned as example of the relational view.
munities outside the boundaries of the firm act inde- Each of the three dissertation papers in the appendix
pendently based on their individual motivations elaborates a different perspective on firm-sponsored
(Dahlander and Wallin, 2006). Therefore, firms have open source projects. The first paper “Enabling
to find other ways than direct control in order to in- Knowledge Creation through Outsiders: Towards a
duce contributions to their open source projects. Push Model of Open Innovation” provides insight on
Dahlander and Magnusson (2005) argue that firms open source development by IBM pointing out
should follows a symbiotic approach. By giving knowledge creation deficiencies in the current con-
knowledge assets to the community and respecting its ception of open innovation. The second paper “Ex-
norms and values, firms are able to exert means of tending private-collective innovation: a case study”
subtle control in their interaction with the community. presents an in-depth qualitative analysis of a corpo-
Nevertheless, successfully building and maintain- rate sponsored open source project by Nokia explain-
ing a thriving community remains a challenging task ing benefits and costs of this distributed type of inno-
for a firm. Academic researchers as well as practi- vation initiative. And the third paper “The credible
tioners are thus interested to better understand the sponsor: Participants’ motivation and organization at-
characteristics of corporate-sponsored open source tributes in collaborative digital innovation” looks at
communities and find the mechanisms how firms are the sponsoring effect on individual's motivation and
able to facilitate the creation of such networks. Re- on their performance.
search so far has identified firm-driven communities
as valuable assets for companies (Dahlander and 2. Communities of Open Source Projects
Wallin, 2006; West and O'Mahony, 2008), but has
omitted in-depth empirical analyses on why and how Research on open source projects is attractive be-
companies benefit from their communities. There- cause of their impact on economy and society, the
fore, this thesis takes up the concept of private-col- theoretical puzzles they pose, the availability of data,
lective innovation, extends its perception of firm- the reflexivity of their communities, and the parallels
driven communities, and suggests an integration with with science (von Krogh and Spaeth, 2007). The cur-
an existing model in the strategic management litera- rent literature on open source software research can
ture. be grouped into three different areas (von Krogh and
As has been explained, an active open source von Hippel, 2006): Motivation, governance, and
community of a firm is a resource difficult to imitate. competitive dynamics. In the following, an overview
Therefore once a company has reached the position of research in these three areas is presented with em-
of an influential participant within a community and phasis on firm-driven open source projects and their
can appropriate value from the community's innova- communities. Then, the concepts of community- and
tions, it has gained so-called interorganizational com- firm-driven projects will be contrasted followed by
petitive advantage. This concept by Dyer and Singh an introduction to the private-collective model of in-
(1998) constitutes an alternative to the industry struc- novation.
ture view by Porter (1980) and the resource-based
view by Wernerfelt (1984) and Barney (1991) and is 2.1. Motivation
also called the 'relational view'. It explains why inter-
The first stream of literature on open source soft-
firm resources and routines embedded in a network
ware treats the issue of motivation. Discovering the
of relationships represent a sustainable competitive
underlying cause for contributing to open source
advantage. As will be shown in the concluding sec-
projects has been one of the most puzzling mysteries
tion, firm-sponsored open source communities in pri-
for social scientists. A single motivational factor has
vate-collective innovation represent a perfect exam-
not been found. However, a thorough review of cur-
ple of such interorganizational competitive advan-
rent research on open source projects shows that the
tage.
motivation among participants is highly diverse. Von
The present doctoral thesis consists of an opening
Krogh and colleagues (2009) found ten different in-
chapter and three independent research papers in the
centives for participation in open source projects,
appendix co-authored by the writer of this thesis.
namely ideology, altruism, kinship amity, enjoyment,
This opening chapter presents a literature overview
reputation, reciprocity, learning, own-use value, ca-
on open source communities, the research framework
reer, and pay. These motives, categorized into intrin-
of the three pivotal papers, a summary of three firm-
sic, internalized extrinsic, and extrinsic motivations,
sponsored projects, and a synthesis of the empirical
have been studied widely by a multitude of re-

How Firms Make Friends: Communities in Private-Collective Innovation 2


searchers (Lerner and Tirole, 2002; Hars and Ou, control, ownership, and the opportunity to utilize the
2002; Lakhani and von Hippel, 2003; Hertel et al., benefits of their investments. In this way governance
2003; Roberts et al., 2006; Steward and Gosain, of an open source project resolves the collective ac-
2006; Wu et al., 2007). tion dilemma by mechanisms such as open source li-
While programmers in the early days of open censes (Franck and Jungwirth, 2003; Lee and Cole,
source development were mostly driven by intrinsic 2003; Henkel, 2006) and non-profit foundations
and internalized extrinsic motives, the importance of (O'Mahony, 2003; O'Mahony and Bechky, 2008). An-
extrinsic incentives has increased in recent years other perspective on overcoming the collective action
(O'Mahony, 2007). One example is the development problem is presented in the private-collective model
of Linux. When Linus Torvalds and colleagues start- of innovation which will be explained below (von
ed to program their operating system, they were Hippel and von Krogh, 2003).
mostly driven by fun and other intrinsic motives Governance substantially differs between commu-
spending much of their spare time on their computer nity-driven and firm-managed open source projects.
hobby (Raymond, 1999). Today, however, at least 73 In a community-initiated project, governance struc-
percent of contributions come from employees of tures evolve bottom-up through a meritocratic
software corporations (Red Hat, Novell, IBM, Intel process providing those authority who engaged in
etc.) or the foundation itself, as reported by the Linux technical contributions and organizational-building
Foundation in a recent study (Kroah-Hartman et al., behavior (O'Mahony and Ferraro, 2007). If a hetero-
2008). This shows that even in initially noncommer- geneous community controls an open source project,
cial open source projects the influence of firms may principles such as independence and decentralized
rise and new motivational structures can become im- decision-making prevail (O'Mahony, 2007). On the
portant. When extrinsic incentives are introduced in other hand, if firms found an open source project, its
intrinsically motivated communities, crowding-out governance structure has to serve a different goal.
effects may occur leading to an overall decrease of Since firms can exist in the long run only if they are
activity (Frey and Oberholzer-Gee, 1997; Frey and profitable, appropriation regimes are required to cap-
Jegen, 2001; Alexy and Leitner, 2008). And when ture the profits generated by innovations (Teece,
software companies start an open source project 1986). Therefore, companies which invest into the
themselves still other incentives are important to at- creation of open source code must somehow govern
tract external contributions (Dahlander and Magnus- their project in a way that allows them to capture the
son, 2005; von Krogh and von Hippel, 2006). One of created value. This suggests that firms need control
the dissertation papers (Stuermer et al., 2009) treats of critical aspects of the innovation process in order
the case where Nokia initiated an own-branded open to appropriate the returns of their investment
source platform and built a community for it. Bene- (Dahlander and Wallin, 2006). Owners of open
fits and costs of this effort are analyzed and best prac- source projects typically have several options to open
tices of Nokia are elaborated. up or restrict access to the software and the develop-
ment process. Shah (2006) coined the term “gated
2.2. Governance communities”. It implies that the project initiator
controls the permeability of the community, e.g. by
The second area of open source research is devot- imposing restrictive property rights of the software or
ed to governance. Governing organizational and pro- by excluding non-corporate members from code inte-
duction processes in open source projects is related to gration and decision-making. As Shah noted, value
motivational issues, but involves also challenges of appropriation may sometimes negatively affect value
its own. Markus (2007) defines governance of open creation by the community. She observed that contri-
source projects as “the means of achieving the direc- butions were low in a tightly controlled environment
tion, control, and coordination of wholly or partially like the gated community but participation was lively
autonomous individuals and organizations on behalf in a broadly accessible open source project governed
of an OSS development project to which they jointly by a heterogeneous community. Also West and O'Ma-
contribute.” Governance of an open source project hony (2008) indicated in their analysis of sponsored
thus includes the management of a more or less inde- open source communities that firms managing an
pendent community by a few individuals who are in open source project need to balance the level of con-
charge of control. Markus (2007) states that gover- trol and opportunities for outside contributors in or-
nance in fact solves the motivational problem by em- der to create a sustainable participation architecture.
powering open source contributors and giving them Dahlander and Magnusson (2005) explained why this

How Firms Make Friends: Communities in Private-Collective Innovation 3


balancing act of managing a community is a key is- Nokia is elaborated where partial revealing of knowl-
sue: With too much control communities may not edge also plays a key role (Stuermer et al., 2009).
contribute with all of their energy, interest, and cre- Intuitively, imitation by competitors may appear
ativity. With too little control the results may not to be a great threat for firms investing in open source
serve the firm's goals. Therefore they suggest that projects. However, the mere use of open source soft-
firms should seek a symbiotic relationship with their ware does not destroy the firm's competitive advan-
communities in order to stimulate contributions while tage since the open source license of the technology
retaining control by subtle means. Interaction be- already has turned it into a commodity (West, 2003).
tween voluntary communities and software firms It is the creation and maintenance of a symbiotic
have allowed O'Mahony and Bechky (2008) to ob- community which may become the source of sustain-
serve the conciliation of divergent interests through able competitive advantage (Dahlander and Magnus-
collaboration in non-profit foundations. Still, only son, 2008). The capability to manage an external
few research has quantitatively explored the impact community enables a firm to access its knowledge
of firm control on communities. As one of the disser- and quickly integrate it into new products. By extend-
tation papers has found (von Krogh et al., 2009a), ing this core idea of open innovation (Chesbrough,
transparency of the project's management and the 2003), one of the dissertation papers explains how
reputation of the firm positively influence contribu- IBM was able to create and maintain a broad commu-
tors' efforts. Accessibility for the open source com- nity for its open source project Eclipse (Spaeth et al.,
munity, however, has only a weakly positive impact. 2009a). Already during the first years after the release
of the platform, programmers not employed by IBM
2.3. Competitive Dynamics started offering contributions. The external communi-
ty has been growing continuously ever since, forming
Thirdly, analyzing open source projects raises an about half of the Eclipse developers' population in
intriguing question on competitive dynamics: Why do 2007, the time of the study. This example shows that
firms give away for free valuable investments in the investing in open source projects and building an ac-
form of source code? The fact that this happens im- tive community can create a valuable resource diffi-
plies that firms benefit in a certain way when they cult to imitate by competitors. Von Krogh (2002) has
freely reveal some of their technologies. Already in defined communities as resources of firms but has
the beginning of the nineties Garud and Ku- also pointed out the dilemma that once a community
maraswamy (1993) noted the success of Sun Mi- displays a high level of voluntary action and self-or-
crosystem's knowledge revealing strategy by being ganization, the ability to control it from the outside
the first to break proprietary barriers. On the one decreases.
hand the company benefited from network externali-
ties of its products which diffused at a higher rate
2.4. Characterizing Firm-Sponsored Open Source
than their competitor's solutions. On the other hand Projects
sponsoring open technologies allowed Sun to be the
first to implement it in new products thus gaining The previous sections distinguish between two
temporal competitive advantage. From an economic types of open source projects: community-driven and
point of view, free revealing of innovations benefits firm-managed initiatives. Although this distinction
firms in various ways, among others through reputa- has been made before in the research on open source
tion gain (Harhoff et al., 2003) and better opportuni- communities by West and O'Mahony (2005 and
ties for R&D partnerships (Muller and Pénin, 2005) 2008), the definition of a firm-sponsored open source
as well as increased technology diffusion (Pénin, project has remained vague or too simplified. A more
2007) and standard setting (Bonaccorsi and Rossi, precise description is necessary because several di-
2003; West and Gallagher, 2006). A common value mensions have to be taken into account in order to
appropriation strategy for open source software in- position a community. O'Mahony (2007) lists five
volves selective revealing of knowledge. Henkel principles of a community-managed open source
(2006) found in his study on embedded Linux firms project which determine its governance structure: in-
that they revealed about half of their code while pro- dependence, pluralism, permeable representation, de-
tecting the other half. Such selective revealing strate- centralized decision-making, and autonomous partici-
gies have been common practice in the IT industry as pation. Therefore, a firm-driven open source project
cases of Apple and Sun Microsystems showed (West, may be defined by the opposite of these characteris-
2003). In one of the dissertation papers the case of tics: dependence on a single sponsor, dominance of
one company, undisputed control by one sponsor,

How Firms Make Friends: Communities in Private-Collective Innovation 4


centralized decision-making by the company's man- Usually companies are free to choose up to which
agement, and strictly restricted participation. level they want to adopt the open source development
While both extreme types of communities do exist model. While some firms such as Motorola have de-
in the open source universe, many intermediary cided to use open source components without deliber-
forms prevail. As has happened in the subsequently ately revealing own developments nor intending to
described case of Eclipse, the initiating firm may for attract outside contributors, other companies such as
example relinquish control by creating an indepen- Nokia or IBM have chosen to publish open source
dent foundation and assigning it with the leadership code and also build an external developer community
of the open source project. Or decision-making can (Stuermer et al., 2009; Spaeth et al., 2009a). Why did
be substantially influenced by the will of the commu- they reveal more knowledge than they were obliged
nity as was the case with the Openmoko project. This to by the software license? The model of private-col-
shows that although a firm-initiated open source lective innovation provides an explanation.
project usually starts off at the extreme position, it
may shift towards a more accessible, community- 2.5. The Private-Collective Model of Innovation
managed structure if the company allows it to.
The private-collective model of innovation repre-
sents a combination of the private investment model
and the collective-action innovation model (von Hip-
pel and von Krogh, 2003). In the private investment
model innovators appropriate financial returns from
Level 3:
innovations through intellectual property rights such
Community Building as patents, copyright, licenses, or trade secrets within
Building of a firm-sponsored the legal framework imposed by society. Any knowl-
open source community edge spillover will reduce the innovator's benefits.
Thus freely revealed knowledge is not in the interest
Level 2:
Knowledge Revealing of the innovator.
Revealing of proprietary source The collective-action innovation model explains
code under an open source license the creation of public goods which are defined by the
non-rivalry of benefits and non-excludable access to
Level 1: Open Innovation the good. In this case the innovators do not benefit
Integration of externally available more than anyone else not investing into the public
open source components
good, thus free-riding may occur. In response to this
problem, the cost of innovation has to be distributed.
Figure 1: Adoption levels of the open source develop- Therefore governments typically invest into public
ment model goods through public funding.
Combining these two concepts the private-collec-
In this context O'Mahony (2007) has introduced tive model of innovation explains the creation of
the concept of primarily independent elements of the public goods through private funding. The model is
open source development model. According to this based on the assumption that an innovator privately
analysis firms can determine their adoption level of creating a public good will benefit more than a free-
the open source model by choosing their position on rider just consuming the public good. While the result
one of three basic levels (see Figure 1): The first level of the investment is equally available to both, the in-
involves integration of externally available knowl- novator benefits from the very process of creating the
edge which is better known as 'open innovation' in public good (Spaeth et al., 2008). In the case of open
the sense of Chesbrough (2003). Alternatively, a source software production, the programming code as
company directing an open source project can em- explicit knowledge is freely revealed and accessible
ploy legal elements by freely revealing source code to all, even to competitors. Nonetheless, the innova-
under an open source license (von Hippel and von tor gains tacit knowledge and expertise through the
Krogh, 2006). The control of the development creation process representing an advantage which is
process, however, remains fully governed by the difficult to imitate (Grand et al., 2004). Therefore,
firm. Finally, to apply the community element, too, private-collective innovation occurs on a sustainable
the firm has to renounce some of the project's gover- basis when the process-related rewards exceed the
nance in order to create a prospering community process-related costs (von Krogh, 2008).
(Shah, 2006).

How Firms Make Friends: Communities in Private-Collective Innovation 5


Typical benefits of firms investing in public goods find options for organizational interventions in order
have been analyzed in open source development (von to increase the contributions of external communities.
Hippel and von Krogh, 2006). These benefits include The strength of these independent research projects
low knowledge protection costs, learning effects for lies in the diverse selection of methods (longitudinal
the firm, reputation gain, high diffusion level of inno- data, grounded theory building, and structured equa-
vations, creation of external innovations, lower man- tion modeling) and dataset (quantitative archival data,
ufacturing costs, and faster time-to-market (Stuermer expert interviews, and online survey) providing a
et al., 2009). However, as the case study on Nokia holistic view on firm-sponsored open source projects.
shows private-collective innovation activities also in- Each study uses a unique set of data and applies a dif-
volve costs such as loss of differentiation from com- ferent method of analysis as will be described subse-
petitors, leakage of business secrets, investments into quently.
community creation and maintenance, loss of control,
and efficiency loss caused by organizational changes. 3.2. Extending the Open Innovation Paradigm

3. Research Framework The first paper “Enabling Knowledge Creation


through Outsiders: Towards a Push Model of Open
This dissertation seeks to shed light on two issues Innovation” (Spaeth et al., 2009a) presents a perspec-
related to community building and firm-sponsored tive on the deficiencies of the current concept of open
open source projects. First, what are the main bene- innovation in explaining external knowledge cre-
fits and costs for firms conducting private-collective ation. Chesbrough (2003) defines open innovation as
innovation by investing in open source projects? And adding external knowledge to internal research and
second, since the collaboration of firms with external development processes and selling unused internal
communities affects their incentive structures, how innovations on external markets. However, this does
are the contributions influenced by the involvement not explain why and how the outside knowledge was
of commercially oriented technology companies? produced in the first place. The extended model of
These questions point to gaps in the literature where open innovation therefore attempts to describe how
previous research has not yet found answers. firms may facilitate the creation of product-related
knowledge receiving it 'pushed-back' into its internal
3.1. Filling Gaps in the Literature innovation processes instead of having to 'pull-back'
this knowledge themselves.
In earlier studies economists have conceptualized The study illustrates IBM's creation of and contin-
possible benefits derived from knowledge revealing ued investments in the open source platform Eclipse.
by firms (Harhoff, 1996; Harhoff et al., 2003; Muller Based on quantitative archival data of communica-
and Pénin, 2005; Pénin, 2007). Analyzing the options tion and production processes and comparing the
and consequences of a firm's openness in its research contributions of IBM versus non-IBM individuals,
and development activities has provided interesting the ratio of knowledge revealing versus knowledge
theoretical insights. However, while the models reception is calculated. The longitudinal analysis
promise attractive returns from investments in pub- shows that during a six year period around 70 IBM-
licly available knowledge resources, the applicability employed developers were constantly programming
of the theory has yet to be proved. for Eclipse. While in the beginning external participa-
By reviewing empirical research in various indus- tion was low, it increased up to a 100 active contribu-
tries and analyzing the case of the open source devel- tors after IBM gave up much of its control of the
opment model in depth, von Hippel and von Krogh open source project. Details on the case are elaborat-
(2006) have found that firms in the past have aban- ed in the next section.
doned monopoly positions by disclosing innovations.
This unintuitive behavior was explained by the un- 3.3. Nokia's Entrance Into the Open Source Industry
derlying benefits firms gain during the knowledge re-
vealing process. Yet, while the conceptual study per- The second paper “Extending private-collective
fectly integrates the models of private-collective in- innovation: a case study” (Stuermer et al., 2009) con-
novations, no new evidence is presented in order to ducts an in-depth qualitative analysis of a corporate-
prove the validity of the arguments. sponsored open source project. The case of Nokia's
Therefore, this dissertation presents three pivotal mobile device platform Maemo represents a success-
studies which empirically analyze the benefit and ful new product development process based on an
cost structure of private-collective innovation and open source technology stack. Through integration of

How Firms Make Friends: Communities in Private-Collective Innovation 6


external innovations Nokia was able to create a new 4. Three Cases of Firm-Sponsored Open Source
mobile device very quickly. To this end, Nokia in- Communities
vested into the building of a user and developer com-
munity. These and other activities of private-collec- In the following section the three firm-sponsored
tive innovation created costs unrecognized until then. open source communities will be characterized. They
Therefore, the paper presents an analysis of benefits were analyzed qualitatively and quantitatively in the
and costs of this distributed type of innovation. dissertation papers in the appendix representing the
To provide an empirical basis 23 in-depth expert empirical evidence of this thesis. In the following
interviews with Nokia managers and engineers, con- case summaries emphasis is placed on relevant inter-
tractors, and voluntary community members were dependencies between individuals, communities, oth-
conducted leading to around 250 pages of transcripts. er stakeholders, and the firm which initiated the open
The semi-structured interview guideline was de- source project. Insight shall be gained in the different
signed to find causal relationships between the firm's reasons of technology companies to found and con-
activities and their consequences. The questions treat- tinuously sponsor their own open source project and
ed issues like e.g. knowledge revealing strategies, how they manage issues such as governance, control
community activity, and organizational issues such as and motivation of developers. First, the case of
recruiting or forms of collaborations. With the sup- Eclipse, the software development environment
port of a computer-aided qualitative data analysis founded by IBM, is portrayed. Next, Nokia's open
software the transcribed interviews were consolidated source community Maemo is described. Finally, the
into 12 constructs following the grounded theory history and background of the mobile phone project
building method (Glaser and Strauss, 1967). Openmoko is presented.

3.4. Antecedents of Intrinsic and Extrinsic Motivation 4.1. Eclipse by IBM

The third paper “The credible sponsor: Partici- On February 22, 1996 IBM announced the acqui-
pants’ motivation and organization attributes in col- sition of Object Technology International Inc. (OTI),
laborative digital innovation” looks at the sponsoring a leading development firm of object-oriented soft-
effect on the contributers' motivation and on their ware technology (Business Wire, 1996). OTI re-
performance. While previous studies have concentrat- mained an autonomous subsidiary of IBM Canada
ed on the causes of motivation within community- employing about 100 highly skilled developers
managed open source projects (Hertel et al., 2003; (O'Mahony et al., 2005). At that time IBM was look-
Stewart and Gosain, 2006; Roberts et al., 2006), this ing for a common technology platform to unite their
research project analyzes the effect of perceived firm various software solutions using Java as a program-
characteristics on the intrinsic and extrinsic motiva- ming language. The OTI team was appointed to de-
tion of voluntary contributors. velop this key platform for IBM and other corporate
In order to assess the significance and impact of partners so they could create complementary prod-
the attributes of firm perception (knowledge reveal- ucts with it. In 2000 the platform was named 'Eclipse'
ing, accessibility, and corporate credibility), a survey and gained internal adoption throughout IBM. Since
was sent to the two independent communities of IBM's strategy was to make money not from client or
Maemo and Openmoko. Within two weeks 1233 indi- development software but from server software (e.g.
viduals filled out the questionnaire corresponding to a WebSphere), they decided to release Eclipse as open
response rate of 27.9 percent. The items were source software in order to gain the maximum market
grouped into constructs and then analyzed by a struc- share in software development tools (West, 2003).
tured equation model. The results show that the firm's With this move they convinced partner firms to use
credibility has the highest impact on the intrinsic mo- and extend Eclipse because they did not have to fear
tivation of contributors while knowledge revealing to be locked into a proprietary platform.
and accessibility were significant only on a low im- The management of IBM was aware that starting
pact level. Counterintuitively the model shows extrin- an open source project involved more than publishing
sic motivation to exert a slightly higher positive ef- the source code on some website. According to
fect on contribution level than intrinsic motivation. O'Mahony and colleagues (2005) the Eclipse team
adapted its processes and practiced for several
months to collaborate in an open source development
mode e.g. by setting up mailing lists and newsgroups
and communicating internally in a transparent way.

How Firms Make Friends: Communities in Private-Collective Innovation 7


The Eclipse managers also wanted to attract corpo- category of Internet tablets. The basic idea was to
rate partners. Thus they invited other software firms provide customers with a convenient device to
to join the board of stewards governing the future di- browse the web, receive emails and communicate via
rection of the open source project. The members were chat and voice-over-IP. Nokia decided to use Linux
granted equal decision rights in the consortium but as a platform for this innovative product because the
IBM remained the legal owner of Eclipse. software provided high flexibility and independence
In November 2001 the public responded positive- from other firms in order to explore the yet unknown
ly to the announcement of IBM releasing its 40 mil- consumer requirements of the new device category.
lion dollar investment Eclipse as an open source Inspired by this vision, Nokia designed an overall
project (see e.g. commentary by Gartner analysts software architecture of the operating system Maemo
Feiman and Driver, 2001). During the following based on open source components. Partly the engi-
months many more companies joined the consortium neers adapted these software components themselves
and started building extensions for the platform. and partly they contracted external open source de-
Eclipse evolved as a typical sponsor-driven open velopers for specific implementation tasks. After hav-
source project including public source code and pub- ing signed non-disclosure agreements these program-
lic mailing lists. However, the effective role of IBM mers were assigned to adapt existing open source
remained unclear because consortium members as components to the hardware requirements of the mo-
well as the public perceived Eclipse as still mainly bile device. Through the strategy of hiring core de-
controlled by IBM. velopers well acquainted with open source technolo-
In conversation with the consortium members the gies and development processes Nokia gained indi-
IBM management proposed a new governance model rect access to existing open source communities.
based on the structures of successful community- Collaboration with and active participation in in-
managed open source organizations such as the cumbent open source communities was a major
Apache Foundation. In February 2004 the new non- process goal since the beginning of the project. Nokia
profit Eclipse Foundation was created owning the in- deliberately chose open source components not only
tellectual property rights of the software platform, due to their functionality, but also based on their li-
taking care of the technical infrastructure, managing censing opportunities and their community structure.
the release process, and undertaking public relations The legal requirements of the open source software
and marketing activities. Simultaneously the founda- allowed Nokia to mix it with proprietary elements. As
tion demanded more formal commitment from its a goal the project's community had to be lively and
members including up to 500'000 dollar annual mem- diverse without the dominance of a single company
bership fee and employment of several full-time (Jaaksi, 2007). Another goal was to establish an own
Eclipse developers. open source community around the Maemo operating
The strategic move by IBM to give up its privi- system. Nokia created a community platform for
leged role as owner of Eclipse and to create an inde- communication and collaboration. Later community
pendent foundation proved successful. Many more building activities included e.g. a logo competition
large corporations joined the Eclipse Foundation for the Maemo brand or the organization of a devel-
since its start (e.g. HP, Intel, Nokia, Motorola, Oracle, opers' conference. As of June 2009, there are over
and SAP), and contributions from non-IBM develop- 18'000 registered users on the community platform
ers increased substantially since 2004 representing sharing 861 applications for the Maemo system.
about half of the programmer population in 2007 The first version of the device, the 770 Internet
(Spaeth et al., 2009a). Tablet, was announced in May 2005. Simultaneously
Nokia started its developers' device program giving
4.2. Maemo by Nokia away 500 products at a low price to qualified open
source developers. In this way Nokia gained rapid ac-
Already in the year 2000 Nokia had started to ex- ceptance from skilled programmers because they en-
periment with the Linux kernel and other open source thusiastically started to develop new applications and
software (Stuermer et al., 2009). Engineers tested port existing Linux programs for the Internet Tablet.
embedded Linux on mobile devices in Nokia's labo- Attracting these technology leaders enabled Nokia
ratory until they felt that this operating system was also to test the device and receive valuable feedback
mature enough to be used on portable products. At and even bug fixes from the community. Eventually
the same time and totally independently, another in November 2005 the public sales of the Internet
stream of research was developing the novel product Tablet began.

How Firms Make Friends: Communities in Private-Collective Innovation 8


While building the device using open source com- Nevertheless, revealing substantial investments of
ponents and releasing source code under open source the new platform worried the management. By ac-
licenses, Nokia deliberately kept certain parts of the cessing the full software stack on a source code basis
device closed as proprietary software. On the one potential competitors were thought to be able to imi-
hand they were not able to publish proprietary third- tate FIC's product development quickly. On the other
party software such as hardware drivers or the Inter- hand implementing the software with the hardware
net browser Opera, on the other hand they retained was known to be very difficult requiring know-how
some of their own developments such as the graphi- and experience with the platform. Therefore the
cal user interface in order to prevent competitors to project leader argued FIC would retain competitive
copy Nokia's end user experience. In addition to this, advantage possessing the knowledge and resources to
Nokia preserved complete control over the develop- integrate the software and manufacture the devices
ment process of the Maemo software release. There- for the mass market.
fore, the process of development was more or less In fact, the market response was overwhelming:
transparent while the direct influence on the feature Soon after the product announcement in November
set or timing was strictly controlled by the Nokia 2006 and the beginning of the sales in July 2007, vol-
management. unteer programmers and other companies started to
However, Nokia granted access to software func- improve FIC's platform, eventually developing differ-
tionality and project governance by other means. ent types of operating systems for the mobile phone.
Technology-wise, Nokia invested substantial effort in Analysts as well as open source developers were en-
the provision of a free software development kit al- thusiastic that a hardware manufacturer had finally
lowing skilled programmers to create applications as announced the production of a completely open mo-
well as improvements at the operating system level. bile phone.
In this way community members were able to en- Similar to Eclipse and Maemo, the Openmoko
hance Maemo according to their needs and without community, too, received an online collaboration
depending on Nokia's permission. And in order to platform in order to exchange knowledge and source
permit external contributors to voice their positions code. Unique to the Openmoko management was that
more clearly, Nokia invited them to elect the Maemo they not only let the outside community participate in
Community Council in 2008. This group of five vol- the software development, but also involved them in
unteers represents the community to the company. the hardware selection process. For instance when the
While this council has no formal authority towards Openmoko engineers were looking for an appropriate
Nokia it allows better communication of the interests Wi-Fi chip for the smart phone, they asked on the
of the community. public mailing list if someone could recommend a
certain manufacturer. Indeed, as many of the sub-
4.3. Openmoko by Openmoko Inc. scribers on this list were working in relevant compa-
nies, the feedback to the request was high and even-
In 2005 the Taiwanese computer and components tually led to the discovery of the best-suited chip set.
manufacturer First International Computer, Inc. (FIC) Recently, in May 2009, Openmoko management re-
intended to create a smart phone based on the Win- leased even the exact hardware specifications includ-
dows Mobile operating system (von Krogh et al., ing architecture files with the computer-aided de-
2009a). However, due to the project leader's lobbying signs. With this step the company wants to test the
the management of FIC became convinced to try a ability of the community to contribute also to the
completely new way of developing a mobile phone. hardware design.
In order to differentiate from established smart phone Unfortunately, the Openmoko project was not so
manufacturers using software from third-party ven- much a success from a business point of view. Al-
dors, FIC decided to invest into the development of a though FIC decided in 2006 to spin-off the Open-
completely open platform based on the Debian moko project as an independent company and found
GNU/Linux operating system and other open source Openmoko, Inc., recently the product planning for
components. Although embedded Linux software 2010 had to be radically changed. Partly because of
manufacturers such as MontaVista existed, FIC the world economic crisis starting in late 2008, partly
thought they were not collaborating sufficiently with because of technical problems, the sales figures of the
the open source community and thus not using the mobile devices did not turn out as well as expected.
full potential of its innovative capabilities. Thus, in February 2009 Openmoko, Inc. announced
that the development of the third generation of the

How Firms Make Friends: Communities in Private-Collective Innovation 9


free mobile phone had to be postponed indefinitely et al., 2009a). The three organizational attributes
while the company had started a client project in or- knowledge revealing, accessibility, and credibility
der to increase cash flow immediately. Therefore, at positively influence the motivation and thus the con-
the time of writing, the commercial future of Open- tribution level of the outside participants.
moko is unclear. Therefore it proves necessary for firms to attract
active contributors in order to benefit the most from
5. Conclusions and Implications their investments in open source technologies. Such a
thriving community cannot be acquired on the market
So far, previous literature on corporate-sponsored but it requires substantial, credible, and long-term
open source communities has been reviewed and the commitment by the sponsor (Dahlander and Wallin,
empirical findings of this thesis have been presented. 2006). In return, such a realm of friends is not easily
This final section completes the picture with a broad- imitated making it a valuable asset of the firm.
er concept that characterizes the unique qualities of
such a firm-influenced network of open source soft- 5.1. The Concept of Interorganizational Competitive
ware producers. It is proposed to see corporate-spon- Advantage: Towards a Related Research Agenda
sored communities as a successful example of how a
firm may gain interorganizational competitive advan- Drawing from literature, Dyer and Singh (1998)
tage through its community. have introduced the concept of interorganizational
By portraying Eclipse, Maemo, and Openmoko, competitive advantage. This so-called relational view
three cases of corporate-driven open source initia- constitutes an alternative to the industry structure
tives are presented in detail. They illustrate the state- view (Porter, 1980) and the resource-based view
ment by O'Mahony (2007) that individuals are no (Wernerfelt, 1984; Barney, 1991). According to this
longer the only parties founding open source projects. concept unique resources and industry position are
In all of the above examples, large technology corpo- not the exclusive requirements for gaining above-av-
rations started their own branded open source project erage rents. Firms may also achieve competitive ad-
with the goal of releasing source code in order to at- vantage through a network of relationships with other
tract external actors to contribute to the future devel- organizations. Such embedded interfirm resources
opment effort. While the specific strategies and pro- and routines are difficult to imitate because they re-
cesses differed, all of the firms attempted not only to quire durable collaboration and cannot be acquired on
integrate code and reveal software, but also to active- the market. However, unlike resources owned by a
ly build a symbiotic community of third-party partici- firm alone these relationships depend on mutual un-
pants (Dahlander and Magnusson, 2005). The firms derstanding and trust of the partners and thus cannot
made use of all three elements of the open source de- be fully controlled by a single organization.
velopment model: integration of innovations, reveal- Reflecting on the qualities of firm-sponsored open
ing of knowledge, and creation of a community source communities it becomes obvious that they rep-
(O'Mahony, 2007). The firms applied the principle of resent an example of such interfirm relations. In all
private-collective innovation and enhanced the model the cases presented the focal firms IBM, Nokia, as
by creating an external community of volunteers and well as Openmoko Inc. are connected to external in-
partner firms. dividuals and organizations creating a network of
In fact, empirical evidence in the dissertation pa- durable relationships. Although each company cannot
pers shows that only by building a prospering com- completely control its partners, it is able to benefit
munity will the benefits of private-collective innova- substantially from the community it has initiated giv-
tion become fully accessible. For instance in the ing it an interorganizational competitive advantage as
Maemo case reputation gain was possible only be- described by Dyer and Singh (1998). Analysis of
cause Nokia successfully managed to start and main- their list of determinants shows that the characteris-
tain a community of volunteers and partnering firms tics of firm-sponsored open source communities in-
(Stuermer et al., 2009). In the case of Eclipse the deed matches the definition of interorganizational
community contributed innovations only because competitive advantage (Table 1). In the following,
IBM gave up control of the open source project and characteristics and sub-determinants of firm-spon-
empowered the members of the Eclipse Foundation sored open source communities are described based
(Spaeth et al., 2009a). As the third dissertation study on the framework of relational rents by Dyer and
shows, the perception of the project-sponsor positive- Singh (1998:663; the terminology of the titles in
ly affects the behavior of the community (von Krogh boldface is quoted from their illustration):

How Firms Make Friends: Communities in Private-Collective Innovation 10


Determinants of relational rents Subprocesses facilitating relational rents
1. Relation-specific assets a) Duration of safeguards
In firm-driven open source communi- In firm-driven open source communities: Open source software licenses protect jointly pro-
ties: A firm creates open source soft- duced open source software code thus ensuring availability of the code for all at all times.
ware components and other artifacts
collaboratively with its community. b) Volume of interfirm transactions
In firm-driven open source communities: Collaborative software development platforms on
the Internet facilitate communication of the firm with its community members.
2. Knowledge-sharing routines a) Partner-specific absorptive capacity
In firm-driven open source communi- In firm-driven open source communities: Overlapping bases of the firm's and community's
ties: Continuous interaction between knowledge evolve over time e.g. when external participants read source code or manuals.
employees of the company with exter-
nal community participants external- b) Incentives to encourage transparency and discourage free riding
izes the firm's tacit knowledge. In firm-driven open source communities: Shared norms and values by the firm and its com-
munity are critical elements favoring open communication and transparent decision pro-
cesses.
3. Complementary resources and a) Ability to identify and evaluate potential complementarities
capabilities In firm-driven open source communities: Through the transparent development process of
In firm-driven open source communi- open source software complementary capabilities become assessable.
ties: Stakeholders collectively develop
software such as the Linux kernel or b) Role of organizational complementarities to access benefits of strategic resource
the Eclipse platform by contributing complementarity
complementary capabilities. In firm-driven open source communities: Collaboration of the firm with its community aligns
their communication and culture over time.
4. Effective governance a) Ability to employ self-enforcement rather than third-party governance enforcement
In firm-driven open source communi- In firm-driven open source communities: As the private-collective model of innovation ex-
ties: Adaptive governance mecha- plains firms often contribute more to open source projects than they are legally obliged to
nisms ranging from informal agree- because of process-related benefits.
ments up to legally binding contracts
enable effective collaboration between b) Ability to employ informal versus formal self-enforcement governance mechanisms
the firm and its community. In firm-driven open source communities: Firms are very much interested to act as credible
members of the community in order to attract contributions from external participants.

Table 1: Firm-driven open source communities as an example of interorganizational competitive advantage (table
based on Figure 1 by Dyer and Singh, 1998:663)

1. Relation-specific assets: As firms must do some- by raising the frequency and scope of transac-
thing special to develop a competitive advantage, tions. In the case of corporate-sponsored open
the creation of assets has to be unique to the part- source communities this is achieved by the
ners. In the case of firm-sponsored communities firms facilitating communication and contribu-
e.g. Nokia has collaborated closely with profes- tions from within the open source community
sional developers and open source firms special- by introducing a collaborative software plat-
ized in certain open source components in order to form.
assemble the integrated software platform. Such
product development processes require intense 2. Knowledge-sharing routines: Interorganizational
knowledge exchange and transaction-specific learning is argued to be critical for competitive
know-how. Thus all stakeholders made substantial success. In open source projects this corresponds
nonrecoverable investments for the asset by e.g. to intense knowledge sharing between organiza-
learning a certain software code and architecture. tions and individuals. It is facilitated by various
While a decade ago physical proximity was essen- mechanisms. On a technical side there exists a tra-
tial for interorganizational cooperation, today it is dition of using newsgroup channels, mailing lists,
superseded by the virtual proximity derived from wikis, forums, film clips, chat rooms and many
the Internet. other communication media in order to broadcast
a) Duration of safeguards: In order to protect knowledge initially residing in one firm. Its diffu-
their investments, the partners seek to safe- sion outside the firm boundaries may then trigger
guard against opportunistic behavior. In the innovations from the community and thus contrib-
case of open source communities it is the soft- ute to the development process of the software.
ware license which ensures that the asset, the Through continuous interaction even tacit knowl-
source code, remains equally available for all edge residing within the community may be exter-
at all times (Stewart et al., 2006; Osterloh and nalized as explicit knowledge (Nonaka, 1994) be-
Rota, 2007). coming available for use by the sponsoring firm.
b) Volume of interfirm transactions: Alliance For instance experimentation by some community
partners increase their collaborative efficiency

How Firms Make Friends: Communities in Private-Collective Innovation 11


members lead to radical innovations in the case of potential partner firms and future employees
the Maemo platform. reducing the risk of a failed choice. E.g. Nokia
a) Partner-specific absorptive capacity: Recipi- used its firm-sponsored open source communi-
ents have to be able to recognize and assimi- ty to spot the most skilled, innovative, and mo-
late valuable knowledge from the sender. Co- tivated individuals. Through their prior partici-
hen and Levinthal (1990) therefore coined the pation in the community they felt attached to
concept of absorptive capacity. In open source Nokia and were very likely to accept its re-
communities the required overlapping knowl- cruitment offer.
edge bases evolve over time through frequent b) Role of organizational complementarities to
and often informal interactions. This enables access benefits of strategic resource comple-
the external community to absorb the knowl- mentarity: Next to the strategic fit the organi-
edge revealed by the firm e.g. by studying the zational complementarity represents a chal-
source code or reading manuals. lenge in forming alliances, too. Intense collab-
b) Incentives to encourage transparency and oration within open source communities facili-
discourage free riding: Knowledge sharing tates alignment of communication and culture
activities are fragile and in their essence can- and therefore increases the probability of suc-
not be fully controlled. Therefore, appropriate cess. E.g. a common norm is reciprocity or the
incentives as well as norms are necessary to culture of gifts which characterizes open
achieve a culture of sustainable knowledge ex- source communities strongly (Bergquist and
change. As the presented cases and other re- Ljungberg, 2001; Lakhani and von Hippel,
search on open source communities has shown 2003).
(Shah, 2006; Stewart and Gosain, 2006),
norms and values are critical elements of the 4. Effective governance: When organizations are
ideology in open source communities encour- interconnected, it is critical in which way they be-
aging transparency and discouraging free rid- have towards each other bringing up the question
ing. of governance. Dyer and Singh distinguish be-
tween third-party enforcement of agreements (e.g.
3. Complementary resources and capabilities: legal contracts) and self-enforcing agreements
Firms benefit in an alliance when they own com- (e.g. strategic alliances). The latter is further di-
plementary resources. In this way the collective vided between formal (e.g. financial investments)
effect is enhanced resulting in greater rents than and informal (e.g. trust and reputation) safeguards
the assets of each individual partner. Companies (1998: 669). The authors propose that transactions
that are not part of the network are thus underpriv- should be appropriately aligned with the specific
ileged. In the software industry examples of such governance structures minimizing transaction
synergies are found as network externalities when costs and maximizing value creation. Looking at
companies use and create compatible technolo- the governance mechanisms present in open
gies. When firms pool their resources and collec- source communities different forms of control are
tively develop a platform such as Linux, they ben- indeed in use, sometimes several of them in the
efit from collective rents (West and Gallagher, same open source project at once. For instance in
2006). Standardized components and file formats Eclipse there exist formal contracts obliging foun-
diffuse more rapidly if they are open and freely dation members to pay certain fees and ordering
accessible (Garud and Kumaraswamy, 1993). On engineers to program using the publicly available
an organizational level foundations of open source software code. Simultaneously, a second gover-
projects (e.g. Eclipse) represent formal networks nance structure regulates the development process
where members share benefits and costs of prod- in a rather self-enforcing way by allowing partici-
uct development. pants to influence the platform's properties
a) Ability to identify and evaluate potential through their contributions.
complementarities: Finding the right alliance a) Ability to employ self-enforcement rather
and integrating into it successfully is a delicate than third-party enforcement governance
endeavor. Firms have to invest substantial re- mechanisms: According to various research
sources in order to identify matching partners. self-governed mechanisms are more effective
Transparency and permeability of open source than is third-party enforcement. They save
projects enable gradual acquaintance between contracting, recontracting and monitoring

How Firms Make Friends: Communities in Private-Collective Innovation 12


costs, prove to be more flexible, and are more fusion by creating communities of volunteers and
probable to induce value-creating initiatives. professionals. On the other hand this dissertation ex-
In addition self-enforced relationships are tends the model of private-collective innovation (von
more difficult to copy than contractual agree- Hippel and von Krogh, 2003) by emphasizing the
ments. Thus they are more likely to represent a crucial role of value appropriation by community
sustainable advantage. The concept of private- building. Investing into the production of public
collective innovation depicts an example of goods without actively creating relationships with
such a self-enforced safeguard. As described other contributors would not make full use of private-
previously the model stimulates (but does not collective innovation. Also, as described in the previ-
force) firms to contribute to the creation of ous section, firm-managed communities are shown to
open source software rather than just using it represent an empirical example of the relational view
as free-rider. as defined by Dyer and Singh (1998) illustrating
b) Ability to employ informal versus formal comprehensively how interorganizational competitive
self-enforcement governance mechanisms: advantage results from the relationship between the
Informal safeguards such as trust or reputation sponsoring firm and its community.
are much more difficult to copy than formal Furthermore communities in private-collective in-
mechanisms like symmetric investments. novation represent in themselves a new form of dis-
While the latter can be imitated by competi- tributed innovation. In the literature on collective in-
tors, informal safeguards require time and novation various concepts have evolved. Two of them
trustful behavior to develop. Similarly, credi- shall be briefly highlighted and delineated against
bility of firms within an open source commu- communities of private-collective innovation. Eric
nity proves to be an important antecedent of von Hippel (1986) coined the idea of lead user com-
intrinsic motivation leading to the conclusion munities pointing out that users of technologies often
that trust and reputation increase the benefits achieve product innovations when they try to solve
of firm-sponsored communities. their personal problems. Such product improvements
in turn can be commercialized by firms manufactur-
As these theoretical and empirical characteristics ing them through mass production. While the concept
of interorganizational relationships have illustrated, a of user innovation is often taken as theoretical expla-
network of partners may result in a source of sus- nation for open source communities (von Hippel,
tained competitive advantage. However, managers of 2001), it does not take into account that in firm-spon-
such firm-sponsored communities have to bear in sored projects the community building activity is
mind that the rents are jointly generated and owned started by a firm, not by the user. Therefore commu-
by both sides, the corporation and the external com- nities in private-collective innovation are a separate
munity. The competitive advantage is extinguished concept of collective innovation because they follow
either if the firm withdraws its resources from the a different path of initiation.
community or if the community decides to separate Another concept are communities of practice
itself by forking the software code (Kogut and Metiu, (Brown and Duguid, 1991; Wenger and Snyder,
2001). Therefore, wise management techniques are 2000). These groups of professionals with similar in-
required by the sponsor in order to sustain a prosper- terests inside an organization meet to find novel solu-
ing community of skilled contributors and thus bene- tions for problems and thereby help to drive strategy,
fit from its competitive advantage. start new lines of businesses, transfer best practices,
develop skills, and recruit and retain talent. Commu-
5.2. Impact on Theory and Practice nities in private-collective innovation differ from
communities of practice mainly by their cross-bound-
On a theoretical level this dissertation contributes ary participation. Although often a majority of contri-
to the understanding of distributed innovation. On the butions stems from the sponsor of the community, it
one hand it shows that today's concept of open inno- is the fundamental goal of communities in private-
vation (Chesbrough, 2003; Chesbrough and collective innovation to enable and encourage exter-
Crowther, 2006) covers only part of the reality at nal participation.
least within the software industry. The examples giv- On the practical side managers of software com-
en show that firms have done far more than just inte- panies is shown the great benefits of building a com-
grating external innovations. Rather they have sub- munity. Though just a rough estimate, the case study
stantially released knowledge and invested in its dif- on Eclipse shows that external firms and individuals

How Firms Make Friends: Communities in Private-Collective Innovation 13


contributed software code valued at approximately not in order to attract external contributions. West
1.7 billion USD (Spaeth et al., 2009a). However, re- (2003) has titled this dilemma bluntly “How open is
vealing knowledge and investing in the creation of open enough?” It represents another balancing act be-
communities also incurs costs which have to be con- tween revealing of knowledge and thus enabling and
sidered in strategic decisions (Stuermer et al., 2009). cultivating participation, and retaining innovations
For policy makers, this dissertation shows possi- and thus being able to appropriate their value more
ble economic advantage of supporting the use and de- easily. So-called selective revealing has been ana-
velopment of open source software. The private-col- lyzed (Henkel, 2006), a generic theory, however, as
lective model of innovation illustrates how firms may well as practically applicable rules are still missing.
be incentivized to invest in the production of public Also related to this issue is the question how much of
good. Therefore, establishing a stimulating environ- a competitive contribution can a community provide
ment for firms to create and integrate open source and how sustainable is it. While e.g. Openmoko Inc.
software may possibly multiply the economic bene- successfully created a helpful community, it could not
fits for society in comparison to allowing incumbent prevent the product sales from dropping. Therefore,
proprietary software firms to collect monopoly rents. knowing the impact of the community's innovative-
ness on the overall business success might help future
5.3. Future Research Agenda product development investments to better appoint its
resources.
Although literature on open source communities Finally, finding sustainable business models of
is abundant, research-wise the phenomenon is firms investing in open source technologies repre-
nowhere near consummated. Within the context of sents an intriguing quest for strategy researchers as
firm-sponsored open source communities many ques- well practitioners. As Dahlander and Wallin (2006)
tions remain regarding governance. E.g. what is the found, open source software development in its origi-
optimal level of control which a firm should exert on nal sense is very diverse. Broadly independent com-
its community? As has been shown it is a precarious munities create open source components which may
balancing act: If the company loosens the reins too even compete for being employed in end-user appli-
much, the software project may not evolve in the in- cations (Spaeth et al., 2009b). Therefore, firms often
tended direction or competitors may imitate the tech- fill in the gap as so-called distributors integrating the
nology all too quickly. If the strings are tightened too components and applications as compact software so-
much, the credibility of the sponsor drops and moti- lutions as in the case of Nokia and Openmoko Inc.1
vation and performance decrease. Or even worse, the For creating the Maemo platform, Nokia for instance
external community actively opposes the sponsoring contracted several specialized individuals and small
firm and spins-off by forking the software code businesses in order to demand low-level changes on
(Kogut and Metiu, 2001). Therefore future studies on some components. Nowadays it is Nokia's strength to
balancing control and value appropriation of commu- integrate all these interdependent parts into a com-
nities are of great interest to researchers as well as plete end-consumer product. Acting as system inte-
managers. grator may be an attractive position in the value chain
Focusing particularly on the forking phenomenon, of complex technology products, as Brusoni and col-
a thorough empirical analysis could clarify what leagues (2001) argued in their empirical study of the
characteristics of control and openness have led to aircraft engine control system industry. Future re-
community splittings in the past and which factors search could advance findings of the evolution of
have determined the success respectively the failure such interconnected systems and provide advice on
of the renegades. Research so far has mostly looked how to manage them efficiently (von Krogh et al.,
at successful communities. Analyzing the behavior of 2009b). Also qualitative and quantitative studies on
firms and individuals which have caused a forking of firms assembling open source software are necessary
the project could lead to a better understanding of to gain more insight into the classical organizational
how to prevent this usually wearing-out process. research question of differentiation and integration
Studying so-to-speak failed projects could also help (Lawrence and Lorsch, 1967).
to clarify the unique factors of successful projects
leading to a clearer definition of sustainable commu-
nity management. 1 It needs to be noted that there are some successfully man-
aged bundles of open source software packages such as the
Another strategic challenge is the decision on
Debian GNU/Linux distribution. This is particularly im-
what kind of knowledge should be revealed and what portant since it represents a comprehensive source for
sampling open source components (Spaeth et al., 2007).

How Firms Make Friends: Communities in Private-Collective Innovation 14


The delicate position of firms starting and manag- Dahlander, L. & Magnusson, M. G. (2005), 'Relationships be-
ing an open source project has been illustrated. It has tween open source software companies and communities: Ob-
servations from Nordic firms', Research Policy 34(4), 481-
become apparent that only when firms sponsor open 493.
source projects in a substantial and credible way
while carefully accepting the norms and rules of the Dahlander, L. & Wallin, M. (2006), 'A man on the inside: Un-
locking communities as complementary assets', Research Pol-
external contributors, a thriving community will icy 35(8), 1243-1259.
evolve. By analyzing and interpreting the myriad
facets of communities in private-collective innova- Dyer, J. H. & Singh, H. (1998), 'The Relational View: Cooper-
tion and by providing a characterization of such com- ative Strategy and Sources of Interorganizational Competitive
Advantage', Academy of Management Review 23(4), 660-679.
munities, this dissertation hopes to help firms to suc-
ceed in the balancing act of entering into sustainable Feiman, J. & Driver, M. (2001), 'Commentary: Eclipse, a de-
relationships with outside participants. veloper's dream?', http://news.cnet.com/2009-1001-
275495.html, accessed June 10, 2009..

Franck, E. & Jungwirth, C. (2003), 'Reconciling Rent-Seekers


and Donators – The Governance Structure of Open Source',
References Journal of Management and Governance 7(4), 401-421.

Frey, B. S. & Jegen, R. (2001), 'Motivation Crowding Theory',


Alexy, O. & Leitner, M. (2008), 'Norms, Rewards, and their Journal of Economic Surveys 15(5), 589-611.
Effect on the Motivation of Open Source Software Develop-
ers', Technical report, Technical University of Munich. Frey, B. S. & Oberholzer-Gee, F. (1997), 'The Cost of Price
Incentives: An Empirical Analysis of Motivation Crowding-
Bae, J. & Cameron, G. T. (2006) 'Conditioning effect of prior Out', American Economic Review 87(4), 746-755.
reputation on perception of corporate giving', Public Relations
Review 32, 144-150. Glaser, B. & Strauss, A. (1967), The discovery of grounded
theory: Strategies for qualitative research, Chicago: Aldine.
Barney, J. (1991), 'Firm Resources And Sustained Competitive
Advantage', Journal of Management 17(1), 99-120. Grand, S.; von Krogh, G.; Leonard, D. & Swap, W. (2004),
'Resource Allocation Beyond Firm Boundaries: A Multi-Level
Bergquist, M. & Ljungberg, J. (2001), 'The Power of Gifts: Model for Open Source Innovation', Long Range Planning 37,
Organizing Social Relationships in Open Source Communi- 591-610.
ties', Information Systems Journal 11(4), 305-320.
Grant, R. M. (1996), 'Toward a Knowledge-based Theory of
Bonaccorsi, A. & Rossi, C. (2003), 'Why Open Source soft- the Firm', Strategic Management Journal, Special Issue:
ware can succeed', Research Policy 32(7), 1243-1258. Knowledge and the Firm 17, 109-122.

Brown, J. S. & Duguid, P. (1991), 'Organizational Learning Harhoff, D. (1996), 'Strategic Spillovers And Incentives For
and Communities-of-Practice', Organization Science 2, 40-57. Research And Development', Management Science 42(6), 907-
925.
Brusoni, S.; Prencipe, A. & Pavitt, K. (2001), 'Knowledge
Specialization, Organizational Coupling, And The Boundaries Harhoff, D.; Henkel, J. & von Hippel, E. (2003), 'Profiting
Of The Firm: Why Do Firms Know More Than They Make?', From Voluntary Information Spillovers: How Users Benefit
Administrative Science Quarterly 46(4), 597-621. By Freely Revealing Their Innovations', Research Policy 32,
1753-1769.
BusinessWire (1996), 'IBM To Acquire Object Technology In-
ternational Inc', Hars, A. & Ou, S. (2002), 'Working for free? Motivations for
http://findarticles.com/p/articles/mi_m0EIN/is_1996_Feb_22/ participating in open-source projects', International Journal of
ai_18023630, accessed June 10, 2009.. Electronic Commerce 6, 25-39.

Chesbrough, H. (2003), Open innovation. The new imperative Henkel, J. (2006), 'Selective Revealing In Open Innovation
for creating and profiting from technology, Harvard Business Processes: The Case Of Embedded Linux', Research Policy
School Press. 35(7), 953-969.

Chesbrough, H. & Crowther, A. K. (2006), 'Beyond High Hertel, G.; Niedner, S. & Herrmann, S. (2003), 'Motivation of
Tech: Early Adopters Of Open Innovation In Other Industries', software developers in Open Source projects: an Internet-
R&D Management 36(3), 229-236. based survey of contributors to the Linux kernel', Research
Policy 32(7), 1159-1177.
Cohen, W. M. & Levinthal, D. A. (1990), 'Absorptive Capaci-
ty: A New Perspective on Learning and Innovation', Adminis- von Hippel, E. (2001), 'Innovation By User Communities:
trative Science Quarterly 35, 128-152. Learning From Open-Source Software', Sloan Management
Review 42(4), 82-86.
Dahlander, L. & Magnusson, M. (2008), 'How do Firms Make
Use of Open Source Communities?', Long Range Planning von Hippel, E. (1986), 'Lead Users: A Source Of Novel Prod-
41(6), 629-649. uct Concepts', Management Science 32(7), 791-805.

How Firms Make Friends: Communities in Private-Collective Innovation 15


Liebeskind, J. P. (1996), 'Knowledge, strategy, and the theory
von Hippel, E. & von Krogh, G. (2006), 'Free revealing and of the firm', Strategic Management Journal, Special Issue:
the private-collective model for innovation incentives', R&D Knowledge and the Firm 17, 93-107.
Management 36(3), 295-306.
Markus, M. L. (2007), 'The governance of free/open source
von Hippel, E. & von Krogh, G. (2003), 'Open Source Soft- software projects: monolithic, multidimensional, or configura-
ware and the "Private-Collective" Innovation Model: Issues tional?', Journal of Management and Governance 11(2), 151-
for Organization Science', Organization Science 14(2), 209- 163.
223.
Muller, P. & Pénin, J. (2006), 'Why Do Firms Disclose Knowl-
Jaaksi, A. (2007), Experiences on Product Development with edge And How Does It Matter?', Journal of Evolutionary Eco-
Open Source Software, in Joseph Feller; Brian Fitzgerald; nomics 16(1-2), 85-108.
Walt Scacchi & Alberto Sillitti, ed.,'Open Source Develop-
ment, Adoption and Innovation', Boston: Springer, , pp. 85-96. Nonaka, I. (1994), 'A Dynamic Theory of Organizational
Knowledge Creation', Organization Science 5, 14-37.
Kroah-Hartman, G.; Corbet, J. & McPherson, A. (2008), 'Lin- O'Mahony, S. (2003), 'Guarding The Commons: How Com-
ux Kernel Development: How Fast it is Going, Who is Doing munity Managed Software Projects Protect Their Work', Re-
It, What They are Doing, and Who is Sponsoring It', Technical search Policy 32(7), 1179-1198.
report, The Linux Foundation.
O'Mahony, S.; Diaz, F. C. & Mamas, E. (2005), 'Harvard Busi-
von Krogh, G. (2008), Researching the Private-Collective In- ness School Case Study: IBM and Eclipse (A) and (B)', Tech-
novation Model, in Daved Barry & Hans Hansen, ed.,'The nical report, Harvard Business School Case Study.
SAGE Handbook of New Approaches in Management and Or-
ganization', Sage, 396-397. O'Mahony, S. & Ferraro, F. (2007), 'The Emergence Of Gover-
nance In An Open Source Community', Academy of Manage-
von Krogh, G. (2002), 'The Communal Resource And Infor- ment Journal 50(5).
mation Systems', The Journal Of Strategic Information Sys-
tems 11, 85-107. O’Mahony, S. (2007), 'The governance of open source initia-
tives: what does it mean to be community managed?', Journal
von Krogh, G. & von Hippel, E. (2006), 'The Promise of Re- of Management and Governance 11(2), 139-150.
search on Open Source Software', Management Science, Spe-
cial Issue on open source software 52, 975-983. O’Mahony, S. & Bechky, B. A. (2008), 'Boundary Organiza-
tions: Enabling Collaboration among Unexpected Allies', Ad-
von Krogh, G. & Spaeth, S. (2007), 'The open source software ministrative Science Quarterly 53(3 Special Issue: Social
phenomenon: Characteristics that promote research', The Movements in Organizations and Markets), 422-459.
Journal of Strategic Information Systems 16(3), 236-253.
Osterloh, M. & Rota, S. (2007), 'Open source software devel-
von Krogh, G.; Spaeth, S.; Haefliger, S. & Wallin, M. (2008), opment—Just another case of collective invention?', Research
'Open Source Software: What we know (and do not know) Policy 36(2), 157-171.
about motives to contribute', ETH Zurich.
Porter, M. E. (1980), 'Competitive Strategy', New York: Free
von Krogh, G.; Spaeth, S.; Stuermer, M. & Hertel, G. (2009a), Press.
'The credible sponsor: Participants’ motivation and firm at-
tributes in collaborative digital innovation', ETH Zurich. Pénin, J. (2007), 'Open Knowledge Disclosure: An Overview
Of The Evidence And Economic Motivations', Journal of Eco-
von Krogh, G.; Stuermer, M.; Geipel, M.; Spaeth, S.; Hae- nomic Surveys 21(2), 326-347.
fliger, S. & Baldwin, C. (2009b), 'How component dependen-
cies predict change in complex technologies', ETH Zurich and Raymond, E. S. (1999), The Cathedral & the Bazaar, 1st Edi-
Harvard Business School. tion. O’Reilly, Sebastopol, CA.

Lakhani, K. R. & von Hippel, E. (2003), 'How open source Roberts, J.; Hann, H. & Slaughter, S. (2006), 'Understanding
software works: "free" user-to-user assistance', Research Poli- the motivations, participation, and performance of open source
cy 32(6), 923-943. software developers: A longitudinal study of the Apache
projects', Management Science 52(7), 984-999.
Lawrence, P. R. & Lorsch, J. W. (1967), 'Differentiation and
Integration in Complex Organizations', Administrative Science Shah, S. (2006), 'Motivation, Governance, And The Viability
Quarterly 12(1), 1-47. Of Hybrid Forms In Open Source Software Development',
Management Science 52(7), 1000-1014.
Lee, G. K. & Cole, R. E. (2003), 'From a Firm-Based to a
Community-Based Model of Knowledge Creation: The Case Spaeth, S.; Haefliger, S.; von Krogh, G. & Renzl, B. (2008),
of the Linux Kernel Development', Organization Science 14, 'Communal resources in open source software development',
633-649. Information Research 13(1).

Lerner, J. & Tirole, J. (2002), 'Some Simple Economics of Spaeth, S.; Stuermer, M.; Haefliger, S. & von Krogh, G.
Open Source', Journal of Industrial Economics 50(2), 197- (2007), Sampling In Open Source Software Development: The
234. Case For Using The Debian Gnu/Linux Distribution, in 'HIC-
SS 2007 40th Annual Hawaii International Conference on Sys-
tem Sciences'.

How Firms Make Friends: Communities in Private-Collective Innovation 16


Teece, D. J. (1986), 'Profiting from technological innovation:
Spaeth, S.; Stuermer, M. & von Krogh, G. (2009a), 'Enabling Implications for integration, collaboration, licensing and pub-
Knowledge Creation Through Outsiders: Towards a Push lic policy', Research Policy 15(6), 285-305.
Model of Open Innovation', International Journal of Technol-
ogy Management (Forthcoming Special Issue on Open Innova- Wernerfelt, B. (1984), 'A Resource-Based View of the Firm',
tion). Strategic Management Journal 5, 171-180.
West, J. (2003), 'How open is open enough? Melding propri-
Spaeth, S.; von Krogh, G.; Stuermer, M. & Haefliger, S. etary and open source platform strategies', Research Policy
(2009b), 'A Lightweight Model of Component Reuse: A Study 32(7), 1259-1285.
of Software Packages in Debian GNU/Linux', Technical re-
port, ETH Zurich. West, J. & Gallagher, S. (2006), 'Challenges of open innova-
tion: the paradox of firm investment in open-source software',
Stewart, K.; Ammeter, A. & Maruping, L. (2006), 'Impacts of R&D Management 36(3), 319-331.
License Choice and Organizational Sponsorship on User Inter-
est and Development Activity in Open Source Software West, J. & O'Mahony, S. (2008), 'The Role of Participation Ar-
Projects', Information Systems Research 17(2), 126-144. chitecture in Growing Sponsored Open Source Communities',
Industry & Innovation 15(2), 145-168.
Stewart, K. J. & Gosain, S. (2006), 'The Impact Of Ideology
On Effectiveness In Open Source Software Development West, J. & O'Mahony, S. (2005), 'Contrasting Community
Teams', MIS Quarterly 30(2), 291-314. Building In Sponsored And Community Founded Open Source
Projects', Proceedings of the 38th Annual Hawai‘i Interna-
Stuermer, M. (2005), 'Open Source Community Building', tional Conference on System Sciences.
Master's thesis, University of Berne.
Wu, C.; Gerlach, J. H. & Young, C. E. (2007), 'An Empirical
Stuermer, M.; Spaeth, S. & von Krogh, G. (2009), 'Extending Analysis of Open Source Software Developers’ Motivations
private-collective innovation: a case study', R&D Manage- and Continuance Intentions', Information and Management
ment 39(2), 170-191. 44(3), 253-262.

How Firms Make Friends: Communities in Private-Collective Innovation 17


Appendix 1
Enabling Knowledge Creation through Outsiders:
Towards a Push Model of Open Innovation*

Forthcoming publication in a special issue on open innovation


in the International Journal of Technology Management

Sebastian Spaeth
Matthias Stuermer
Georg von Krogh

Chair of Strategic Management and Innovation


Department of Management, Technology, and Economics
ETH Zurich, Switzerland

Open innovation is increasingly being adopted in business and describes a situation in which firms
exchange ideas and knowledge with external participants, such as customers, suppliers, partner
firms, and universities. This article extends the concept of open innovation with a push model of
open innovation: knowledge is voluntarily created outside a firm by individuals and organizations
who proceed to push knowledge into a firm's open innovation project. For empirical analysis we ex-
amine source code and newsgroup data on the Eclipse Development Platform during a six year
timespan. We find that outsiders invest as much in the firm's project as the founding firm itself.
Based on the insights from Eclipse, we develop four propositions: “the preemptive generosity” of a
firm, “continuous commitment”, “an adaptive governance structure”, and “a low entry barrier” are
contexts that enable the push model of open innovation. We conclude with a summary of the push
model and discuss implications for research and management practice.

Introduction existing ideas and knowledge outside the firm's


boundary.” While literature on this topic has mostly
Since its introduction by Chesbrough (2003), the focused on utilization of readily available external
framework of 'open innovation' as a major source of knowledge, how that knowledge was originally creat-
knowledge and contributor to the firm's competitive- ed has often been neglected. However, it is important
ness has attracted attention. The framework “assumes to ask who and why outsiders would produce knowl-
that firms can and should use external ideas as well edge for open innovation in the first place? What is
as internal ideas, and internal and external paths to the motivation of individuals and firms to create and
market, as they look to advance their technology" freely reveal knowledge that is of use to other (even
(Chesbrough et al., 2006, p.2). According to Ches- competing) innovators? West and Gallagher (2006)
brough “inflow is concerned with the exploitation of

* The authors would like to thank Peter Amhof for his valuable research assistance as well as Daniel Megert, Ian Skerrett, Sonali
Shah, the editors of the special issue and two anonymous reviewers for their helpful feedback. This research was supported by the
Swiss National Science Foundation (grant #100012-101805).

How Firms Make Friends: Communities in Private-Collective Innovation 18


asked what would happen to open innovation if ev- 3) the coupling of the former processes. In the fol-
eryone sought to be a 'free rider' absorbing others’ in- lowing section, we build upon these core processes
novations. and categorize them according to two dimensions: the
We believe there could be a form of open innova- main actors in innovation and their relationship char-
tion that does not merely rely on the inflow of readily acterized by knowledge creation and utilization. The
available existing knowledge, but one that stimulates resulting framework is summarized in Table 1.
or enables the voluntary creation of knowledge by in- First, firms sell internal knowledge for further ex-
dividuals and organizations outside of the firm, con- ploitation by outsiders (inside-out). In this process,
tributing to the firm's open innovation projects. The firms profit from existing internal knowledge or
purpose of this study is to empirically investigate knowledge creation with other partners, as was exam-
such a push model of open innovation, by asking ined, for example, by Gassmann and Enkel (2004) in
what are the enabling contexts that make this model the case of IBM's Industry Solution Laboratory and
of open innovation work? Thus we aim to find orga- by Hounsell and Smith (1988) in the case of cello-
nizational and technical decisions and activities firms phane licensing. The firm appropriates returns from
can make in order to get external constituents to their existing knowledge by licensing it to third par-
proactively push knowledge into open innovation ini- ties. Another, well-researched entity which bases
tiatives by the firm. The next section briefly reviews their business model on this process is InnoCentive
work on open innovation and identifies the research (see Allio, 2004).
gap. The third section describes the case of the Second, the firm takes advantage of available ex-
Eclipse Developer Platform, initiated by IBM. The ternal knowledge to improve competitiveness for ex-
fourth section describes the research design and ample, by licensing in intellectual property (IP) or in-
methodology. The fifth section presents the findings tegrating the knowledge of suppliers early on (out-
and the sixth section develops four propositions on side-in). This process is frequently mentioned in as-
enabling contexts. The last section discusses implica- sociation with open innovation (Chesbrough, 2003;
tions for research and management practice. 2006). Licensing in of intellectual property, rather
than creating it through investment in R&D, is one
Open Innovation way for the firm to benefit from pools of external
knowledge. Tapping into freely available knowledge,
Distributed knowledge creation (Gibbons, 1994) for example, in the form of open source software, is
and the open innovation framework (Chesbrough, another method analyzed by several authors (Henkel,
2003) have increasingly gained the attention of schol- 2006; Dahlander and Wallin, 2006; von Hippel and
ars and practicing managers alike. Firms such as Proc- von Krogh, 2003, 2006; Chesbrough, 2006). Occa-
ter & Gamble (Dodgson, Gann and Salter, 2006) and sionally, firms set up “listening posts” or use various
platforms such as InnoCentive (Allio, 2004) have forms of technology and perform research scans to
been actively pursuing and benefiting from open inno- identify pools of knowledge which can be of use to
vation. Chesbrough (2006) and Chesbrough, Van- their internal innovation (Gassmann and Gaso, 2004).
haverbeke, and West (2006) provide excellent over- Such methods may also be of use for firms that
views on this emerging field of research and theoriz- source external innovations created by users of tech-
ing. nologies (von Hippel, 1988; von Hippel and Katz,
Gassmann and Enkel (2004) provide an in-depth 2002). In this case, the firm benefits from exploiting
analysis of the framework, suggesting three core pro- knowledge in the relationship with an outsider. The
cesses. They classify by the direction of knowledge outside-in process can, therefore, be found in the
flows and call these: 1) inside-out, 2) outside-in, and lower left-hand corner of Table 1.

Actor
Firm External constituents
Inside-out process Push model
Knowledge Creation e.g., licensing of unsolicited knowledge creation through
intellectual property outsiders
Activity
Outside-in process Knowledge spillover to outsiders
Knowledge Utilization e.g., technology sourcing, e.g., reverse engineering
using open source software

Table 1: Open innovation categories

How Firms Make Friends: Communities in Private-Collective Innovation 19


Third, firms also couple the previous two process- acteristics of public goods (Bessen, 2002; Myatt and
es and innovate jointly with other organizations that Wallis, 2002). Although there is a general interest in
have complementary characteristics. Knowledge cre- the motivation behind such contributions (Lakhani
ation and utilization in these settings have been ex- and von Hippel, 2003; Bagozzi and Dholakia, 2006),
amined exhaustively, for example, in research on little is known about why programmers would con-
strategic alliances and joint ventures (Kogut, 1988; tribute to a project that is tightly associated with a
Hamel, 1991; Mowery et al., 1996, Hagedoorn 1993; commercial firm. Such voluntary (but “not-called-
Hagedoorn, Link and Vonortas, 2000). Combining for”) contributions by outside participants, targeted to
both knowledge creation and utilization and referring a specific open innovation project, are not well ex-
to knowledge that flows between actors inside-out plained by current research (West and O’Mahony,
and outside-in this process spans both corners on the 2008). Understanding the motivation of contributors
left-hand side in Table 1. and identifying enabling contexts for the creation of
Depicted along the two dimensions, actor and external knowledge pools is in the interest of a firm
type of relationship, two more settings need to be ex- seeking to benefit from such knowledge and allocate
amined: knowledge creation and utilization through resources to such activity (Grand, von Krogh,
externals warrant more attention. Knowledge utiliza- Leonard and Swap, 2004). To summarize, our re-
tion by outsiders is concerned with the usage of firm- search question can be formulated as: What are the
internal knowledge by outside actors. In contrast to enabling contexts that make a push model of open in-
the licensing of intellectual property to third parties, novation work?
such knowledge spillover usually does not result in In order to explore the setting of a push model and
financial remuneration for the firm. Knowledge identify the enabling contexts, we examined the
spillovers can be voluntary or involuntary (e.g., in the Eclipse Project founded and dominated by IBM as an
form of reverse engineering). Although research has example of such a push model project inductively. In
identified some positive aspects of knowledge the next section, we describe the project in detail be-
spillovers for firms (Harhoff 1996; Harhoff, Henkel, fore turning to the research design.
von Hippel, 2003), for example, through the creation
of industry standards or positive effects on the firm's The Eclipse Platform
reputation, firms usually seek to prevent this type of
uncompensated knowledge outflows (Mayer, 2006). Eclipse is an Integrated Development Environ-
While the literature on open innovation claims ment (IDE) and consists of a hierarchical structure of
that this framework can save development cost and software components called top- and sub-projects.
time by "leveraging external development" (Ches- Initially designed for developing applications based
brough, 2006, p.17), it has thus far neglected the on the programming language Java, the Eclipse plat-
process of how readily available external knowledge form today supports a wide range of programming
which could be exploited was created in the first languages and software development frameworks. It
place. Questions remain as to how and why external is used and developed by many major software com-
individuals and firms create pools of readily available panies, who are often direct competitors, and also by
external knowledge and how and why they make it academic institutions and individuals.
possible for a company to use it. West and Gallagher The initial version of Eclipse was developed by
(2006, 2006a) recognized this and underscored that the Canadian company Object Technology Interna-
motivating the supply of external knowledge is one tional (OTI) which was acquired by IBM in 1996. In
of the critical factors to make open innovation work. order to increase the adoption of the platform and sell
This shows an open and promising space for fur- complementary products, IBM released the source
ther research, namely the situation where outsiders code of Eclipse, then valued at $40 million (Fitzger-
voluntarily create knowledge which is pushed into ald, 2006), as open source software on November 7,
the firm’s open innovation project. An example 2001. A board of directors staffed with representa-
serves to highlight the importance of this category: tives from major software companies such as Borland
while the firm's use of open source software is often and Rational Software exerted some influence on the
highlighted as a case of knowledge exploitation by development path of Eclipse.1 However, IBM still re-
the firm (Chesbrough, 2006), voluntary contributions mained in control of the overall development strate-
to open source software are left out of the analysis. In gy, as well as of the technical infrastructure of the
open source software, individual developers and or-
ganizations contribute technologies that exhibit char-
1 http://www.eclipse.org/org/pr.html

How Firms Make Friends: Communities in Private-Collective Innovation 20


Eclipse project. This changed on February 2, 2004 actively participate in the ongoing development of
when the Board of Stewards announced the formation the platform, although the intellectual property of the
of the Eclipse Foundation, an independent gover- committed source code has to be assigned to the
nance body.2 With this decision, IBM ceded the con- Eclipse Foundation and licensed under the "Eclipse
trolling position it had held since the start of the Public License," an open source license.
project and allowed other firms and institutions to be-
come equal members in the project governance. To- Research Design
day the Foundation is responsible for all the technical
infrastructure, the coordination of the development In the following section, we outline the research
process such as central release management and design adopted to explore the enabling contexts for
project life cycle steering, the handling of the source the push model of open innovation. Due to the ex-
code's intellectual property rights, and the promotion ploratory nature of our research and the open re-
of Eclipse and its wider ecosystem e.g., through the search question, we performed a single case study
organization of developer conferences.3 As will be and inductively generated propositions on enabling
shown subsequently, IBM still invests a vast amount contexts (Campbell, 1975; Yin, 2003). The main pur-
of resources into the development of the open source pose of the design is to establish a “real-world case”
software. From a business perspective this may be of a push model and to identify and operationalize the
explained by strategic as well as operative reasons. enabling contexts. In effect, we create a push model
On the one hand Eclipse enabled IBM to focus their of innovation from the case (Eisenhardt, 1989). We
software product portfolio while entering rapidly into sampled the case of the Eclipse Development Plat-
the IDE industry and gaining a key position (O'Ma- form for two reasons. First, Eclipse is an open inno-
hony et al., 2005). On the other hand IBM earns rev- vation project started by a dominant and unambigu-
enues by selling licenses for the WebSphere product ous sponsor, IBM, which aims at facilitating external
line which is based on Eclipse. adoption and attracting outside developers to contrib-
In the Eclipse development process there are three ute to the software platform (O'Mahony et al., 2005).
different groups of individuals involved: first, the Thus, it fits with the broad initial category of a “push
users of the software who are able to freely download model” that was established by examining and posi-
the executable application as well as the source code. tioning existing research contributions (Table 1). Sec-
These individuals may also contribute by submitting ond, Eclipse was released as open source software in
bug reports to the central issue tracking system, by 2001 and since then it has been developed in a trans-
providing bug fixes and code enhancements to the parent manner, providing our research team access to
core developers, or by communicating through news- data accumulated over a period of six years.
groups and mailing lists. Second, there is the 'neutral' Yin (2003) suggests identifying key informants
Eclipse Foundation which governs everything but the that can shed extensive light on the context of a case.
source code itself. For instance, within the Founda- Following initial desk research on the Eclipse project,
tion, there is a Project Management Council for each we gained a deeper understanding of the project by
top-level Eclipse project. The Council, is required to interviewing an Eclipse core developer employed by
accomplish numerous administrative and coordina- IBM who has been involved with the project since
tive tasks as elaborately defined in the Eclipse Devel- 1999. The key informant relayed a narrative of how
opment Process.4 Third, there is the type of developer the project had evolved over time (Pentland, 1999).
community usually associated with open source He provided his viewpoints on topics such as IBM's
projects, allegedly based on meritocratic principles relationship with external participants, the overall
(Fielding, 1999; Scacchi, 2004), which requires de- governance and structure of the project, communica-
velopers to prove their skills through code contribu- tion within the community, as well as the norms and
tions before they are granted commit access to the actual behavior within the community.
code repository (von Krogh et al., 2003). Those de- Before proceeding to the different data sources in
velopers can be independent, or employed by IBM or this study, it is necessary to clarify the terminology
other companies. There is no need to be a formal applied in order to distinguish between the roles and
member of the Eclipse Management Organization, actors involved. 'Contributors' is the term used to de-
the coordinating body of the Foundation, in order to scribe individuals that participate in an open source
project though programming, testing, or commenting.
A 'committer' is a contributor and software developer
2 http://www.eclipse.org/org/press-release/feb2004foundationpr.php
3 http://www.eclipse.org/org/ with permission to alter the official source code of
4 http://www.eclipse.org/projects/dev_process/development_process.php

How Firms Make Friends: Communities in Private-Collective Innovation 21


the project in question. We also distinguish between ments of knowledge creation (Nonaka, 1994). Hence,
'employed' versus 'external' contributors or commit- measures of dialogue provide one type of access to
ters as people who are employed or not employed by the level of knowledge creation within and outside of
IBM. 'Employed' and 'voluntary' contributors differ- IBM over time. In this work, we limit our focus to
entiate between individuals who develop open source explicit knowledge (e.g., documentation, software
code as part of their job and those who are not direct- code, answers to queries) and exclude the sharing of
ly paid for doing so. tacit knowledge between project participants. In order
One crucial construct in this study is the measure- to capture knowledge creation based on dialogue, we
ment of internal and external knowledge creation. analyzed the traffic of the 90 distinct newsgroups
Two data sources were used for this. First, we mea- (discussion forums) centered on several topics within
sured the evolution of the source code itself, an estab- the Eclipse project. Most top- and lower-level
lished and common source of data in open source projects offer such a newsgroup intended for ex-
software projects (MacCormack et al., 2006). All changing advice within the respective knowledge do-
source code contributions within the Eclipse Concur- main. There are ten top-level projects, such as 'core'
rent Versions System (CVS) code repository were an- Eclipse, web tools, performance tools etc. At the time
alyzed, using the tool CVSAnaly (Herraiz et al., of the writing 73 sub-projects dedicated to more nar-
2007; Robles et al., 2004). This effort provided all row topics or technologies were associated with one
added lines of code minus all deleted lines of code of these top level projects. Examples of such sub-
per developer over the entire development period. projects are integrations of specific programming lan-
The timespan of the development activities collected guages, specific clients, or embedded platforms. Al-
ranged from April 20015 until February 2007. We ac- though mailing lists exist, we decided to focus on dis-
cumulated the code repositories of all Eclipse top- cussions in the newsgroups. The key informant sug-
level projects which covered 63 million lines of code gested that mailing lists are used exclusively for in-
contributed by 605 distinct committers. By identify- ternal discussions on software development issues
ing the developers' names and comparing them to and address only core developers of the platform.
participants in the Eclipse newsgroups, we were able However, newsgroups serve general programmers by
to identify the organizational affiliation of 565 devel- providing support services directly from the develop-
opers, categorizing them into IBM and non-IBM de- ers on how to use the Eclipse platform. Examining
velopers. Code contributions for both groups were such support activities within newsgroups enabled us
examined on a monthly basis and split into the times- to focus on knowledge flows since the structure of
pan before and after the creation of the Eclipse Foun- conversations consists mostly of questions asked and
dation in January 2004. The purpose of this split was answers provided as will be shown subsequently.
to better identify the role of the firm and external par- We downloaded all the messages from the Eclipse
ticipants in knowledge creation, as well as adaptive newsgroup servers starting from February 2001 until
governance structures in the push model of open in- July 2007, which resulted in 371,942 unique mes-
novation. sages. Out of these, 116,973 messages started a new
Second, given our interest in contexts that enable discussion thread and 254,969 messages replied to
the push of knowledge in open innovation, there was them. No established coding scheme exists in the lit-
a need to focus on knowledge flows between the erature that use thread starts and replies to differenti-
firms and its outside contributors. Hence, we exam- ate between knowledge-seeking versus knowledge-
ined knowledge flows by measuring the communica- providing messages. Therefore, we assume that start-
tion between what scholars have categorized as ing a new discussion thread contains a question and
“knowledge seekers” (people with a need to learn, responding to such a message presents an answer.
obtain information, help, and insights, etc.) and This assumption was confirmed by the key infor-
“knowledge providers” (people with knowledge who mant, and it has also been used in recent research on
are willing and able to share it) within the Eclipse communication within communities (Dahlander and
community (Constant et al., 1996). We adopt Ikujiro Frederiksen, 2007). To further test the assumption in
Nonaka's view that “dialogue” between knowledge our case, we coded the content of a random sub-sam-
providers and seekers is one amongst several ele- ple of 500 newsgroup messages. The results con-
firmed that thread starts are mostly knowledge-seek-
ing activities and thread replies are generally knowl-
5 Although the project was not publicly available until No- edge-providing messages.
vember 2001, development in the central code repository
was started in April 2001 by members of IBM

How Firms Make Friends: Communities in Private-Collective Innovation 22


In order to classify the content, we assigned each would be quickly noticed and reprimanded. In the
message to one of five different types: to qualify as case of OTI (see below), which was acquired by
"question," a message had to contain a request for IBM, early members communicated using OTI email
knowledge, while an "answer" provided a response to addresses. Therefore "OTI" affiliation was coded as
a specific demand. Since the solution to a technical IBM affiliation. With newsgroup messages, we also
problem often required a certain amount of informa- distinguished between a pre- and post-Eclipse Foun-
tion, sometimes a clarification question was neces- dation period.
sary. Further, the response provided occasionally did
not solve the problem directly. Thus, the knowledge Findings
seeker answered with a more elaborated question. We
coded these two types of messages as "follow-up" In this section, we present the findings from the
questions. Thread starts which included announce- case study examining the results of the two sources of
ments or coordination issues as well as thread replies data. These cover the number of active source code
containing non-specific information or 'Thank you' developers and their programming productivity, as
without verification of the success of the answer were well as the knowledge-seeking and -providing pat-
classified as "comment." Finally, wrongly addressed terns within the community. The data from the case
emails or spam mails were assigned the category leads to the formulation of propositions on contexts
"noise." that enable the push model of open innovation.
Two authors coded the randomly selected mes- The first data source consists of the code analysis
sages independently and inter-rater reliability was analyzing who programmed how much in which
measured using Fleiss Kappa (Fleiss, 1971). The timeframe. Figure 1 shows the increase of active de-
Kappa of 0.816 is well above 0.7, usually recom- velopers over time, where an 'active developer' is de-
mended as the acceptable minimum (Straub et al, fined as a person who creates (or removes) at least
2004). As Table 2 illustrates, knowledge-seeking one line of source code of the project within the cal-
messages (questions) constitute 93.6% of all 'thread endar month. Before launching the Eclipse founda-
start' messages, while "answer" messages represent tion in 2004, IBM involved on average 41.8 active
70.8% of all replies - the rest mostly being comments committers per month (sd:18.6). IBM's commitment
such as 'Thank you, it worked' or others. The cate- increased in the period with the Eclipse Foundation
gories "follow-up," "comment," and "noise" were ex- to an average of 98.3 active developers per month
cluded since knowledge-seeking, respectively -provi- (sd:8.3). The number of committers without IBM af-
sion, is unclear or not present. filiation also increased. Including the unidentified
Analogous to the source code analysis, we distin- committers, it went from an average of 7.4 active de-
guished between messages from IBM and non-IBM velopers per month before 2004 (sd:4.6) to 63.8
authors. Newsgroup messages contain the real name (sd:38.5), multiplying by a factor of 8.62. While IBM
of the sender as well as their email address and insti- provided more manpower in the beginning, the
tutional affiliation. According to the key informant, growth of external committers is apparent. Since May
IBM has an explicit company policy to disclaim com- 2006, but for one month, the number of external com-
pany affiliation, and any attempt to deviate from this mitters has always remained higher than the number
of IBM committers.

question answer follow-up comment noise sum share


thread start 140 ; 139 3;3 0;1 5;6 1;1 149 (140+139) / (149*2) = 0.9362
thread reply 29 ; 32 249 ; 248 26 ; 43 46 ; 27 1;1 351 (249+248) / (351*2) = 0.7080
sum 169 ; 171 252 ; 250 26 ; 44 51 ; 33 2;2 500

Table 2: Inter-rater coding (coder A; coder B) of content of 500 messages

How Firms Make Friends: Communities in Private-Collective Innovation 23


Figure 1: Active committers per month Figure 2: Total messages per month by IBM and non-
IBM newsgroup participants

Figure 3: Thread reply over thread start ratio Figure 4: External knowledge flow ratio over time

While the number of external contributors is im- The second data source is concerned with knowl-
portant, previous research found that employed con- edge flows through newsgroup message postings as a
tributors are more active than voluntary contributors proxy for knowledge creation through dialogue (Non-
(Dahlander and Wallin, 2006). Thus, it is also neces- aka, 1994). We look at knowledge flows from inter-
sary to examine the productivity of contributors. Al- nal and external sources and distinguished between
though much debated, one of the standard measures knowledge-seeking and -providing messages. As ex-
of productivity in software engineering is the number plained in the research design section, a message
of added lines of code (LOC) (see Albrecht and posting containing a question ("thread start") is con-
Gaffney, 1983). Table 3 give an overview of LOC sidered as knowledge-seeking, while sending an an-
contributions from both IBM and non-IBM parties swer ("thread reply") as knowledge-provision. Figure
using the LOC-added metric. The percentage of ex- 2 shows the aggregation of the monthly newsgroup
ternal code contributions increased significantly after traffic from 2001 until 2007. While the number of
2004 and represents about a third of all added lines of messages from IBM senders remains at an approxi-
code. mately constant rate, the newsgroup postings from
non-IBM individuals continuously increase over
time.

How Firms Make Friends: Communities in Private-Collective Innovation 24


LOCs before Foundation LOCs after Foundation Total LOCs
(4.01–12.03: 33 months) (1.04-2.07: 37 months) (4.01-2.07: 70 months)
IBM 8,410,972 73.94% 31,346,054 60.42% 39,757,026 62.85% 313 committers
Non-IBM 2,042,365 17.95% 19,507,145 37.60% 21,549,510 34.07% 252 committers
Unidentified 922,488 8.11% 1,027,085 1.98% 1,949,573 3.08% 40 committers
Total 11,375,825 100.00% 51,880,284 100.00% 63,256,109 100.00% 605 committers

Table 3: Lines of code from IBM and non-IBM developers

Thread Replies Thread Starts Relative External


Net Knowledge
(Knowledge Pro- (Knowledge Knowledge Flow
Flow
vision) Seeking) (non-IBM / IBM)
IBM 28,649 3,406 25,243
Pre-Foundation 0.743
Non-IBM 45,040 26,287 18,753
IBM 52,156 5,116 47,040
Post-Foundation 1
Non-IBM 129,124 82,164 46,960
Sum 254,969 116,973
Share 68.55% 31.45%

Table 4: Knowledge flow of IBM and non-IBM developers

A first knowledge flow pattern can be observed by if both non-IBM and IBM contributors provide the
analyzing thread start versus message reply ratios. As same amount of knowledge in the open innovation
Figure 3 demonstrates, the proportional rate of thread project and becomes larger than 1 if non-IBM partici-
replies over thread starts remains constantly consis- pants provide more knowledge than IBM members.
tently higher (mean: 9.56) for IBM than for non-IBM The values of relative external knowledge flow in the
people (mean: 1.63). For example, in February 2006 project before and after the creation of the Eclipse
non-IBM individuals answered 1.8 times as many Foundation can be found in Table 4.
discussion threads as they initiated (4145/2289), In the period before the Eclipse Foundation was
whereas IBM members replied to 9.9 times more dis- installed in the open innovation project, the external
cussion threads than they started (1327/134). Assum- knowledge flow ratio measured by this method is be-
ing that replies provide answers while thread starts low 1, meaning that more knowledge was provided to
represent questions, IBM members provided signifi- the Eclipse user community by IBM employees than
cantly more knowledge than non-IBM participants by external participants. However, the value for the
during the timespan investigated. second period is approximately 1. Thus, the external
In a second analysis we examined the newsgroup contributors provided just as much explicit knowl-
data in absolute numbers (see Table 4 and Figure 4) edge as IBM did. The monthly values are plotted in
showing that knowledge provision was initially car- Figure 4 and show that the ratio has been larger than
ried out by IBM members but is increasingly accom- 1 since the middle of 2006 until mid 2007. Measured
plished by external individuals. In order to measure during this period non-IBM contributors provided
the amount of knowledge flow we created the metric more knowledge within the Eclipse project than IBM
of net knowledge flow for each group of knowledge employees.
providers subtracting questions from answers both We also analyzed knowledge flow ratios for each
for IBM and for non-IBM individuals. This resulted of the 90 different newsgroups. Interestingly, the pat-
in an estimation of a net knowledge flow value per terns in the single newsgroups differ widely. Some of
month, bearing in mind that we focus narrowly on the “sub-communities” communicating through a
explicit knowledge only that can be exchanged particular newsgroup are dominated by IBM mem-
through electronic media. Taking the ratio of non-IB- bers, whereas others are almost completely run by ex-
M's over IBM's values results in one measure of rela- ternal non-IBM programmers. For example, the
tive external knowledge flow. This measure equals 1 newsgroup for the top-level project BIRT (Business

How Firms Make Friends: Communities in Private-Collective Innovation 25


Intelligence and Reporting Tools) is mostly contrib- USD was not done for altruistic purposes. It was
uted to by non-IBM developers, both in terms of rather a generous first signal which in turn attracted
source code as well as in newsgroup messages. A other software firms and stimulated the creation and
possible interpretation of this finding is that being sharing of explicit knowledge by external contribu-
able to create an architecture in which outsiders can tors, including software patches or bug reports. The
take the lead may reduce the entry barrier for new- source code analysis revealed that non-IBM develop-
comers (external and voluntary contributors). ers begun to participate in the development process
immediately after the release of the source code by
Enabling contexts and the push model IBM. Our proposition relates to existing research in
various areas. Based on observations of IBM's release
In this paper, we seek to inductively generate of more than 500 patents to open source software de-
propositions (Eisenhardt, 1989) on contexts that en- velopers, von Krogh (2006) suggested that companies
able the push model of open innovation. A push mod- need to show preemptive generosity in order to moti-
el of open innovation refers to knowledge creation by vate customers to contribute to an external knowl-
external contributors that is uncompensated by the edge pool. Similarly, Sawhney and Prandelli (2000)
firm but that pushes knowledge into the open innova- commented that communities of customers that en-
tion process of the firm. As such, the push model of gage in innovation need to obtain sponsorship of re-
open innovation refers to an extension of the process- sources from firms. Nahapiet and Ghoshal (1998,
es of open innovation proposed by Gassmann and p.250) suggest that “social capital” plays a role in
Enkel (2004). A summary of the case observations, knowledge creation which, according to Coleman
corresponding propositions on enabling contexts, and (1988), in turn facilitates certain actions. Thus, social
the related literature can be found in Table 5. capital for the firm can be understood as the value of
First, we propose that preemptive generosity en- its relationships. It facilitates the creation of knowl-
ables the push model of open innovation. In the case edge by effecting the conditions necessary for ex-
we found that by releasing the source code of Eclipse change and combination knowledge as it builds trust
under a liberal open source software license IBM re- and establishes norms of sharing (Nahapiet and
vealed valuable knowledge to the public. The deci- Ghoshal, 1998).
sion giving away source code valued at 40 million

Findings Propositions on enabling con- Literature


texts
• Revealing of initial Eclipse source code Preemptive generosity James Coleman (1988)
by IBM Nahapiet and Ghoshal (1998, p.250)
Sawhney and Prandelli (2000)
Harhoff et al. (2003)
Spencer (2003)
von Krogh (2006)
Muller and Pénin (2006)
• Constant number of IBM programmers Continuous commitment Rizova (2006)
involved in Eclipse development Gächter et al. (2006)
• Constant level of participation in news- Shah (2006)
groups
• Launching non-profit foundation with Adaptive governance Shah (2003)
equal membership of firms and other in- structures West (2003)
stitutions concerning organizational con- Dahlander and Magnusson (2005)
trol, technical infrastructure, and man- Shah (2006)
agement of intellectual property Fauchart and von Hippel (2006)
Sawhney and Prandelli (2000)
O'Mahony and Ferraro (2007)
• Sub-projects run by non-IBM developers Lowering barriers to entry Rullani (2006)
• Modular architecture Lakhani and von Hippel (2003)
Shah (2003)
von Krogh, Spaeth and Lakhani (2003)

Table 5: Contexts enabling the push model of open innovation

How Firms Make Friends: Communities in Private-Collective Innovation 26


Second, we propose that continuous commitment pants to the Eclipse project. A first indication is seen
enables the push model of open innovation. The find- in some sub-projects which are run by and contrib-
ings indicate that IBM showed such continuous com- uted to almost exclusively by IBM-external develop-
mitment. Opening up a project in terms of architec- ers. Entry barriers, or the cost of joining and con-
ture and intellectual property rights is a required, but tributing to an open innovation project, affect the lev-
maybe not a sufficient, condition in order to receive el of contributions of external participants. According
outside contributions. IBM continuously contributed to previous research in the context of open source
to Eclipse with more than 40 developers until 2003, projects, external and voluntary contributors typically
and afterwards with more than 80 people. Thus, IBM join by conducting peripheral tasks and then later
permanently invested significant resources in their gravitate towards more important and technically
open innovation project although external source complicated tasks (Rullani, 2006; Lakhani and von
code commits remained low until the initiation of the Hippel, 2003; Shah, 2003). Moreover, a modular ar-
independent non-profit foundation. This finding also chitecture which allows for such plug-ins to be easily
receives some support from existing research. It has added would seem to have an effect on entry barriers
been shown that successful research projects receive (von Krogh et al, 2003). Thus, entry barriers may be
continuous resource commitment by the firm (Rizo- of a technical as well as of an organizational nature
va, 2006). Shah (2006) also finds that stable norms of and can be influenced, for example, by the choice of
reciprocity in a community of contributors, that is, programming language or creation of acertain soft-
the need to both give and receive ideas and knowl- ware architecture.
edge, is key in order to motivate contributors.
Third, we propose that an adaptive governance Discussion and conclusion
structure enables the push model of open innovation.
IBM ceded control over the administration and intel- Enabling contributions of external participants, be
lectual property of the software code in the Eclipse they individuals or organizations, and benefiting from
project. Not until IBM let external contributors be- these is crucial for any effective open innovation
come equal members of the governance structure, did project. Extending the processes identified by
outside contributions increase significantly. IBM em- Gassmann and Enkel (2004), we show that the litera-
ployees continued to provide more knowledge to the ture to date has mainly examined how already exist-
technology development (software code) than non- ing external knowledge is exploited by firms. Profit-
IBM developers. Yet, the steady growth of participa- ing from internal knowledge by making it available
tion in communication by external participants led to to selected parties against direct financial reimburse-
a higher share than that of IBM employees in the cre- ments has also been part of the open innovation
ation of explicit knowledge through dialogue, mea- framework since its inception (Chesbrough, 2003).
sured by thread starts and replies. Thus, an adaptive However, the creation of knowledge with outside
governance structure matters for open innovation contributors and the provision of contexts that enable
projects to succeed in attracting continuous contribu- this process represents a research gap. We extended
tions from external participants (see also Shah, 2006). the open innovation framework by proposing a push
Based on the case findings, we can conclude that the model of open innovation. In this model, knowledge
establishment of a foundation is not only used to pro- specific to the open innovation project is created by
tect intellectual property rights (O'Mahony, 2003) and external constituents, without direct financial remu-
to resolve IP conflicts but also to facilitate continued neration from the firm. We conducted a case study of
contributions from external participants by providing the Eclipse Development Platform and focused, in
a fair governance body. While it might well be that particular, on the creation of explicit knowledge that
voluntary and external contributors would not enjoy can be exchanged through electronic media. Based on
working for IBM for free, external contributors seem the findings, we inductively developed four proposi-
to be less hesitant to assign their IP to an independent tions on contexts that enable the push model of open
foundation even if IBM remains the dominant player innovation: the preemptive generosity of a firm, con-
in the open innovation project. tinuous commitment by the firm, an adaptive gover-
Fourth, based on some preliminary findings from nance structure, and a low entry barrier.
the case study, we propose that low entry barriers en- By releasing valuable source code under a liberal
able the push model of open innovation. While they open source software license, IBM revealed valuable
were not strong, we found indications that low entry knowledge to the public and showed preemptive gen-
barriers could be important to attract external partici- erosity. While managers will often hesitate to make

How Firms Make Friends: Communities in Private-Collective Innovation 27


the firm's knowledge available without direct finan- in open innovation needs to take the full project life
cial compensation, “preemptive generosity” can be cycle into account, including the continuous commit-
economically advantageous under conditions where ment by the founding firm and cumulative contribu-
future returns may ensue (Harhoff et al., 2003; Muller tions by external constituents.
and Pénin, 2006). Thus, firms enable a push in open In addition it would be interesting to investigate
innovation through preemptive generosity, contribut- values comparable to the "net knowledge flow" in
ing some initial knowledge (including intellectual other projects in order to verify the usefulness of the
property) which is considered valuable by outsiders method. Furthermore, our research takes a very re-
and, thus, motivates others to reciprocate by con- strictive view of explicit knowledge that can be ex-
tributing knowledge. Empirical evidence of the posi- changed through electronic media. Our concept of di-
tive influence of knowledge sharing on innovation alogue is also tied to electronic media (used by
performance has also been found by Spencer (2003). knowledge seekers and providers) and does not cover
Future research should investigate preemptive dialogue and collective reflection in face-to-face in-
generosity across a range of projects with different teraction as proposed by Nonaka and Takeuchi
objectives, constituents, and technologies. Motivating (1995). In line with knowledge creation theory, future
people to contribute by first giving may be one of the research needs to focus on both explicit and tacit
most fundamental factors in open innovation projects. forms of knowledge and investigate more extensively
Moreover, more research is needed on the targets of processes of knowledge creation including socializa-
preemptive generosity. Firms may choose to release tion, externalization, combination, and internalization
technology to the public, but one can imagine many (Nonaka, 1994, Nonaka et al 2006).
other “gifts” too, including workshops and seminars, Third, while open innovation projects are often set
free labor, brands, infrastructure, etc. up outside the firm, the firm may tend to govern the
The second enabling context, continuous commit- project (e.g., Sawhney and Prandelli, 2000). The level
ment was found to be crucial to the push model. of control varies: while some firms might prefer to
Gächter et al. (2006) argue that "social norms in the keep tight governance in which they continue to
community reward innovators' reciprocal contribu- wield control, others hope to benefit from the “open-
tions." Following this reasoning, preemptive generos- ness” of their projects by receiving third-party contri-
ity will not be sufficient to sustain external knowl- butions and, thus, save on development costs. West
edge creation by external participants in the innova- (2003) examined the tension inherent in governance
tion process. Hence, firms enable outside knowledge structures. The firm is caught between retaining con-
creation in open innovation through continuous com- trol and benefiting from openness and, in particular,
mitment, that is, contributing knowledge (including he showed situations in which it is beneficial to fol-
intellectual property) over time which is considered low the one or the other strategy. Shah (2006; see
valuable by outsiders, and that creates norms of reci- also 2003; see also Dahlander and Magnusson, 2005)
procity. also finds that companies will either try to retain con-
Moreover, further research is needed in order to trol over a project (lead a gated community) or focus
establish a minimum (or maximum) threshold of con- on openness. Governance structure impacts on the
tribution by founding firms in order to sustain open contribution levels of external contributors in open
innovation projects. The benefit of investing in open innovation: the more open the project, the more em-
innovation projects may be tremendous as we esti- phasis contributors put on a 'fair' governance struc-
mated in a simple calculation: While IBM initially ture (Shah, 2006). As shown in the case of Eclipse, a
contributed software that was valued at 40 million transparent and independent legal structure was re-
USD, external contributors to the Eclipse Develop- quired to attract contributions. It seems that while un-
ment Platform project created software representing a der some conditions purely norm-based intellectual
value of roughly 1.7 billion USD over the examined property systems work (Fauchart and von Hippel,
period.6 Thus, any calculation of return on investment 2006), open innovation projects require a legal gover-
nance structure considered 'fair' by outside contribu-
6 Applying the COCOMO model (Boehm, 1981), a somewhat tors (O'Mahony and Ferraro, 2007). However, what is
simplifying yet common metric for estimating the effort of cre- “fair” might differ amongst outside participants, de-
ating software, external contributions of 21.5 million lines of pending on their short- and long-term interests, the
code in Eclipse represent a work effort of approximately
214,000 man-months, or an investment of about 1.7 billion actions of the firm, alternative projects where they
USD. For the calculation we assumed the parameters for a can work, and so forth. Thus, firms enable a push in
semi-detached software project and defined the costs of a high- open innovation through adaptive governance struc-
ly qualified Java software developer at USD 8000 per month.

How Firms Make Friends: Communities in Private-Collective Innovation 28


tures, that is, governance that over time will be con- gle case within the realm of open source software.
sidered fair by outside contributors. Future research should replicate and externally vali-
It is interesting to observe that the introduction of date our inductively generated propositions. Large-
an independent, non-profit foundation did not imme- scale quantitative studies could evaluate the effective-
diately increase the number of active outside partici- ness and impact of specific contexts that enable exter-
pants. Only some months after the launch of the nal knowledge creation for the purposes of a specific
Eclipse Foundation, could we register a significant domain of innovation. Additional qualitative research
increase in numbers. Thus, external knowledge cre- is also needed on the motivation of external contribu-
ation through dialogue needs time to evolve in open tors in order to identify factors that increase levels of
innovation. This observation lends support to Wasko outside participation.
and Samer's (2005) analysis of knowledge sharing in Managers wanting to benefit from knowledge cre-
electronic networks of practice. The authors found ation with external contributors should consider the
that individuals' experience in knowledge sharing and four contexts that enable the push model of open in-
their structural embeddedness in the network corre- novation. Although the initial revealing of knowl-
late with the amount they share. The current research edge, continuous commitment, adaptive governance
cannot conclude which aspects of governance struc- structure, and the lowering of entry barriers require
tures matter most to external knowledge creation in substantial investments and are associated with risks,
open innovation. Possible explanations could be the the benefits of receiving knowledge from outside par-
ceding of organizational control, the technical infra- ticipants can often be extensive. However, as the case
structure run and owned by a neutral organization, or of IBM shows, managers need to take the “long
the management of intellectual property by the non- view” of open innovation: addressing, inviting, and
profit foundation. Future research needs to examine attracting external contributors into the project re-
these factors in more detail. quires substantial investment over time. External par-
There were indications in the case study that entry ticipants have many alternative projects to which they
barriers potentially impact on the level of contribu- can contribute, and they make their choices based on
tions by outside contributors. Von Krogh, Spaeth, and their expectation as to whether or not projects will
Lakhani (2003) identified four technical dimensions succeed. Seeing long-term sustained commitment by
of entry barriers in open source projects that impact- the firm may make them more inclined to join the
ed on newcomers' propensity to join: the ease of open innovation project.
modifying a module (that is the difficulty of the task
at hand), the extent to which a developer could
choose the computer language to fulfill the task, the
ease with which to plug in a new module of code into
a software architecture, and the extent to which a
module of code is intertwined with or works indepen-
dently of other modules. Firms impact on the choice References
of programming language used or the amount of de-
veloper information to hand. Thus, firms can enable Albrecht, A. and Gaffney, J. "Software function, source lines
of code, and development effort prediction: A software science
outside knowledge creation in open innovation validation," IEEE Transactions on Software Engineering (9),
through lowering the entry barriers for external par- 1983, pp. 639-648.
ticipants who seek to join and contribute. However,
Allio, R. J. "CEO interview: The InnoCentive Model of Open
more research on a large sample of projects will be Innovation," Strategy & Leadership (32:4), 2004, pp. 4-9.
needed in order to test the proposition.
To conclude, researchers will benefits from cate- Bagozzi, R. P. and Dholakia, U. M. "Open Source Software
gorizing open innovation projects using the simple 2- User Communities: A Study of Participation in Linux User
Groups," Management Science (52:7), 2006, pp. 1099-1115.
by-2 framework presented in section 2 of this paper.
The framework allows the ordering of the investiga- Bessen, J. "Open Source Software: Free Provision of Complex
tion into the motives and sources of knowledge Public Goods", Technical report, Research on Innovation,
http://www.researchoninnovation.org/opensrc.pdf, 2002.
which play out in open innovation. In particular, giv-
en the current state of research, more empirical work Boehm, B. Software Engineering Economics, Prentice-Hall,
should be targeted at the push model of open innova- 1981.
tion. The case of Eclipse is an exemplary case of such
Campbell, D. T. "Degrees Of Freedom And The Case Study,"
a push model. However, the study is limited to a sin- Comparative Political Studies (8:1), 1975, pp. 178-193.

How Firms Make Friends: Communities in Private-Collective Innovation 29


Grand, S., von Krogh, G., Leonard, D. and Swap, W. "Re-
Chesbrough, H. W. Open Business Models: How to Thrive in source allocation beyond firm boundaries: A multi-level model
the New Innovation Landscape, Harvard Business School for Open Source innovation," Long Range Planning (37:6),
Press, Boston, MA, 2006. 2004, pp. 591-610.

Chesbrough, H. W. Open Innovation: The new imperative for Gächter, S., von Krogh, G. and Haefliger, S. "Private-Collec-
Creating and Profiting from Technology, Harvard Business tive Innovation and the Fragility of Knowledge Sharing",
School Press, Boston, 2003. http://www.nottinghamnetlearning.com/economics/cedex/pa-
pers/2006-21.pdf, University of Nottingham, CeDEx Discus-
Chesbrough, H. W., Vanhaverbeke, W. and (eds), J. W. Open sion Paper No. 2006-21, 2006.
Innovation: Researching a New Paradigm, Oxford University
Press, 2006. Hagedoorn, J. "Understanding the Rationale of Strategic Tech-
nology Partnering: Interorganizational Modes of Cooperation
Coleman, J. S. "Social Capital In The Creation Of Human and Sectoral Differences," Strategic Management Journal
Capital," The American Journal of Sociology (94), 1988, pp. (14:5), 1993, pp. 371--385.
95-120.
Hagedoorn, J., Link, A. N. and Vonortas, N. S. "Research part-
Constant, D., Sproull, L. and Kiesler, S. "The Kindness Of nerships," Research Policy (29:4-5), 2000, pp. 567--586.
Strangers: The Usefulness Of Electronic Weak Ties For Tech-
nical Advice," Organization Science (7:2), 1996, pp. 119-135. Hamel, G. "Competition for Competence and Interpartner
Learning Within International Strategic Alliances," Strategic
Dahlander, L. and Frederiksen, L. "Communication And Management Journal (12:S1), 1991, pp. 83-103.
Boundary Spanning In A Firm-Hosted Online User Communi-
ty", Technical report, Imperial College London, 2007. Harhoff, D. "Strategic spillovers and incentives for R&D,"
Management Science (42:6), 1996, pp. 907-925.
Dahlander, L. and Magnusson, M. G. "Relationships between
Open Source Companies and Communities: Observations Harhoff, D., Henkel, J. and von Hippel, E. "Profiting from vol-
from Nordic Firms," Research Policy (34), 2005, pp. 481-493. untary information spillovers: how users benefit by freely re-
vealing their innovations," Research Policy (32:10), 2003, pp.
Dahlander, L. and Wallin, M. W. "A man on the inside: Un- 1753-1769.
locking communities as complementary assets," Research
Policy (35:8), 2006, pp. 1243-1259. Henkel, J. "Selective revealing in open innovation processes:
The case of embedded Linux," Research Policy (37:7), 2006,
Dodgson, M., Gann, D. and Salter, A. "The Role Of Technolo- pp. 953-969.
gy In The Shift Towards Open Innovation: The Case Of Proc-
ter & Gamble," R&D Management (36:3), 2006. Herraiz, I., Gonzalez-Barahona, J. M. and Robles, G. "Fore-
casting The Number Of Changes In Eclipse Using Time Series
Eisenhardt, K. M. "Building Theories from Case Study Re- Analysis"'29th International Conference on Software Engi-
search," Academy of Management Review (14:4), 1989, pp. neering Workshops (ICSEW'07)', 2007.
532-550.
von Hippel, E. The Sources of Innovation, Oxford University
Fauchart, E. and von Hippel, E. "Norms-based intellectual Press, 1988.
property systems: the case of French chefs", http://web.mit.e-
du/evhippel/www/papers/vonhippelfauchart2006.pdf, Working von Hippel, E. and Katz, R. "Shifting Innovation To Users Via
Paper, 2006. Toolkits," Management Science (48:7), 2002, pp. 821-833.

Fielding, R. T. "Shared Leadership In The Apache Project," von Hippel, E. and von Krogh, G. "Free revealing and the pri-
Communications of the ACM (42:4), 1999, pp. 42-43. vate-collective model for innovation incentives," R&D Man-
agement (36:3), 2006, pp. 295-306.
Fitzgerald, B. "The Transformation Of Open Source
Software," MIS Quarterly (30:3), 2006, pp. 587-598. von Hippel, E. and von Krogh, G. "Open Source Software and
the "Private-Collective" Innovation Model: Issues for Organi-
Fleiss, J. L. "Measuring nominal scale agreement among many zation Science," Organization Science (14:2), 2003, pp. 209-
raters," Psychological Bulletin (76), 1971, pp. 378-382. 223.

Gassmann, O. and Enkel, E. "Towards a Theory of Open Inno- Hounshell, D. A. and Smith, J. K. Science and Strategy: Du
vation: Three Core Process Archetypes"'Proceedings of the Pont R&D, Cambridge University Press, 1988.
R&D Management Conference (RADMA)', 2004.
Kogut, B. "Joint Ventures: Theoretical and Empirical Perspec-
Gassmann, O. and Gaso, B. "Insourcing Creativity with Lis- tives," Strategic Management Journal (9:4), 1988, pp. 319-
tening Posts in Decentralized Firms," Creativity and Innova- 332.
tion Management (13:1), 2004, pp. 3-14.
von Krogh, G. "Customers Demand Their Share of IP," Har-
Gibbons, M. The New Production of Knowledge: The Dynam- vard Business Review (February), 2006, pp. 45-46.
ics of Science and Research in Contemporary Societies, Sage,
1994.

How Firms Make Friends: Communities in Private-Collective Innovation 30


von Krogh, G., Spaeth, S. and Lakhani, K. R. "Community, Robles, G., Koch, S. and Gonzalez-Barahona, J. "Remote
Joining, And Specialization In Open Source Software Innova- Analysis And Measurement Of Libre Software Systems By
tion: A Case Study," Research Policy (32:7), 2003, pp. 1217- Means Of The Cvsanaly Tool", Technical report, Universidad
1241. Rey Juan Carlos, Wirtschaftsuniversität Wien, 2004.

Lakhani, K. R. and von Hippel, E. "How Open Source Soft- Rullani, F. "Dragging developers towards the core. How the
ware Works: 'Free' User-to-User Assistance," Research Policy Free/Libre/Open Source Software community enhances devel-
(32:6), 2003, pp. 923-943. opers' contribution"(2006/22), Technical report, Laboratory of
Economics and Management (LEM), Sant'Anna School of Ad-
MacCormack, A., Rusnak, J. and Baldwin, C. Y. "Exploring vanced Studies, Pisa, Italy, available at
The Structure Of Complex Software Designs: An Empirical http://ideas.repec.org/p/ssa/lemwps/2006-22.html, 2006.
Study Of Open Source And Proprietary Code," Management
Science (52:7), 2006, pp. 1015-1030. Sawhney, M. and Prandelli, E. "Communities of creation:
managing distributed innovation in turbulent markets," Cali-
Mayer, K. J. "Spillovers and Governance: An Analysis of fornia Management Review (42:4), 2000, pp. 24-54.
Knowledge and Reputational Spillovers in Information Tech-
nology," Academy of Management Journal (49:1), 2006, pp. Scacchi, W. "Free And Open Source Development Practices In
69-84. The Game Community," IEEE Software (21:1), 2004, pp. 59-
66.
Mowery, D. C., Oxely, J. E. and Silvermann, B. S. "Strategic
Alliances and Interfirm Knowledge Transfer," Strategic Man- Shah, S. "Understanding the Nature of Participation & Coordi-
agement Journal (17), 1996, pp. 77-91. nation in Open and Gated Source Software Development
Communities", Doctoral dissertation, 2003.
Muller, P. and Pénin, J. "Why Do Firms Disclose Knowledge
And How Does It Matter?," Journal of Evolutionary Econom- Shah, S. K. "Motivation, Governance, and the Viability of Hy-
ics (16:1-2), 2006, pp. 85-108. brid Forms in Open Source Software Development," Manage-
ment Science (52:7), 2006, pp. 1000-1014.
Myatt, D. P. and Wallis, C. "Equilibrium Selection and Public-
good Provision: The Development of Open Source Software," Spencer, J. W. "Firms' Knowledge-Sharing Strategies In The
Oxford Review of Economic Policy (18:4), 2002, pp. 446-461. Global Innovation System: Empirical Evidence From The Flat
Panel Display Industry," Strategic Management Journal
Nahapiet, J. and Ghoshal, S. "Social Capital, Intellectual Capi- (24:3), 2003, pp. 217-233.
tal, and the Organizational Advantage," Academy of Manage-
ment Review (23), 1998, pp. 242-266. Straub, D. W., Boudreau, M. and Gefen, D. "Validation Guide-
lines for IS Positivist Research," Communications of the Asso-
Nonaka, I. "A Dynamic Theory of Organizational Knowledge ciation for Information Systems (13), 2004, pp. 380-427.
Creation," Organization Science (5), 1994, pp. 14-37.
Wasko, M. M. and Samer, F. "Why Should I Share? Examin-
Nonaka, I., von Krogh, G. and Voelpel, S. "Organizational ing Social Capital and Knowledge Contribution in Electronic
knowledge creation theory: Evolutionary paths and future ad- Networks of Practice," MIS Quarterly (29:1), 2005, pp. 35-37.
vances," Organization Studies (27:8), 2006, pp. 1179-1208.
West, J. "How Open is Open Enough? Melding Proprietary
Nonaka, I. and Takeuchi, H. The Knowledge-Creating Compa- and Open Source Platform Strategies," Research Policy (32:7),
ny: How Japanese Companies Create the Dynamics of Inno- 2003, pp. 1259-1285.
vation, Oxford University Press, 1995.
West, J. and Gallagher, S. "Challenges of open innovation: the
O'Mahony, S., Diaz, F. C. and Mamas, E. "IBM and Eclipse paradox of firm investment in open-source software," R&D
(A)", Technical report, Harvard Business School Case Study, Management (36:3), 2006b, pp. 319-331.
2005.
West, J. and Gallagher, S. "Patterns of Open Innovation in
O'Mahony, S. and Ferraro, F. "The Emergence Of Governance Open Source Software"'Open Innovation: Researching a New
In An Open Source Community," Academy of Management Paradigm', Oxford University Press, 2006a, pp. 82-109.
Journal (50:5), 2007.
West, J. and O'Mahony, S. "The Role of Participation Archi-
Pentland, B. T. "Building process theory with narrative: From tecture in Growing Sponsored Open Source Communities,"
description to explanation," Academy of Management Review Industry & Innovation (15:2), 2008, pp. 145-168.
(24:4), 1999, pp. 711-724.
Yin, R. K. Case Study Research: Design and Methods, Sage,
Rizova, P. "Are You Networked for Successful Innovation?," 2003.
MIT Sloan management review (47:3), 2006, pp. 49-55.

How Firms Make Friends: Communities in Private-Collective Innovation 31


Appendix 2
Extending Private-Collective Innovation:
A Case Study*

Published 2009 in R&D Management,


volume 39, number 2, pages 170-191

Matthias Stuermer
Sebastian Spaeth
Georg von Krogh

Chair of Strategic Management and Innovation


Department of Management, Technology, and Economics
ETH Zurich, Switzerland

The private-collective innovation model proposes incentives for individuals and firms to privately
invest resources to create public goods innovations. Such innovations are characterized by non-ri-
valry and non-exclusivity in consumption. Examples include open source software, user-generated
media products, drug formulas, and sport equipment designs. There is still limited empirical re-
search on private-collective innovation. We present a case study to 1) provide empirical evidence of
a case of private-collective innovation, showing specific benefits, and 2) to extend the private-col-
lective innovation model by analyzing the hidden costs for the company involved. We examine the
development of the Nokia Internet Tablet, that builds on both proprietary and open source software
development, and that involves both Nokia developers and volunteers who are not employed by the
company. Seven benefits for Nokia are identified, as are five hidden costs: difficulty to differentiate,
guarding business secrets, reducing community entry barriers, giving up control, and organizational
inertia. We examine actions taken by the management to mitigate these costs throughout the devel-
opment period.

Introduction where firms and individuals expend private resources


to create public goods innovations. Such innovations
In a private investment model of innovation, firms are characterized by non-rivalry and non-exclusivity
use internal processes to create ideas, knowledge, and in consumption (von Hippel and von Krogh, 2003;
technologies and commercialize these in the market 2006). It is similar to “open innovation,” pertaining
place. Firms appropriate returns from private invest- to models of innovation where firms frequently ex-
ment in innovation through intellectual property change ideas, knowledge, and technology with out-
rights (Granstrand, 1999). This model is contrasted side firms and individuals (Chesbrough, 2003). How-
with the private-collective model of innovation, ever, open innovation does not assume that intellectu-

* The research was supported by the Swiss National Foundation (grants 100012-101805 & 105512-106932). The authors wish to
thank all interview partners for their time and comments, as well as Eric von Hippel and Gideon Markman for providing us with
insightful comments and reviews.

How Firms Make Friends: Communities in Private-Collective Innovation 32


al property rights are forfeited and the resulting inno-
vation is offered to the public for free. Examples of The private-collective innovation model
private-collective innovation model include collabo-
rative composing of music on the Internet by many There are two predominant models of innovation
musicians, the open and collective development of a incentives in the technology and innovation manage-
drug formula for treating malaria, or the creation and ment literature. The private investment model as-
sharing of new designs for sporting equipment among sumes that innovators step forward and invest in in-
sports enthusiasts. An oft-cited example of the model novation if and when they can appropriate returns
is open source software development resulting in from these investments. Intellectual property rights is
products such as Linux, MySQL, or Apache. Open a necessary condition for the model because it safe-
source software comes with licenses that make it guards returns appropriation (Arrow, 1962; Dam,
non-exclusive: the software is free for all to down- 1995). In contrast, the collective action model as-
load, use, modify, and redistribute. Open source soft- sumes that innovators, provided with the right public
ware is also characterized by non-rivalry as one per- subsidy, contribute to public goods innovations
son's use of the product does not diminish anyone (David, 1992; 1998; Stiglitz, 2006). Public goods are
else's benefits from using it. characterized by non-rivalry and non-exclusivity in
Although researchers have examined individuals' consumption. Innovations are made freely available
motivations to participate in open source software de- to all as public goods. Science is often cited as an ex-
velopment, to date there has been limited empirical ample of this model. However, companies have the
examination of a firm's incentives for private-collec- option to free-ride on public goods innovations, such
tive innovation. Moreover, the literature has empha- as, for example, a biotechnology company commer-
sized the benefits the model brings to the innovator cializing scientific knowledge on genetics without
rather than the costs and has not discussed how the contributing research back to the scientific communi-
latter could be mitigated. Research has shown that the ty. Therefore, society elects to subsidize the activity
implementation of new models of innovation often of innovators, e.g., university-based research on the
have unintended consequences, including “hidden human genome funded by the government.
costs” (e.g., Crawford, 1992), and there is a need for Recently, a third model, called the private-collec-
more empirical work on the benefits and costs of im- tive model of innovation incentives, has been sug-
plementing private-collective innovation. gested where public subsidy is absent and where the
In this paper, we advance empirical research on innovator expends private resources for public goods
the incentives and costs of the private-collective innovation (von Hippel and von Krogh, 2003). The
model of innovation. A case study design permits an model is counter-intuitive: why should I make my in-
in-depth investigation of benefits, costs, and actions novations available to all and why pay for something
only partly discussed in prior work (von Hippel and that anyone else can use for free? Generally speaking,
von Krogh, 2006). Our case draws on quantitative in the case of private-collective innovations the inno-
and qualitative data from the creation of the Nokia vator receives higher benefits when contributing to
Internet Tablet. Nokia based the development of this the public goods creation than by only free-riding on
product mainly on open source software and made a its production by others. One aspect is the privately
large part of the research and product development retained tacit knowledge innovators receive through
transparent and accessible as a “public- goods inno- the production of freely available knowledge which
vation.” Outside contributors involving firms and in- distinguishes them from pure users of the explicit
dividuals, unpaid by Nokia, expended a significant knowledge. This implies that firms receive certain
amount of private resources on its development. benefits during the process of creating publicly avail-
The paper is structured as follows. In the next sec- able innovations, while the mere application of such
tion, we discuss the private-collective model for in- knowledge bears less incentives. (see also Grand et
novation incentives. The third section describes the al., 2004; Gächter et al., 2006)
research design, and the fourth section contains the
case description. The fifth section presents the find- Benefits when applying private-collective innovation
ings organized along the topics of benefits and costs
A closer examination of the model outlined in von
incurred in the implementation of the model and
Hippel and von Krogh (2006) and other literature re-
strategic actions to mitigate these costs. Finally, we
veals six complementary benefits for firms to inno-
conclude the paper and outline implications for man-
vate in this manner: the cost of controlling knowl-
agement practice and future research.
edge, learning, reputation gains, and fast and wide-

How Firms Make Friends: Communities in Private-Collective Innovation 33


spread diffusion of innovations, as well as lower cost brough (2003) argued that involving outside firms,
of innovation and manufacturing. First, in the long organizations, and individuals in the development of
run, the cost of protecting knowledge needed to inno- products reduces the direct labor cost in innovation.
vate (Liebeskind, 1996) might outweigh the benefits In addition, when the firm contributes to public goods
of doing so. Often extensive investments in knowl- innovations, such as open source software, it can also
edge management systems are needed to protect in- effectively reuse existing technology found in the
formation which ultimately and inevitably spills over public sphere. Research has shown that, in software
to the public (Foray, 2004; Alavi and Leidner, 1999). development, the reuse of open source software is
For example, although the source code for Sony's ro- considerably higher than the reuse of firm-internal
bot dog Aibo was protected, it was ultimately hacked software, which should have a positive impact on the
and published by Sony's customers. cost of innovation (see Haefliger et al., 2008). How-
Second, innovators that contribute to collective ever, much of this software comes with restrictive
goods innovations benefit from learning from their open source licenses, which requires the firm to make
own and others' contributions. In addition to benefits any combination between this and other software ad-
garnered from the public goods itself (products and here to this license. Hence, while the reuse of such
services), innovators also benefit from learning in the products may reduce costs of innovation, it also
process of creating it (Allen, 1983; Nuvolari, 2004; “forces” the firm to contribute to public goods inno-
Baldwin and Clark, 2006). Thus, it should come as no vations.
surprise that many contributors to open source soft- Sixth, the supply of public goods innovations to
ware projects are computer science students. By pro- the market makes it possible for manufacturers to
viding open source software to the public, contribu- learn about these innovations and reduce manufactur-
tors may get others to use it, test it, and provide feed- ing-incurred fixed costs related to research and devel-
back on how to improve it (Lakhani, et al. 2002). opment. Free access to innovations may incentivize
Some authors have even referred to open source soft- manufacturers to ramp up manufacturing capacity,
ware projects as “epistemic communities,” where pursue economies of scale, and reduce the price of
people create shared knowledge of software develop- manufactured products. Additional benefits to the
ment (Edwards, 2003). customers of the manufactured product may include
Third, innovators may gain a positive reputation enhanced product quality and product warranties
by privately expending resources for public goods in- (Kotha, 1995; Harhoff et al. 2003).
novations (Allen, 1983; Lerner and Tirole, 2002;
Muller and Penin, 2006). For example, Pricewater- Empirical evidence and hidden costs
houseCoopers gains a positive reputation amongst
regulators, peers, and customers when they provide Empirical research on the private-collective inno-
research to the public on practices of corporate gov- vation model is mainly found in the field of open
ernance. The firm's reputation is further enhanced source software development where the focus has
when regulators actively use and reference the re- been on individual contributors and projects (e.g.,
search during standard setting in principles of audit- Shah, 2006; Baldwin and Clark, 2006; Roberts et al.
ing and corporate governance (PricewaterhouseCoop- 2006; Gambardella and Hall, 2006). Research on the
ers, 2005). application of the model by firms is rare with some
Fourth, being the first to contribute a public goods exceptions: Dahlander (2004) explored the network
innovation increases the likelihood of benefiting from effects available to firms that provide open source
fast and widespread adoption of the innovation software to the public. Jeppesen and Frederiksen
(Allen, 1983). As a consequence, firms may gain an (2006) explored users' motives to contribute voluntar-
advantage over competitors stemming from network ily to the development of media products by firms.
effects (von Hippel and von Krogh, 2003). Establish- Henkel (2006) investigated firms that revealed open
ing a “dominant design” or “open standard” onto source software embedded in their devices to other
which the firm can fit other technologies and even firms and found incentives for them to do so. Given
preempt the introduction of competing technology the focus on public goods innovations in these works,
may provide the firm with additional advantages the authors have tended to focus on the cost of for-
(Economides, 1996; see also Economides and Kat- feiting intellectual property rights in the private-col-
samakas, 2006) lective innovation model.
Fifth, by contributing to public goods innovations, Innovation research has pointed to several types
the firm may lower the cost of innovation. Ches- of “hidden costs” in the implementation of new inno-

How Firms Make Friends: Communities in Private-Collective Innovation 34


vation models that are not inherent to or captured in product based on open source software. While Nokia
the models themselves. For example, Crawford is not the first manufacturer to create a mobile device
(1992) and Smith (2004) pointed to the costs of im- based on the open source operating system Linux, its
plementing the accelerated product development strategy of a conspicuous commitment to open source
model in US manufacturing industries. Firms often software and its devotion to building a community of
find that the rapid launching of products that have not volunteers is unique for a multinational corporation
been properly tested leads to costly recalls, or that in the consumer electronics industry. In order to de-
fast product development leads to significant delays termine the distinctiveness of the case, we compared
in pilot- and full-scale manufacturing. Related to it to a wider sample including Sharp, Motorola,
open innovation, Kessler et al. (2000) found some in- Hewlett-Packard, and Sony.
dication that the cost of product development rose In the 1990s, Sharp introduced a Personal Digital
with increasing dependency on external sources of Assistant called Zaurus which ran on an adapted Lin-
technology in the innovation process. Empirical re- ux operating system. Soon after, voluntary developers
search is needed that validates the specific benefits in programmed a variety of applications for the device
the implementation of private-collective innovation, and started a vivid community of developers (among
examines the costs incurred in such innovation, and them one of the interviewees). However, in contrast
identifies firms' strategies to mitigate these costs. We to Nokia, Sharp did not reveal much of its own
contribute to this research by investigating the imple- source code, scarcely supported the community and
mentation of private-collective innovation in the case its interests, and eventually stopped selling the device
of the Nokia Internet Tablet development. in Europe and the United States.
Motorola brought Linux-based cell phones to the
Research design market in 2003. The development of their operating
system was done by MontaVista, a vendor of embed-
The research on the development of Nokia's Inter- ded Linux software, and TrollTech, the provider of
net Tablet focuses on the process of implementing the graphical user interface ‘Qt.’ In contrast to Mo-
private-collective innovation. We investigate benefits torola, Nokia did not rely on a few service providers
in implementation and extend the private-collective to implement the software but collaborated with
innovation model by identifying implementation many, mainly small, software firms in an open fash-
costs and strategies to mitigate these. Research on ion. By using open source software, Motorola expect-
implementation processes typically require longitudi- ed to cut costs and speed up software development,
nal observation (Pettigrew, 1990), prompting a case since they did not pay per-unit royalties and built ap-
study design. In order to obtain insights into the de- plication software on the existing open source soft-
velopment process, we gathered different types of ware (Shankland and Charny, 2003). Similar to
data and performed quantitative and qualitative anal- Sharp, however, Motorola neither revealed source
ysis. Done properly, such combined analysis offers code beyond legal requirements, nor did the firm pro-
valuable insight, as Shah and Corley (2006) have re- vide extensive developer documentation of the soft-
cently argued. In the following sections, we describe ware on its devices. Moreover, their Linux appliances
sampling, data sources, and data analysis. granted no administrator access to the user, inhibiting
the installation or modification of native applications.
Sampling Other mobile devices running embedded Linux in-
cluded HP’s iPAQ and the Sony Mylo. However, as
Our research design is a single-case study de-
we discovered in interviews and in press articles,
manding particular attention to sampling (see Eisen-
these companies retained the software’s source code
hardt, 1989). There are three reasons for selecting a
and we could find no evidence that these firms at-
particular case: fit, distinctiveness, and its revelatory
tempted to build up a community of outside develop-
nature (see also Yin, 1999; Siggelkow, 2007). First,
ers as proposed by the private-collective innovation
Nokia's Internet Tablet development represents a case
model (von Hippel and von Krogh, 2003).
that both serves to explore and extend the private-col-
Third, given the research gap on the implementa-
lective model. The Internet Tablet was created by pri-
tion of private-collective innovation, we also
vate investments by Nokia, building on existing open
searched for a revelatory case. The main criterion for
source products as well as releasing a substantial
selecting a revelatory case is the researchers' access
amount of knowledge (source code).
to a previously inaccessible setting for scientific ob-
Second, the distinctiveness of the case is provided
servation. Establishing ties to Nokia and the develop-
through Nokia's unique approach in producing a

How Firms Make Friends: Communities in Private-Collective Innovation 35


er community surrounding the Internet Tablet, the re- with Nokia employees, 5 with contractors, and 8 with
searchers gained access to a variety of data including unpaid volunteers (see Table 1). In order to protect
documents, interviews, prototypes, and online con- their privacy, interviewees were anonymized. The in-
versations. Shedding light on the reasons for and ef- terviews lasted on average 75 minutes. The interview
fects of this innovation project going open and ab- guidelines included questions on the firm-community
stracting these underlying intentions into a model that relationship, strategies Nokia used to reveal knowl-
can be used in future research, as well as raising the edge and technology to the community, motivation,
attention of practitioners to this mode of product de- and other issues (see Appendix for examples of two
velopment, motivated the selection of the case. distinct interview guidelines). The initial interview
guidelines were updated and enhanced over time, in-
Data sources tegrating and building upon the results of interviews
already analyzed. All interviews were transcribed
This study relies on several sources of data. First verbatim and, using the software Max.QDA, codified
and most importantly, we conducted semi-structured using an open coding technique (for a discussion, see
interviews, allowing participants the opportunity to also Strauss and Corbin, 1998). This led to the cre-
narrate stories, provide anecdotes, and state opinions. ation of 80 codes, which were subsequently merged
Through an initial reading of the mailing lists, rele- and reduced to 12 categories including seven incen-
vant stakeholder groups and data sources in the de- tive and five cost categories.
velopment project were identified as “Nokia employ- The second data source consisted of the project’s
ees,” “Nokia-paid contractors,” and “independent in- user and developer mailing list. The monthly archives
dividuals.” Interviews were conducted with partici- were downloaded from their inception in May 2005
pants from all the stakeholders identified. The initial until the end of December 2006. The archival data
participants were selected from the developer and was examined using the statistical software 'R' in or-
user mailing lists, and subsequent interviewees were der to indicate the size and activity of the community.
identified through snowball sampling (Heckathorn, The resulting statistics are presented later in the text.
1997). In total, 23 interviews were conducted, 10

# Key Role Date Duration Contribution, Function


1 N1 Nokia Nov 15, 2006 89 min Head Open Source Software Operations at Nokia
2 C1 Contractor Nov 22, 2006 71 min Developed Window Manager
3 V1 Volunteer Dec 6, 2006 55 min Linux Distribution Release Manager
4 V2 Volunteer Dec 7, 2006 52 min Developed Mapping Software
5 V3 Volunteer Dec 13, 2006 54 min Developed Music Player
6 C2 Contractor Dec 14, 2006 89 min Performance Measurements and more
7 V4 Volunteer Dec 15, 2006 79 min Developed Swap Memory Feature
8 N2 Nokia Dec 15, 2006 58 min Maemo Product Manager
9 N3 Nokia Jan 12, 2007 92 min Software Architecture Team Leader at Nokia
10 N4 Nokia Feb 19, 2007 83 min GNOME Desktop Developer at Nokia
11 V5 Volunteer Feb 20, 2007 85 min Ported Remote Control Software
12 N5 Nokia Feb 20, 2007 96 min GNOME Desktop Developer for Nokia
13 N6 Nokia Feb 28, 2007 80 min Multimedia Player Developer for Nokia
14 C3 Contractor Mar 5, 2007 81 min GNOME C++ Bindings Developer
15 C4 Contractor Mar 13, 2007 email Software Developer at Contracted Firm
16 N7 Nokia Mar 21, 2007 60 min GNOME Desktop Developer for Nokia
17 V6 Volunteer Apr 4, 2007 71 min Developed Geolocation Software
18 C5 Contractor Apr 10, 2007 email CEO of Contracted Software Company
19 N8 Nokia Apr 11, 2007 69 min X Windows Developer for Nokia
20 N9 Nokia Apr 12, 2007 77 min Maemo Community Manager
21 N10 Nokia Apr 12, 2007 91 min Testing Team at Nokia, Volunteer in Browser Project
22 V7 Volunteer Apr 15, 2007 email Linux kernel patching for Maemo
23 V8 Volunteer Apr 23, 2007 66 min GNOME Foundation Board Member

Table 1: List of interviews including identifier keys, role, date of the interview, duration, and the person's
contribution or function in the context of this case study

How Firms Make Friends: Communities in Private-Collective Innovation 36


Third, one co-author followed the developer mail- torola as the world's largest mobile phone manufac-
ing list over the course of several months and ob- turer (ICFAI, 2005). R&D operations at Nokia have
served the project’s chat channel on IRC (Internet always been scattered across the world, working in a
Relay Chat) which is not publicly archived. Often, in- dispersed, non-hierarchical structure, allegedly to
formal discussion takes place on IRC giving the re- prevent the development of ’tunnel vision.’ In 2003
searchers a feeling for what “really happens” in the and 2004, Nokia suffered a decline in market share of
community. Such online participant observation is its mobile phone business. The company had misin-
rather uncommon in the research on open source soft- terpreted the market demand for ‘clamshell’ devices
ware development but is deemed both necessary and and camera phones and had failed to adapt fast
helpful in understanding the unfolding dynamics of enough to these new developments (ICFAI, 2005).
the Internet Tablet development. No formal analysis However, in 2007, Nokia posted EUR 51.1 bn of net
was done with the data but being immersed in the sales and an operating profit of EUR 8.0bn, spending
community helped to interpret mails and understand EUR 5.6 bn on research and development.
issues raised in the interviews.
Fourth, secondary sources, such as news reports, Internet Tablet History
blogs of Nokia members and volunteers, and corpo-
rate Web sites of Nokia and contracting firms were According to N1, Nokia started to experiment
included in the case study database. For example, with incorporating open source products, specifically
these sources provided additional information on the based on the Linux kernel, into their devices in 2000.
extent to which Nokia sponsors other open source At the same time, the company sought to develop a
projects. Some Web articles were used to get back- device that would take advantage of the increasing
ground information on the Nokia device and potential availability of wireless access points and give access
competitors. We also searched an independent Web to Internet appliances everywhere. N1, the head of
forum for discussions and opinions of users of the In- the software development team, summarized the vi-
ternet Tablet. sion for this new type of mobile device as follows:
“At the same time, totally independently, we had
Nokia and the development of the Internet Tablet another stream of thought which was this kind of cat-
egory of Internet Tablets. The big idea behind that
In this section, we present a short overview of the was really the same way mobile phones liberated
history of Nokia and the Internet Tablet development, voice. Not only houses or offices have phone num-
and we provide a descriptive analysis of Maemo, the bers, but people have numbers. So you can take the
community for the Internet Tablet software platform. phone wherever you go. We have the same vision that
The purpose of this analysis is to confirm the correct- we want to do the same thing with the Internet and
ness of the case to examine the implementation of the Internet use cases. You don’t need to fire up a PC and
private-collective model for innovation incentives. you don’t need to go to your desk. You have this very
Nokia was originally set up in 1865, producing light portable device that gives you access to the In-
pulp and paper. It underwent a series of remarkable ternet. Whether that is browsing, email, chat, VoIP.”
transformations in its business model. In 1967, it (N1)
merged with the Finnish Rubber Works Ltd. and the Inspired by this vision, Nokia designed an overall
Finnish Cable Works, forming the Nokia Corporation software architecture of the operating system based
with four major businesses: forestry, rubber, cable, on open source components and partly adapted these
and electronics. A diversified company, with a prod- components themselves and partly contracted devel-
uct portfolio ranging from tires to television sets, it opers for specific implementation tasks. In 2002, in-
first started producing mobile phones in 1981, manu- dividuals who were active in architecture-crucial
facturing car phones for the first international cellular open source projects were approached by Nokia and
mobile phone network. The first hand-portable phone asked to perform some tasks as contractors for the
sets were introduced in 1987. During the 1990s, company (e.g., C1, C2, and C3). They had to sign a
Nokia focused on telecommunications, especially on Non-Disclosure Agreement (NDA) which prevented
mobile phones based on the then emerging GSM the leaking of much information until the Internet
standard, which had been published by the European Tablet went public.
Telecommunications Standards Institute (ETSI) in The prototype device was first publicized on May
1990. During this period, the company also divested 25, 2005. At the GNOME User and Developer Euro-
its other businesses. In 1998, Nokia overtook Mo- pean Conference on May 31, 2005 Nokia announced

How Firms Make Friends: Communities in Private-Collective Innovation 37


that it would give away 500 devices for about a third behind the platform was to provide open source com-
of the regular sales price to selected software devel- ponents usually deployed on Linux desktop distribu-
opers and donate the sales to the GNOME founda- tions and to adapt and enhance these for the environ-
tion, a not-for-profit organization dedicated to sup- ment of handheld devices (Maemo.org, 2006). Nokia
porting the GNOME graphical desktop environment had downloaded the open source GTK graphical tool-
(Nokia, 2005). The sales of the product did not com- kit and other components such as the GStreamer
mence until November 2005 when the 770 Internet framework for multimedia and modified these to fit
Tablet was officially released with a price of USD the needs of an embedded device with restricted
439. Customers could order the product via a Nokia hardware resources. Nokia also added their own soft-
Internet Tablet dedicated Web site or through the offi- ware developments and parts from independent soft-
cial Nokia Web shop. However, little effort went into ware vendors as proprietary software, protected by
promoting the Internet Tablet to the wider public. The commercial software licenses and released as binary
product was not available through other distribution code only. Figure 1 gives an overview of the software
channels such as local Nokia mobile phone shops2. architecture distinguishing between software pub-
While the Internet Tablet 770 was still being sold lished under an open source software license, com-
throughout 2006, its successor was announced by mercial software components published by third-par-
Nokia’s CEO Olli-Pekka Kallasvuo on January 8th, ty vendors, and Nokia's proprietary components.
2007, at the International Consumer Electronics As-
sociation show in Las Vegas. The successor product
was added to the N-series of Nokia devices, a popular
brand for Nokia’s major communication products,
and marketed to the ‘mainstream’ public as the N800.
On October 17, 2007, the third generation of the In-
ternet Tablet, the N810, was presented in San Fran-
cisco at the Web 2.0 Summit. Nokia executive VP
and head of the company's multimedia business unit,
Anssi Vanjoki, stated "The N810 is the first of these
devices targeted at a 'normal' consumer group, be-
yond the geeks." (Martin, 2007)
The hardware of the Internet Tablet differs in one
main aspect from other Nokia products: it does not
contain mobile phone functionality. It offers a 4.13” Figure 1: Maemo software stack
display with – given its size – an unusually high reso-
lution of 800x480 pixels which, using a stylus, can be
utilized as a touch screen. It connects wireless The Maemo platform uses its own infrastructure,
through Bluetooth (connecting to a mobile phone) or such as a revision control system, a software bug
through a common WiFi to the Internet. It is also pos- tracker, and mailing lists that allow communication
sible to connect to another PC through the integrated between developers and users. In November 2006,
USB port. The device has no hard disk, but flash- the Maemo.org site had 54,000 unique visitors. While
based storage is included which can be extended with the operating system contains proprietary software
external flash storage media. Since the N800, a VGA (some hardware drivers and applications such as the
webcam has been integrated and in the N810 a GPS Opera Web browser), the main software platform,
receiver is also included. Maemo, was open source and developed by the Mae-
mo community.
The Maemo Community A source code repository allows developers to add
The operating system and the software of the new software components to the product and upload
Nokia Internet Tablet are based on the Maemo soft- improved software that resolves problems and bugs
ware platform, an effort led by Nokia and announced in previous versions. The Maemo source code has its
the same day the device was launched. The intention own repository of source code. As of January 2007,
our descriptive analysis shows that 33 developers
added more than 7.2 million lines of code to this
2 A brief anecdote: in 2006, when one of the co-authors of repository, forming the core of the operating system
this paper asked for accessories in such a shop, none of the
shopkeepers were even aware of the product's existence. (although much of this is unmodified code from other

How Firms Make Friends: Communities in Private-Collective Innovation 38


open source projects). Figure 2 visualizes growth in list ranged from 98 to 548 with a median of 328
the source code over time. It should be noted that the (mean 339.8). 19.6% of all mails were sent by Nokia
developers were exclusively employees of Nokia or addresses. Out of the top fifteen posters, only four
formally affiliated with the Maemo project, indicat- had an official email address from Nokia/Maemo.
ing that Nokia retains tight control over the actual Communication on mailing lists is organized in so-
changes that happen to the core system in the soft- called “threads.” Often, a thread is started by some-
ware architecture (see also Kuk, 2006). one asking a technical question, followed by replies
by others who attempt to discuss and answer the
question. 183 threads were started by Nokia affiliates,
while 931 mails were replies to an existing thread.
Non-Nokia participants, on the other hand, started
1,670 threads and posted 4,010 replies. A chi-square
test of the frequency contingency table confirms that
Nokia participants differ significantly in the thread
start/reply frequencies from non-Nokia participants:
Nokia members were more likely to post a reply to an
existing thread rather than start a new one.
Besides discussing development, there is also a
Maemo user mailing list which serves to support gen-
eral user discussion. This list follows similar patterns
although it is less frequented than the developer mail-
Figure 2: Lines of code in the Maemo Subversion ing list. Its number of postings per month ranged
code repository from 3 to 253, with a median of 133 and a mean of
127 mails. As can be expected, user-related discus-
sions started to pick up after the device was intro-
In order to help identify and remove software duced to the market in November 2005, whereas de-
bugs, Maemo also has a so-called “bug tracker” of its velopment discussions preceded this point. Our anal-
own. This is used to enter software errors, - or bugs, ysis shows that 511 participants contributed to the
and keep track of the bug-fixing process. Our analy- list, 6.46% of them from Nokia. Only two out of the
sis shows that in June 2008, this tracker contained top fifteen posters identified themselves as belonging
3228 bugs of which 1133 bugs were marked as to Nokia. 10.9% of all mails were sent from Nokia
“open,” meaning that they were waiting to be fixed. employees. These started 29 new threads and replied
The Maemo project infrastructure also offers a to 220 existing threads. Non-Nokia participants, on
repository, garage.maemo.org, where people indepen- the other hand, started 775 new threads and sent
dently of Nokia can register their projects and use the 1,510 mails to existing ones. Also on the user mailing
developer infrastructure for free. Our analysis shows list, a chi-square test confirmed that Nokia affiliates
that in June 2008, Garage consisted of 615 active were significantly more likely than non-Nokia partic-
projects and 12,446 registered users. Projects led and ipants to reply to existing threads rather than start
contributed to by independent individuals ranged new ones.
from GPS navigation solutions and system adminis- Our brief examination of the Maemo community,
tration tools to an Electronic Flight Information Sys- including source code development and technical and
tem for use with small aircrafts. user discussions, supports the statement that the de-
Most communication happens through mailing velopment of the Nokia Internet Tablet is an imple-
lists. For discussion purposes, Maemo has a develop- mentation of private-collective innovation. Large
er list where actual technical development issues are parts of the product are a public goods innovation and
discussed and a user mailing list for discussing issues Nokia employees as well as external individuals and
related to the operation and use of the product. We organizations expend considerable private resources
examined the Maemo developer mailing list from its (time, knowledge, and technology) to contribute to
inception in June 2005 until December 2006. In total, the innovation. In the mailing list geared towards de-
832 participants identified by their email address con- velopment, more than 80% of all emails were sent by
tributed to the list, with 79 participants (9.5%) post- non-Nokia affiliates. In the next section, we present
ing from an official @Nokia.com or @maemo.org the findings from the case study.
email address. The monthly number of postings to the

How Firms Make Friends: Communities in Private-Collective Innovation 39


Findings benefits derived above from the existing literature
and additionally identify a benefit in the case: faster
The following section presents findings on the im- time-to-market. In addition, we extend the model of
plementation of private-collective innovation in the private-collective innovation, highlighting the hidden
case of the Nokia Internet Tablet development. The costs related to the implementation of private-collec-
aim of this section is twofold: first, we illustrate a tive innovation and strategies to mitigate these costs.
case of private-collective innovation with empirical The benefits and costs of this extended model are
data, providing specific benefits for the company in- summarized in Tables 2 and 3.
volved. The findings confirm the six conjectures on

Benefits Findings in the Nokia case


Low knowledge protection costs Revealing source code rather than protecting it; however, undetermined
costs for revealing.
Learning effects Collaboration with external firms and individuals
Reputation gain Increased attraction of Nokia as an employer and for building their own de-
veloper community
Adoption of innovation Standard setting of the platform configuration
Increased innovation at lower costs Reuse of open source software, outsourcing of software testing and bug fix-
ing and maintenance to open source communities. Experimentation and con-
tributions of new applications by lead users
Lower manufacturing costs No licensing fees for software platform
NEW: Faster time-to-market Tapping of distributed technology expertise and high flexibility of software
platform

Table 2: Benefits in the implementation of Private-Collective Innovation and findings in the Nokia case

Cost Findings in the Nokia case Mitigation strategy


Difficulty to differentiate Released source code can be Partial revealing of source code to retain
reused by competitors control of look and feel
Guarding business secrets Plans for new products Selective revealing of future plans and pro-
tection of information through NDAs
Reducing community entry barriers Investments for Software Develop- Sharing the costs with other actors in the
ment Kit, preview version of plat- community
form, device program, staff for com-
munity management, and increased
communication effort
Giving up control Development direction such as Hiring of key developers and participation
scope of functionality of open source in upstream communities. No single vendor
projects is controlled by external controls platform
parties
Organizational inertia Required internal restructuring of Adapting and opening up processes
processes

Table 3: Costs and possible mitigation strategies in the implementation of Private-Collective Innovation

How Firms Make Friends: Communities in Private-Collective Innovation 40


Benefits in implementation project within Nokia, seeking to explore new possi-
bilities with open source components in order to
First, considering the reduced cost of knowledge
build new competences. Although the company could
protection, product development managers at Nokia
have chosen a commercial open source software inte-
were conscious that costs stemming from protecting
grator such as Red Hat or Montavista, delivering
proprietary knowledge and costs of voluntarily re-
ready-made solutions in order to release a new prod-
vealed knowledge imply a trade-off. For the Internet
uct quickly, Nokia decided to learn about the novel
Tablet platform, Nokia decided to keep some compo-
technologies internally in order to gain the potential
nents proprietary, while developing other components
for future innovations. N3 reported:
as open source software. According to the interview
“[We wanted to make] something that is more
with N1, about 25% of the software is unmodified
than just putting a product into the market through
open source software. Another 50% consists of exist-
learning about the possibilities of leveraging open
ing open source code to which Nokia made improve-
source in deeper and more substantial ways. As a re-
ments or adaptations and which was again released
sult, we didn’t take the quick and simple approach,
under an open source license. Only 25% comprises
we went a little bit deeper and we learned more.”
closed code that Nokia either implemented from
Building upon technologies developed mainly by
scratch or which is closed commercial (proprietary)
external contractors and volunteers implied that many
software from a commercial vendor. N1 stated a rea-
outsiders, rather than the internal research and devel-
son for not releasing source code under an open
opment staff, had the most intimate knowledge of
source license was that Nokia wanted to keep control
certain software components. The interview with
of the look and behavior of end-user applications,
contractor C2 revealed that in the beginning, Nokia
rather than fearing a loss of valuable knowledge. Dis-
tried to solve technical issues internally. However,
closing the knowledge was not free either, all source
when the company faced severe time pressure on a
code had to be checked by Nokia's department to en-
product launch date, it would outsource the request to
sure that Nokia did not reveal intellectual property or
a trusted external firm which specialized in the area
patents they did not have the right to. Altogether, it
of technical development. For specific parts of the In-
remains unclear whether Nokia saved costs by reveal-
ternet Tablet software, Nokia contracted several small
ing knowledge rather than spending effort to protect
enterprises with experience in open source compo-
it. Nokia seems to see it as beneficial:
nents (see Table 6). Working with these experts and
“We have evidence that some of our competitors
listening to feedback from the Maemo community
are now looking at our code and they are investigat-
helped Nokia to rapidly learn new open source tech-
ing if they could use our code in their products. You
nologies and how those were produced by their com-
might say that we help them now to get their products
munities.
out fast. [...]But if we had not put it out then we could
Nokia’s strategy was deliberately focused on
not have used the OSS communities who have al-
small companies, rather than on a few, large contrac-
ready helped us to develop that code.” (N1)
tors. Collaborating with small firms all over Europe
Second, Nokia obtained learning effects by ex-
enabled Nokia to fill gaps in the company-internal
perimenting with open source software technologies
technical knowledge needed for development, while
and by collaborating with contractors, firms, and vol-
still retaining control of the core software architecture
unteers. The platform had started as a research
by having internal experts in these areas (C2).

Company Expertise Country


KernelConcepts GPE and Embedded Linux Germany
OpenedHand Matchbox United Kingdom
Collabora Telepathy United Kingdom
Imendio GNOME and D-BUS Sweden
Fluendo GStreamer Spain
Movial Scratchbox Finland

Table 6: Some contractors to Nokia in the software development process

How Firms Make Friends: Communities in Private-Collective Innovation 41


In addition to learning about the technologies in for jobs. Thus, most of the Nokia employees working
use and acquiring the skills to work with them, Nokia on the Internet Tablet were previously active partici-
learned to cooperate with a diverse community of pants in open source projects and known to the Nokia
employees, volunteers, and contractors rather than re- managers through prior collaboration. This fact re-
lying on contractors only who could be forced to flects that for many, Nokia became an interesting
keep schedules and timelines: company to work for, as C1 testified:
“It’s all about the process... You develop this “At that time, it was a dream come true. I did not
openly within the communities and you try to syn- have a good job. I was spending all my spare time
chronize your own work with the heartbeat of the hacking Matchbox [window manager of the Internet
communities. Some companies now understand this Tablet]. I was really enjoying it. And then you’re giv-
better than others. We certainly have done our learn- en a chance to get paid to do that full time. It was
ing. We have made some mistakes too on this front.” pretty fantastic and an amazing piece of luck.”
(N1) According to interviews with Nokia development
For example, one of Nokia's major learnings con- managers, Nokia employees selected through the
cerned the early decision to independently continue community were highly motivated to continue to
development of the GIMP Toolkit (GTK), a collec- work on technologies they already knew. Possibly in-
tion of software constituting the core of the graphical trinsic motivation, such as fun - often a primary cause
interface. Among others, C3 stated “the disadvantage of contributions to open source development - played
of doing this of having a forked or large patch was a role in their continued high-level efforts (see e.g.,
experienced by Nokia.” By following a separate de- Luthiger Stoll, 2006; Torvalds and Diamond, 2001).
velopment stream Nokia became disconnected from Fourth, the private-collective innovation model
the code maintenance effort by the community and, proposes that being first to contribute a public goods
thus, realized they had to move their changes back innovation increases the likelihood of fast and wide-
into the main open source project as architecture team spread adoption of the innovation. According to the
leader, N3, reported: interviews, since the community was already familiar
“It was also a learning field that we had to go with underlying technologies, adapting existing ap-
through. If we had known then what we know today, plications from other projects to the Internet Tablet
we would have been able to do it without such a platform proved a relatively easy task. Nokia also in-
large patch. We would have been able to do more di- vited competitors to participate in the creation and
rectly upstream in technically better ways with less use of their platform, citing a “the more the merrier
effort for changing code.” approach.” By initiating a vendor-independent em-
Third, the growing commitment to open source bedded software platform intended for use in other
software development led to a reputation gain for mobile devices, Nokia made it easier for volunteers,
Nokia, and interviewees V2, V3, and V4 suggested contractors, and competitors to contribute. Spreading
this led to the increasing attachment of volunteers the innovation and inviting others to participate was
and recruiting benefits. Particularly the openness of seen as crucial:
the Maemo platform encouraged volunteer users and “We believe the world is changing and the com-
developers to buy such a device, to improve the oper- petitive advantage comes from how many others you
ating system, and to create applications on top of it. can get to participate in this network.” (N1)
Nokia was accredited with pioneering development In fact, in July 2007, Intel announced they were
of an open embedded platform by users who enjoyed adopting the Internet Tablet's user interface frame-
running applications from third parties as for example work Hildon into their new product category called
V6 stated: Mobile Internet Devices (Paul, 2007), which will
“Yeah, I definitely have a more positive view of eventually lead to a higher developer and user basis
Nokia this way. Especially, I think they are handling of Nokia's Maemo platform.
the open source interaction quite well. I think they Fifth, by contributing to public goods innovations
are quite a good open source citizen.” firms can lower the cost of innovation. Building on
For Nokia, recruiting active open source contribu- existing and mature technologies which could be in-
tors enabled them to select the best individuals based tegrated into the new hardware, Nokia enabled the de-
on their prior contribution to the various projects that velopment of a solid, yet cheap operating system for
constituted the development of the Internet Tablet. In embedded platforms. Collaborating within existing
some cases, Nokia directly contacted skilled open open source projects allowed Nokia to benefit from
source software developers and invited them to apply the collective programming efforts:

How Firms Make Friends: Communities in Private-Collective Innovation 42


“So what is happing in the D-BUS, in the GTK, in Sixth, supply by anyone of public goods innova-
the GStreamer, in the Linux kernel is that I put two tions to the market enables manufacturers to learn
guys there, IBM puts two guys, Motorola maybe puts about innovations and, thereby, reduce costs in man-
one guy, maybe Novell puts a couple of guys. So for ufacturing. This conjecture indicates particular bene-
the price of two guys, I get four or six guys working fits private-collective innovation in software offers to
on the same problem.” (N1) computer hardware manufacturers (von Hippel and
Additionally, the company benefited from volun- von Krogh, 2003). Through choosing a software plat-
tary contributions leading to enhancements of the de- form which is available under an open source license
vice. The volunteers contributed several innovations as indicated, Nokia reduced fixed costs related to re-
including applications, user interface improvements, search and product development. Nokia manufac-
translations, bug reports and fixes, testing different tured and sold the Internet Tablet, but the product's
peripherals, and making feature requests. Letting vol- functionality and, thus, ability to fulfill user needs
unteers experiment with the software also created were to a large extent shaped by the users themselves.
“proof of concepts” which enabled innovations previ- User-developed applications, such as mapping and
ously seen as unrealistic by Nokia engineers (V3). navigation software, could easily be installed by the
One example of this is the swap memory enhance- end users themselves for free, keeping Nokia's costs
ment (using the flash memory as extended virtual down. Interestingly, in addition to fixed cost reduc-
memory) which was initiated by volunteer V4 and in- tion, Maemo also has a positive impact on variable
cluded in a subsequent official version of the Internet costs in manufacturing since Nokia did not have to
Tablet's operating system. Thus, the likelihood of pay a per-device license fee to an intellectual proper-
finding a “killer application” (see e.g., Downes and ty owner. For example, at the beginning of 2006, a
Mui, 1998) in the process (by evolution or sheer comparable proprietary operating system, Symbian,
luck) increased. An open source software developer demanded USD 7.5 per device for the first 2 million
observed: units.
“I think from my point, if you let people change In addition to confirming theoretical conjectures
things [...] and document them and open them up so on benefits, one more benefit emerged in the study
people can hack their own stuff, you never know what that should be considered crucial for private-collec-
is going to happen, what kind of things people are tive innovation: faster time-to-market. Using exter-
going to write for your device which ultimately could nal, modular technologies not only impacted on costs,
make it sell millions and millions if someone writes it also led to the creation of a new operating system
the killer application for it.” (C1) working on a new hardware platform in a short time.
As such, both the costs of innovation were kept According to interviews with N1 and N3, this fast de-
low and ideas which would not have been developed velopment created flexibility which, combined with
otherwise could be tested and integrated: user feedback of pre-releases, allowed for a quick
“I can develop, say, twenty ideas a day and this time-to-market compared to other devices the compa-
community can develop a hundred ideas a day. So it’s ny had launched.
more important to be part of the community with a However, integration of the numerous open
hundred ideas than by yourself with twenty ideas.” source components into a executable environment is
(N1) very challenging as, for example, C3 explained. One
Investigating the software architecture, we found way to rapidly tap into this knowledge of open source
most contributions by volunteers were separate appli- communities was achieved by Nokia's strategy of
cations which could be installed independently from contracting open source developers and small firms –
the core operating system. The interviews showed the “bridges between Nokia and the communities”
that volunteers who made such contributions showed (N3):
high commitment to and responsibility for their work, “It’ll get done quicker and probably better if they
listening to user feedback and, in some cases, when pay us to do it. [...] Over the years we have been in-
others demanded it, even enhancing their software volved, we have so much experience and knowledge,
against their personal belief of the usefulness of the we know all the tricks. We know how to get things
features (explained by V2). Through their contribu- like X and Matchbox up and running quickly on
tions and feedback from users, volunteers slowly hardware. We are just basically selling that knowl-
gravitated towards more development work in the edge as well. Although they could very well likely fig-
community. ure it out themselves, we just can get them there a lot
quicker.” (C1)

How Firms Make Friends: Communities in Private-Collective Innovation 43


Costs of implementation and strategies to mitigate these ment of new software features, they contracted sever-
al small enterprises or even motivated individual vol-
While the benefits of private-collective innovation
unteers with unique knowledge critical to the devel-
have been spelled out previously, the hidden costs of
opment of the Internet Tablet to sign a NDA. The
implementing the model have been neglected in pre-
NDA protected Nokia against the leakage of informa-
vious work or remain unknown. In this section, we
tion about planned new product developments. The
present the findings from the case study along five
agreements effectively created a three-tiered commu-
categories of costs together with Nokia's strategies
nity of people “who knew,” which meant Nokia in-
(where applicable) to mitigate these. First, when soft-
siders, “those with a clue,” which meant contractors,
ware is freely available even to direct competitors, it
and the “regulars” (C3). However, this information
is possible for current and future competitors to de-
imbalance led to strong tensions for Nokia with its
sign clones that look like and behave in a very similar
external community members. Thus, Nokia hired N9
way to the original product. Competitors are in the
e.g., in order to update the technical roadmap of the
position not only to imitate but to replicate the prod-
Internet Tablet with all the information necessary for
uct (see Kogut and Zander, 1992). This potential lack
software developers.
of differentiation of products represents a cost to the
Third, in order to facilitate the increasing involve-
firm as it forfeits an opportunity to gain competitive
ment of volunteers, Nokia needed to carry the costs
advantage (Granstrand, 1999). Such reuse of open
of reducing community entry barriers. The compa-
source software is completely legal and, in fact, took
ny invested in the creation of a Software Develop-
place as explained in this illustrative example:
ment Kit that enabled new volunteers to easily start
“The OpenMoko actually bases on our stuff. [...]
development for the Internet Tablets. It is common
the Moko window is a kind of base window for their
practice for software manufacturers to provide such
applications I think. There in the comments of the
an SDK at high costs (Jacobson et al., 1999). Howev-
source file you can find that this is based on the
er in the case of the Internet Tablet, Nokia offered the
Hildon window by Nokia. They even have this copy-
development tools for free. In order to allow volun-
right “Nokia corporation.” So they are based on our
teers to adapt their software for upcoming platform
stuff. But, on the other hand, our stuff is not that rev-
releases, Nokia also offered a development snapshot
olutionary. I mean that’s how it goes. We base on
of their work-in-progress (often including software
somebody else’s stuff too.” (N4)
for yet unannounced features) which could be used to
Nokia’s strategy to mitigate this cost was to selec-
ensure that an application would also run on future
tively open up its software development. Figure 1
releases. Employees were sent to related conferences
(see above) visualized how Nokia revealed the mid-
in order to increase awareness of the platform and an-
dle layer of software, the basic infrastructure, under
swer questions from current and future volunteers
an open source license, while keeping parts of the
and contractors. Additional staff, such as “community
bottom layer (hardware specific software) and much
representative” N9, were hired in order to communi-
of the user-visible applications under their own pro-
cate between Nokia internal developers and the exter-
prietary license. N1 argued that it wanted to ensure a
nal community members. In order to mobilize more
unique “Nokia look” by retaining control over appli-
volunteers to join the Maemo community, Nokia sold
cations, for example of the device's email program. In
1,500 heavily subsidized devices to active open
doing so, they kept crucial parts closed (e.g., power
source developers. While such direct costs by Nokia
management and other hardware drivers) and pre-
cannot be mitigated easily, investments in community
vented replication of these parts by competitors. In all
building, knowledge diffusion, and marketing may be
the interviews conducted with volunteers, the respon-
lower in the future through sharing the effort with
dents demonstrated understanding for Nokia's deci-
other community members. Some interviewees ex-
sion to selectively reveal source code, although some
plained that, for example, with increasing popularity,
stated their strong preference for releasing all soft-
well-integrated Maemo community members started
ware under an open source license.
to support new volunteers who were getting involved
Second, Nokia was concerned with potential costs
in the development process.
stemming from losing business secrets, such as plans
Fourth, by contributing source code to open
for future devices. In order to mitigate these costs,
source projects which were not managed by the com-
Nokia revealed knowledge in terms of software but
pany, Nokia gave up control of the future develop-
kept future key product innovations and business fig-
ment direction of core technologies deployed in the
ures, such as devices sold, number of employees, or
company's hardware. According to N1, the company
investments expended, confidential. For the develop-

How Firms Make Friends: Communities in Private-Collective Innovation 44


traded having full control of the technology for par- forking.3” When attempting to balance control and
ticipation in joint development, thus benefiting from openness, Nokia considered the threat of defection of
sharing the cost of innovation with outsiders. For ex- volunteers to other open platforms used by emerging
ample, GTK was originally intended for use on desk- competing “open devices.”
top PCs, and according to interviewees N1, C1, V2, Fifth, since the private-collective model of inno-
V4, and N3, it was necessary to adapt GTK to the vation incentives breaks with the traditional private-
low resource environment of the Nokia Internet investment model that is prevalent in industry, it is
Tablets by decreasing memory consumption. In order reasonable to expect that the implementation of the
to regain some control of these critical software com- model in an established firm incurs costs of organi-
ponents, Nokia hired key developers from the GTK zational inertia (see Sorenson and Stuart, 2000).
community and contracted small enterprises with Since the Internet Tablets also include software writ-
deep knowledge in this area. According to N3, the ten by third-party vendors, Nokia needed to ensure
contribution of code improvements and modifica- that their software revealed to the Maemo community
tions, combined with a meritocratic organization of did not infringe on intellectual property rights. As
the projects involved, gave Nokia enough influence mentioned above, interviewees commented that
on the direction of software development. Yet, the Nokia's internal approval of open source software
practice of contracting developers from incumbent was slow and bureaucratic. In addition, the complex
communities such as GNOME in order to gain repu- internal processes of a large multinational organiza-
tation and control raised concerns from some Maemo tion often made it difficult for Nokia's developers to
community members as to how Nokia would influ- collaborate with external open source projects. One
ence the future of the projects. For example: interviewee commented:
“Obviously when they are sponsoring a project, “The biggest problem is that Nokia is a very big
then they are going to have some control over the di- company and that Maemo is a very small group. I
rection and what gets into it. [...] I think, they truly think it’s like ten or twenty people in total working on
want to work with the community and want to keep it. A lot of software which they use on the 770 is de-
them happy as well, so it’s all a bit of give and take, I veloped in some other groups in Nokia. For example,
suppose.” (C1) the movie plug-ins and MP3 plug-ins. Those are the
The situation was different for projects that were same I think as they use on their phones in Symbian.
initiated and controlled by Nokia itself. According to So that’s a Symbian decision. You can’t tell the Sym-
C3, Nokia granted only write-access to its software bian people to use that bug system instead of the in-
repository to its employees, thus retaining control of ternal bug system. They would say, 'Why?' ” (V1)
the actual published source code. It is worth noting Nokia employees also commented that the inter-
that this strategy created some tensions in the com- nal Nokia firewall would not allow connection to the
munity, with the non-Nokia members complaining official developer chat room of the Maemo communi-
that they could not help or contribute if they did not ty, which was located outside Nokia’s network. In or-
know about the future direction of the software de- der to mitigate the costs of organizational inertia,
velopment or about the estimated timeframes of fu- Nokia employees stated that they would work on a
ture software releases. The Maemo community tried case-by-case basis to remove obstacles. Employees
to influence Nokia's behavior, mostly through provid- were, for example, assigned to make sure that entries
ing intense feedback on mailing lists. For instance, in in the external, publicly accessible bug tracking sys-
2000 Nokia released a new version of the Maemo tem would be paid attention to.
platform as a binary download as V4 remembers. Altogether, Nokia managers were conscious of the
However, Nokia did not publish the source code at trade-offs between revealing knowledge and technol-
the same time but staved off the community by argu- ogy and the benefits from participating in private-col-
ing that the legal department needed three more lective innovation. The following statement summa-
weeks to clear the code for reasons of intellectual rizes well the experiences had regarding the trade-
property protection. A strong, negative reaction from offs between cost and benefits in the model:
the community taught Nokia to proceed differently “Some people might say that one of the problems
next time, releasing both binary and source code si- is that you are leaking and giving out your secrets
multaneously. Sanctions on the part of the community
were mainly “withdrawal of love” and the “threat of 3 A so-called “fork” results when “dissatisfied program-
mers” copy the original source code from a project and
continue its development in an alternative, competing
project.

How Firms Make Friends: Communities in Private-Collective Innovation 45


and so forth, but it’s more like a trade-off. What is Nokia enjoyed reputation benefits both as an attrac-
more important to you: to give some of your secrets tive employer and as an “open-source friendly com-
an internal work-out or how much help in creating pany” amongst open source software developers who
these products you get for free. I think, if you calcu- contributed software to the platform. By creating the
late, you are far more on the positive side when you vendor-independent GNOME embedded platform
decide to share.” (N1) and inviting competitors to contribute and use the
software platform, Nokia facilitated the adoption of
Discussion and conclusion the software as a common platform for embedded de-
vices. Moreover, Nokia built upon existing technolo-
In this paper, we identified a research gap in the gy and took advantage of users' previous contribu-
literature on private-collective innovation (von Hip- tions to open source software projects. By taking ad-
pel and von Krogh, 2003; 2006): little is known about vantage of existing open source components, Nokia
the implementation by firms of the private-collective managed to create a complete operating system with
model of innovation incentives. We argued that the only a handful of developers and was able to inte-
implementation of the model will be associated with grate ideas and improvements from other Maemo
benefits, “hidden” costs, and strategies to mitigate community members. In terms of costs, Nokia's man-
these. In order to examine and extend the model ufacturing could benefit from low cost software de-
through empirical work, we employed a case study velopment and avoided paying the common per-de-
design. Using quantitative and qualitative data, we vice license fees. Finally, our study found that in-
demonstrated that the development of the Internet creased flexibility and a faster time-to-market is a
Tablet is a case of private-collective innovation. benefit in implementing private-collective innova-
Next, we analyzed data from several sources in order tion.
to identify benefits and costs incurred in the imple- Thus, although the six benefits at first glance
mentation of private-collective innovation and strate- make it rational for the firm to choose private-collec-
gies by Nokia to mitigate these costs. tive innovation amongst alternative models, previous
Nokia launched the Internet Tablet as a private- work also raised the awareness of unintended conse-
collective innovation project and as a low-cost probe quences or “hidden costs” resulting from implement-
(Brown and Eisenhardt, 1997). At the time of product ing private-collective innovation. For example, in or-
launch, neither a product category nor a market for der to obtain outside contributions, a firm may need
these devices existed. Rather than following existing substantial investments in documenting the released
market demand, Nokia targeted technology pioneers software, training potential contributors, and develop-
to find out who would use the Internet Tablet and ing online tutorials. These costs of implementation
how it would be used in real-life applications (similar may offset the benefits to private-collective innova-
to what Zander and Zander, 2005, called “exploiting tion4.
the inside track’). Nokia opened up the product's soft- We briefly reviewed literature that indicated “hid-
ware using externally developed open source tech- den costs” associated with the implementation of
nologies, allowed for and encouraged contributions novel innovation models (Crawford, 1992; Kessler et
by outsiders and, in the process, created a new market al. 2000; Smith, 2004). The study found that the im-
for a product it had envisioned. When the product plementation of the private-collective model of inno-
proved successful, Nokia moved from targeting tech- vation incentives in Nokia's development of the Inter-
nology pioneers towards the mainstream market with net Tablet incurred costs and that the company found
the subsequent release of the Internet Tablet N800 strategies to mitigate these. In particular, the potential
and N810. lack of product differentiation as well as revealed
This study confirmed most incentives to innovate business secrets incurred costs to the company. Nokia
identified in previous literature (von Hippel and von mitigated these costs by selectively revealing knowl-
Krogh, 2003; 2006). It remains inconclusive whether edge and technology. Another cost concerns the low-
Nokia saved knowledge protection costs by revealing ering of entry barriers to the Maemo community. The
most of their software. However, the company gained company invested in several measures to reduce such
skills and knowledge through collaborating with out- barriers, including discounted devices and a free
side volunteers and contractors and, thus, also acted
as a system integrator coordinating a loosely coupled
4 In addition, Osterloh and Rota (2004) pointed out that the
network of component providers (see also Brusoni et mere presence of a firm in private-collective innovation
al., 2001). As the main contributor in the project, may “crowd out” intrinsic motivation by voluntary con-
tributors (e.g., fun and enjoyment).

How Firms Make Friends: Communities in Private-Collective Innovation 46


SDK, as well as allocating employees responsible for is largely based on tacit knowledge acquired through
community communication in order to attract further extensive and costly apprenticeship (e.g., luxury
volunteers. Moreover, using technologies that are goods), volunteers who join product development
partly maintained externally has the advantage of products may be rare. Future research will also have
shared innovation costs but implies giving up full to investigate the impact of company age and size on
control of the future development of that technology. the innovation incentives in the private-collective
Through hiring key developers of software for the model. As we found in the case of an established
product, Nokia regained some influence and control. company, the process represented costs of organiza-
Internal processes sometimes proved inadequate to tional inertia.
enable a transparent and open development process, Managers who want to experiment with flexible
incurring some delays and costs as well as frustration solutions, while keeping their own product develop-
in the community. Nokia acknowledged this chal- ment costs low, should investigate ways to implement
lenge that the interviewees described as a “learning private-collective innovation. Sharing development
process.” costs and enabling contributions from third parties, as
The extended model of private-collective innova- well as boosting organizational learning, are powerful
tion provides additional insights for researchers. reasons why the model is attractive in practice. How-
First, implementing private-collective innovation ever, there are potential “hidden costs” in implement-
may enhance organizational learning and renewal, in ing the model. Learning from Nokia's successful ap-
addition to being a form of “open product develop- proach, managers should think ahead about possible
ment” (Chesbrough, 2003). During the development costs and create strategies to mitigate them. The ex-
of the Internet Tablets, Nokia adapted and learned to perience from the development of the Internet Tablet
work with a community of volunteers. An open re- provides possible mitigation strategies.
search question is to what extent such learning can
enable firms to work with outside volunteers across
generations and categories of products. Second,
Harhoff et al. (2003), modeling the payoff for innova- References
tors to freely reveal their innovations find that one of
four conditions, “greater generality of the Alavi, M. & Leidner, D. E. (1999), 'Knowledge Management
Systems: Issues, Challenges, And Benefits', Communications
knowledge,” reduces the likelihood of free-revealing of the AIS 1(1), 1-37.
by the innovator (see also Muller and Pénin, 2006).
However, this case study showed that much knowl- Allen, R. (1983), 'Collective Invention', Journal of Economic
Behavior and Organization 4(1), 1-24.
edge revealed proved to be generic frameworks: it
laid the foundation for a generic embedded Linux Arrow, K. J. (1962), 'The Economic Implications of Learning
desktop environment. More specific components, by Doing', The Review of Economic Studies 29(3), 155-173.
such as power management and some end-user appli-
Baldwin, C. Y. & Clark, K. B. (2006), 'The Architecture of
cations were kept proprietary. This contradicts Participation: Does Code Architecture Mitigate Free Riding in
Harhoff’s et al. findings and the issue of what kind of the Open Source Development Model?', Management Science
knowledge is likely to be revealed under what condi- 52(7), 1116-1127.
tions requires more attention in future research. Brown, S. L. & Eisenhardt, K. M. (1997), 'The Art of Continu-
Limited by its design, the current study can only ous Change: Linking Complexity Theory and Time-paced
generalize findings to theory. The extended private- Evolution in Relentlessly Shifting Organizations', Administra-
collective innovation model provides a set of bene- tive Science Quarterly 42, 1-34.
fits, costs, and mitigation strategies that must be test- Brusoni, S.; Prencipe, A. & Pavitt, K. (2001), 'Knowledge
ed on a larger sample in future research using cross- Specialization, Organizational Coupling, And The Boundaries
sectional as well as longitudinal designs. Thus, it will Of The Firm: Why Do Firms Know More Than They Make?',
Administrative Science Quarterly 46(4), 597-621.
be important to garner insights on technological, in-
dustry, and market conditions that provide different Chesbrough, H. (2003), Open innovation. The new imperative
levels of benefits and costs of innovation. For exam- for creating and profiting from technology, Harvard Business
ple, in industries of non-virtual goods or where prod- School Press.
uct development constitutes a minor share of fixed Crawford, C. (1992), 'The Hidden Costs of Accelerated Prod-
costs in production (e.g., cement manufacturing), uct Development', Journal of Product Innovation Manage-
companies may find it more attractive to pursue pri- ment 9(3), 188-199.
vate-investment innovation. Moreover, if innovation

How Firms Make Friends: Communities in Private-Collective Innovation 47


Dahlander, L. (2004), 'Appropriating the Commons: Firms in Henkel, J. (2006), 'Selective Revealing In Open Innovation
Open Source Software', Technical report, Chalmers University Processes: The Case Of Embedded Linux', Research Policy
of Technology. 35(7), 953-969.

Dam, K. (1995), 'Some economic considerations in the intel- von Hippel, E. & von Krogh, G. (2006), 'Free revealing and
lectual property protection of software', Journal of Legal Stud- the private-collective model for innovation incentives', R&D
ies 24(2), 321-377. Management 36(3), 295-306.

David, P. (1998), 'Knowledge spillovers, technology transfers, von Hippel, E. & von Krogh, G. (2003), 'Open Source Soft-
and the economic rationale for public support of exploratory ware and the "Private-Collective" Innovation Model: Issues
research in science''Background Paper for the European Com- for Organization Science', Organization Science 14(2), 209-
mittee for Future Accelerators'. 223.

David, P. (1992), 'Knowledge, property, and the system dy- ICFAI (2005), 'Nokia and the Global Phone Industry', Case
namics of technological change', Proceedings of the World study by ICFAI Center for Management Research, Hyderabad,
Bank Annual Conference on Development Economics, 215- India.
247.
Jacobson, I.; Booch, G. & Rumbaugh, J. (1999), The Unified
Downes, L. & Mui, C. (1998), Unleashing the Killer App: Software Development Process, Addison-Wesley, Boston.
Digital Strategies for Market Dominance, Harvard Business
School Press. Jeppesen, L. B. & Frederiksen, L. (2006), 'Why Do Users
Contribute to Firm-Hosted User Communities? The Case of
Economides, N. (1996), 'The Economics Of Networks', Inter- Computer-Controlled Music Instruments', Organization Sci-
national Journal of Industrial Organization 14(6), 673-699. ence 17(1), 45-63.

Economides, N. & Katsamakas, E. (2006), 'Two-Sided Com- Kessler, E.; Bierly, P. & Gopalakrishnan, S. (2000), 'Internal
petition Of Proprietary Vs. Open Source Technology Plat- vs. External Learning In New Product Development: Effects
forms And The Implications For The Software Industry', Man- On Speed, Cost, And Competitive Advantage', R&D Manage-
agement Science 52(7), 1057-1071. ment 30(3), 213-224.

Edwards, K. (2003), 'Epistemic Communities, Situated Learn- Kogut, B. & Zander, U. (1992), 'Knowledge of the Firm, Com-
ing And Open Source Software Development', Technical re- binative Capabilities, and the Replication of Technology', Or-
port, Technical University of Denmark. ganization Science 3, 383-397.

Eisenhardt, K. M. (1989), 'Building Theories From Case Study Kotha, S. (1995), 'Mass Customization: Implementing the
Research', Academy of Management Review 14(4), 532-550. Emerging Paradigm for Competitive Advantage', Strategic
Management Journal 16, 21-42.
Foray, D. (2004), Economics Of Knowledge, MIT Press.
Kuk, G. (2006), 'Strategic Interaction and Knowledge Sharing
Gambardella, A. & Hall, B. H. (2006), 'Proprietary Versus in the KDE Developer Mailing List', Management Science
Public Domain Licensing Of Software And Research 52(7), 1031-1042.
Products', Research Policy 35(6), 875-892.
Lakhani, K.; Wolf, B.; Bates, J. & DiBona, C. (2002), 'The
Grand, S.; Georg von Krogh, D. L. & Swap, W. (2004), 'Re- Boston Consulting Group hacker survey', www.bcg.com/open-
source Allocation Beyond Firm Boundaries: A Multi-Level source/BCGHackerSurveyOSCON24July02v073.pdf.
Model for Open Source Innovation', Long Range Planning 37,
591-610. Lerner, J. & Tirole, J. (2002), 'Some Simple Economics of
Open Source', Journal of Industrial Economics 50(2), 197-
Granstrand, O. (1999), The Economics and Management of 234.
Intellectual Property, Edward Elgar, Cheltham.
Liebeskind, J. P. (1996), 'Knowledge, strategy, and the theory
Gächter, S.; von Krogh, G. & Haefliger, S. (2006), 'Private- of the firm', Strategic Management Journal, Special Issue:
Collective Innovation and the Fragility of Knowledge Knowledge and the Firm 17, 93-107.
Sharing'.
Luthiger Stoll, B. (2006), 'Spass und Software-Entwicklung:
Haefliger, S.; von Krogh, G. & Spaeth, S. (2008), 'Code Reuse Zur Motivation von Open-Source-Programmierern', PhD the-
in Open Source Software', Management Science 54(1), 180- sis, University of Zurich.
193.
Maemo.org (2006), 'Maemo website', http://maemo.org, ac-
Harhoff, D.; Henkel, J. & von Hippel, E. (2003), 'Profiting cessed Dec 2006.
From Voluntary Information Spillovers: How Users Benefit
By Freely Revealing Their Innovations', Research Policy 32, Martin, R. (2007), 'Nokia Launches 'Context-Aware' Internet
1753-1769. Tablet', InformationWeek.

Heckathorn, D. D. (1997), 'Respondent-Driven Sampling: A Muller, P. & Pénin, J. (2006), 'Why Do Firms Disclose Knowl-
New Approach to the Study of Hidden Populations', Social edge And How Does It Matter?', Journal of Evolutionary Eco-
Problems 44, 174-199. nomics 16(1-2), 85-108.

How Firms Make Friends: Communities in Private-Collective Innovation 48


Nokia (2005), 'Nokia Makes Donation to GNOME Founda-
tion', Nokia press release. Shankland, S. & Charny, B. (2003), 'Linux to power most Mo-
torola phones', accessed Jan 24th, 2007, accessed January
Nuvolari, A. (2004), 'Collective invention during the British 2007 from http://news.com.com/2100-1001-984424.html.
industrial revolution: the case of the Cornish pumping engine',
Cambridge Journal of Economics 28(3), 99-119. Siggelkow, N. (2007), 'Persuasion With Case Studies', Acade-
my of Management Journal 50(1), 20-24.
Osterloh, M. & Rota, S. (2004), Trust and Community in Open
Source Software Production'Trust and Community on the In- Smith, P. (2004), Accelerated product development: Tech-
ternet', Baumann, M./Matzat, U., . niques and traps, Wiley, New Jersey.

Paul, R. (2007), 'Intel launches site for open source mobile Sorenson, O. & Stuart, T. (2000), 'Aging, Obsolesence, and
Linux development', Ars Technica . Organizational Innovation', Administrative Science Quarterly
45(1), 81-112.
Pettigrew, A. (1990), 'Longitudinal Field Research On
Change: Theory And Practice', Organization Science 1(3), Stiglitz, J. E. (2006), 'Scrooge and intellectual property rights',
267-292. British Medical Journal 333(7582), 1279-1280.

PricewaterhouseCoopers (2005), 'Governance, risk, and com- Strauss, A. L. & Corbin, J. (1998), Basics of qualitative re-
pliance series: Global Best Practices', www.globalbestprac- search: Techniques and procedures for developing grounded
tices.com. theory, Sage, Thousand Oaks, CA.

Roberts, J.; Hann, H. & Slaughter, S. (2006), 'Understanding Torvalds, L. & Diamond, D. (2001), Just for Fun, Texere,
the motivations, participation, and performance of open source London, UK.
software developers: A longitudinal study of the Apache
projects', Management Science 52(7), 984-999. Yin, R. K. (2003), Case Study Research: Design and Methods,
Sage.
Shah, S. (2006), 'Motivation, Governance, And The Viability
Of Hybrid Forms In Open Source Software Development', Zander, I. & Zander, U. (2005), 'The Inside Track: On the Im-
Management Science 52(7), 1000-1014. portant (But Neglected) Role of Customers in the Resource-
Based View of Strategy and Firm Growth', Journal of Man-
Shah, S. K. & Corley, K. G. (2006), 'Building Better Theory agement Studies 42(8), 1519-1548.
By Bridging The Quantitative–Qualitative Divide', Journal of
Management Studies 43(8), 1821-1835.

How Firms Make Friends: Communities in Private-Collective Innovation 49


Appendix 3
The Credible Sponsor: Participants’ Motivation
and Organization Attributes in Collaborative Digital Innovation

Publication under review,


please do not circulate or quote without the explicit consent of the authors

Georg von Krogh


Sebastian Spaeth
Matthias Stuermer
Guido Hertel

Chair of Strategic Management and Innovation


Department of Management, Technology, and Economics
ETH Zurich, Switzerland

Collaborative digital innovation projects often involve a sponsoring firm and voluntary, unpaid par-
ticipants. While important to both the firm and the participants, such projects are challenging be-
cause participants are not exposed to pay or career-related incentives offered by firms. Recently,
various research projects have focused on the motivation of voluntary participants in innovation en-
abled by digital technology, for example, open source software projects. However, little is known
about how the sponsoring firm affects this motivation. In this paper, we use a self-determination
framework of motivation (intrinsic/extrinsic motivation) and research on developers’ motivations in
open source software, to create a model that explains how perceived firm attributes impact on par-
ticipants’ motivation. We test the model with data from two collaborative digital innovation projects
sponsored by Nokia and OpenMoko Inc. Both projects are formed around the development of mo-
bile consumer devices. We found that attributes of the sponsoring firm as perceived by individual
participants affected both their intrinsic motivation and the effort they invested in the project.
Among firm attributes, corporate credibility mattered most, while the perceived level of information
revealed by the firm had a moderate impact on motivation. Finally, no effect on motivation and ef-
fort was found for the degree to which participants perceived that they had an influence on the firm.
In summary, the effort voluntary contributors invest in firm-sponsored projects seems to be a func-
tion of both extrinsic and intrinsic motivation.

Introduction through learning about new technologies, increasing


their reputation, lowering the cost of product devel-
In many industries, such as software, computer opment, obtaining useful information about user
hardware, or the media, firms increasingly invest in needs, and decreasing the time to market (Stuermer et
collaboration with outside communities of partici- al., 2009; Henkel, 2006). At the heart of this collabo-
pants who often seek personal rewards in return for ration are digital technologies, such as the Internet,
voluntary contributions to digital innovation. Firms which mitigate the time spent and costs of communi-
benefit from such collaborative digital innovation

How Firms Make Friends: Communities in Private-Collective Innovation 50


cating and coordinating work with outsiders, and engagement can be expected to be low when a firm
therefore enhance the resulting benefits for firms and and its actions are perceived as unfair, exploitative,
communities (Kogut, 2003; Osterloh and Rota, and in conflict with personal values or the values and
2007). Open source software (OSS) is a case in point: norms of a community to which an individual be-
firms (e.g. IBM, Nokia, Hewlett-Packard, Sun Mi- longs. Since voluntary engagement in collaborative
crosystems, or Oracle) draw upon the contributions digital innovation projects resembles organizational
of hundreds of thousands of volunteer software de- citizenship behavior, the predictors of this voluntary
velopers, creating open source software that runs on engagement might also be similar.
hardware sold by these firms or in conjunction with Second, as Dahlander and Magnusson (2005) ar-
their other software products or services. A prerequi- gued, sponsoring firms exert influence on the com-
site for collaborative digital innovation is a communi- munity of participants for commercial and strategic
ty of highly motivated participants (e.g., Kogut and reasons, such as ensuring that the collaboratively de-
Metiu, 2001; von Hippel, 2005). veloped technology provides a sufficient level of eco-
In a recent literature review on collaborative digi- nomic benefits. The firm may attempt to make the
tal innovation and open source software develop- technology compatible with other software or hard-
ment, von Krogh and von Hippel (2006) show that a ware, or to add features that are of importance to the
fast-growing stream of research focuses on partici- firm but not necessarily wanted by the community.
pant motivation. The authors conclude that most of One strategy to influence the developer communities
the existing literature examines the motivation for is to finance “men on the inside” of the community,
OSS developers in general, without taking into ac- who contribute to the collaborative digital innovation
count how a given project’s context shapes and influ- (Dahlander and Wallin, 2006). Yet, little is known
ences the participant’s motivation. Although many about how such strategies are perceived by partici-
scholars recognize that firms sponsor OSS projects pants, and how they affect their motivation. However,
and collaborate with communities of project partici- the heated discussion of perceived attempts by com-
pants (Dahlander and Wallin, 2006; Shah, 2006; Os- mercial software firms to interfere with OSS develop-
terloh and Rotha, 2007), little is known about how ment (e.g., Feller and Fitzgerald, 2002; Raymond,
participants’ perception of the firm and its involve- 1999) suggests that the consequences for their moti-
ment in the project affect their motivation to contrib- vation may be rather negative.
ute. Yet, there are at least three reasons why the per- Third, Gruber and Henkel (2006) showed that
ceived attributes of a firm might affect people’s moti- companies expect to increase their reputation when
vation to participate in digital innovation projects and they reveal source code and other knowledge to the
contribute on a voluntary basis. community of participants. Yet, an empirical question
First, prior studies in organizational psychology is to what extent such activities yield the expected ef-
have provided many examples of how employees’ fect, as well as higher effort from the participants.
perception of their work context affects their willing- In this paper, we develop and test a model that
ness to show high effort for a firm (e.g., perceived predicts participants’ contribution to collaborative
procedural and distributive fairness, perceived orga- digital innovation as a function of the motivational
nizational support, perceived fit between personal consequences of their perception of the sponsoring
and organizational values, level of participation) firm. In creating the model, we draw upon research
(e.g., Rhoades and Eisenberger, 2002). This is partic- on collaborative digital innovation processes and mo-
ularly true for voluntary behavior that extends be- tivations in OSS development, as well as a self-deter-
yond what is written in a formal job contract. On the mination framework. To test the model, we designed
one hand, so-called “extra-role” or “organizational a cross-sectional survey study that was conducted in
citizenship behavior” (e.g., Cohen-Charah and Spec- two communities of collaborative digital innovation
tor, 2001; Coyle-Shapiro, 2002; Organ and Moorman, projects: the Maemo project, developing software to
1993; Organ et al., 2006; Robinson and Morrison, run on Nokia’s Internet Tablet devices N800/810, and
1995) is particularly likely when persons perceive a the OpenMoko project, developing software for
firm as fair, nurturing, and pursuing values similar to OpenMoko mobile phones. The results revealed that
the employee’s own. In addition to reciprocity pro- participants’ motivation was not affected by the per-
cesses (e.g., Blau, 1964), these effects can be addi- ceived accessibility of the firm and the perceived in-
tionally amplified by individuals’ affective commit- fluence they had on the firm. However, participants’
ment to and identification with the organization (e.g., motivation was significantly affected both by the de-
van Dick et al., 2006). On the other hand, voluntary gree of knowledge and information revealed by the

How Firms Make Friends: Communities in Private-Collective Innovation 51


firm and by the perceived credibility of the firm, the frequently including intrinsic motivation as well as
latter showing the strongest effect. extrinsic motivators.5 Intrinsic motivation (e.g., Deci
Our study makes two important contributions to and Ryan, 1985) is given when people perform a task
the literature. First, we advance the emerging work for its own sake, because it is inherently joyful (they
on participant motivation in collaborative digital in- feel capable and competent, and have fun) and satis-
novation by integrating participants’ perception of fies explorative needs (learning new things, personal
firm attributes as additional predictors of their level growth). Extrinsic motivation, on the other hand, is
of contribution to the project. Second, we contribute given when people perform a task for incentives or
to the literature on collaborative digital innovation by outcomes that are distinct from the task itself, such as
exploring how the effects of firm strategies might im- making money or advancing their career by gaining
pact community participants’ motivation, and in turn reputation as software developer. Given that many
affect innovation outcomes. From this study it can be developers work without direct financial remunera-
concluded that perceived firm attributes strongly af- tion, the inclusion of intrinsic motivation has been
fect the extent to which participants contribute to col- quite useful to capture a broad range of motives in
laborative digital innovation. We provide extensive OSS development (Hars and Ou, 2002; Lakhani and
evidence for the conjecture that firms need to manage Wolf, 2005; Roberts et al., 2006; Wu et al, 2007). In-
their “image” (Dutton et al., 1994; Dutton and Duk- deed, Lakhani and Wolf (2005) presented data show-
erich, 2006) in collaborative digital innovation, and ing that intrinsic motivation affected work motiva-
devote particular attention to their credibility as per- tion, in addition to extrinsic motives. Lakhani and
ceived by the individual developer. von Hippel (2003) showed that participants’ experi-
In the next section, we start with a brief literature ence of competence and fun during support tasks was
review and develop the research model. The third significantly related to their willingness to help users
section presents our sample and research design, and of open source software. Roberts et al. (2006)
the fourth section presents the results. Finally, we dis- demonstrated the central role of intrinsic motivation
cuss implications of the study for theory, future re- in a longitudinal study of the Apache project.
search, and practice. As observed by Lerner and Tirole (2002), open
source software is a successful outcome of collective
Development of the Research Model action by people who may or may not be paid for
their contributions. While organization theory mostly
Phenomena such as OSS projects have opened construes innovation as hinging on work by firm em-
new and unprecedented access for researchers to ployees (e.g., Miller and Friesen, 1982; Kogut and
large amounts of data that enable a thorough investi- Zander, 1992; Zenger and Hesterly, 1997), open
gation of a plethora of management and organization source software is an interesting phenomenon for the
theories (e.g., Fleming, 2001). A recent review of re- study of complementary intrinsic motivation in col-
search on collaborative digital innovation and open lective action (e.g. Lakhani and Wolf, 2005; Hars and
source software grouped this work in three streams; Ou, 2002).
the motivation of voluntary and often unpaid partici- In addition to research on the motivation of partic-
pants, the process of collaborative digital innovation, ipants, academic work has also focused on the pro-
and the competitive dynamics resulting from the cesses within collaborative digital innovation
availability of products with public good’s character- projects, including coordination and governance, the
istics in the market (von Krogh and von Hippel,
2006). The two first streams are relevant for the cur-
rent study. 5 Alternative frameworks are often closely related to the dis-
Initially, characterizing open source software as a tinction between intrinsic and extrinsic motivation. For ex-
ample, building on Feller and Fitzgerald’s taxonomy
public good that is non-exclusive and non-rival, (2002), Bonaccorsi and Rossi (2006) separate economic,
Lerner and Tirole (2002) asked why top-notch soft- social, and technological motivation. Economic motiva-
ware developers contribute to creating it for free. tion and extrinsic motivation are similar, and social moti-
vation resembles intrinsic motivation. In addition, “tech-
With a traditional understanding of homo economi-
nological motivation” includes benefits from learning and
cus, where humans seek to maximize economic bene- working with a new technology. A broader and more inte-
fits, such behavior is difficult to explain, even when grative framework is provided by Hemetsberger (2004),
advantages for personal career development are taken who viewed motivation as “self-interest” and “others-ori-
entation” and Hertel et al. (2003) who used a model of
into account. The question initiated a fast-emerging voluntary action in social movements proposed by Klan-
stream of research using various frameworks, most dermans (1997), including collective, norm-oriented, re-
ward, and identification motives.

How Firms Make Friends: Communities in Private-Collective Innovation 52


knowledge and roles of developers and users, the tributes as the antecedent of developers’ contribution
structure of communication, or the evolution of tech- to OSS development. Moreover, we specify develop-
nology (e.g., Franck and Jungwirth, 2003; von Krogh ers’ intrinsic and extrinsic motivation, as well as their
et al. 2003; Kuk, 2006; MacCormack et al., 2006; Os- commitment to and identification with the communi-
terloh and Rotha, 2007). Recently, several authors ty as important mediating mechanisms. The model is
have analyzed the economic benefits for firms partic- shown in Figure 1.
ipating in collaborative digital innovation and OSS
development and have concluded that firms “spon- Knowledge Revealed by the Firm
sor” collaborative digital innovation (Shah, 2006) to
lower the cost of development, accelerate introduc- Many authors argue that knowledge revealed by
tion of a new product into the market, diffuse a tech- firms (for example, of product and process technolo-
nology, obtain new information about user needs, or gy, software code, algorithms) leads to multiple eco-
improve product quality (Bonaccorsi and Rossi, nomic benefits, such as downstream product im-
2003, 2006; Dahlander and Magnusson, 2005; Lerner provements (Harhoff, 1996), increased innovation ac-
and Tirole, 2005). On the one hand, a sponsoring firm tivity by users and manufacturers (Harhoff et al.,
can pay developers to perform certain tasks and sup- 2003), ease of access to innovation networks (Muller
port projects in other ways, financially or with a tech- and Pénin, 2006), reciprocal knowledge revealing by
nical infrastructure. On the other hand, in order to re- other firms (Pénin, 2007), enhanced reputation and
alize economic benefits, the sponsoring firm might low-cost marketing activity (Gruber and Henkel,
influence the innovation process and developer com- 2006), network externalities and technological stan-
munity by controlling the knowledge revealed to the dard-setting (Bonaccorsi and Rossi, 2003), and en-
community (e.g., Harhoff et al., 2003), by providing hanced ability to recruit new participants (Stuermer et
or restricting access to the sponsored project and its al., 2009). Moreover, providing information about
source code (e.g., West and O’Mahony, 2008), and by processes and source code can also trigger reciprocal
building credibility within the developer community behavior by community members (Blau, 1964). By
(e.g., Dahlander and Magnusson, 2005). voluntarily revealing software code and related
In the past, research on the motivation of partici- knowledge, firms can attract and motivate volunteer
pants in collaborative digital innovation projects, and participants to write, test, and debug code, and also
research on firm participation in these projects, have enhance services (Dahlander and Magnusson, 2005).
been pursued independently from each other. Our Yet, firms choose to reveal knowledge selectively;
study seeks to connect and integrate these two sharing detailed engineering knowledge and process-
streams of work, leading to the interesting question of es is often delayed or prohibited (Pénin, 2007). When
how participants’ perceptions of the attributes of the sponsoring OSS projects, firms often blend open and
sponsoring firm might affect their motivation to con- proprietary source code in their products and choose
tribute to collaborative digital innovation projects. to reveal some but not all software with open source
Based on the literature we introduced above, we licenses (Henkel, 2006; West, 2003).
develop a model that links perception of firm at-

Extrinsic
Motivation H9
Knowledge
Contribution
Revealing H1 H7
Intrinsic H10
H2 H3 Motivation
Accessibility H8

H4
Identification
Corporate
H6
Credibility

H5

Perceived Firm Attributes Inidividual Identification, Motivation, and Contribution

Figure 1: Proposed effects of sponsor firm attributes on participant contribution


How Firms Make Friends: Communities in Private-Collective Innovation 53
As Osterloh and Rota (2007) argued, the threshold nity. Stewart and Gosain (2006) supported this find-
for voluntary contributions is low, due to digital tech- ing and argued that in communities where partici-
nology that decreases efforts for communication and pants adhere to an “open source ideology,” free re-
coordination in the innovation project. Previous reve- vealing of software and knowledge, information shar-
lation of knowledge by a sponsoring firm gives par- ing and helping others are collective beliefs and rep-
ticipants access to software code or other knowledge resent important values. Their study showed that
that they can either use directly or extend to create norms and values increase participants’ group identi-
their own enhancements (Lerner and Tirole, 2002). fication because adherence to them enhances emo-
Revealed documentation or technical specifications tional attachment to the OSS project. West and
enable participants to program for some hardware O’Mahony (2008) suggest that sponsoring firms also
more easily and demonstrates the firm’s willingness become more “transparent” when they reveal infor-
to cooperate with external participants. Surveys of mation and source code. In their comparative case
participants have shown that previous contributions study of 12 firm-sponsored OSS projects, the authors
by others impact positively on their motivation to found that the sponsoring firm’s change from private
contribute (Lakhani and von Hippel, 2003; Lakhani to public communication was positively acknowl-
and Wolf, 2005). We expect that both the access to edged by both the firm and the developer community.
knowledge and cooperative gestures from a firm’s The developer community perceived that following
side will increase participants’ intrinsic motivation to and contributing to technical discussions, understand-
contribute to collaborative digital innovation, such as ing how decisions are made, and using revealed soft-
OSS projects, because these activities provide the ware across projects would increase the firm’s adher-
necessary information to trigger explorative as well ence to the norms and values of OSS development.
as reciprocal behavior. Therefore, we suggest: On the other hand, a firm’s loss of transparency may
Hypothesis 1: The perceived revelation of knowl- lead to reduced community identification by partici-
edge by the firm in a firm-sponsored open source pants (Lattemann and Stieglitz, 2005). If participants
project positively affects the intrinsic motivation of perceive the firm is acting in accordance with com-
participants to contribute to the project. munity norms and values by revealing knowledge,
In addition to the expected direct effects of firms’ the firm is perceived as “similar” to other community
knowledge revelation on participants’ intrinsic moti- participants and this in turn should positively affect
vation, there are also conceivable indirect effects that commitment and identification processes. Thus, we
will lead to higher engagement. One central factor propose:
explaining participants’ voluntary engagement is Hypothesis 2: Perceived revelation of knowledge
whether they feel personally connected with “their” by the firm, in a firm-sponsored open source project,
community. For instance, social identification pro- affects positively participants’ commitment to and
cesses have been demonstrated to be the strongest identification with the project.
predictors of voluntary engagement in a study of Lin-
ux developers by Hertel et al. (2003). However, de- Perceived Accessibility of Firm-Sponsored Projects
veloping emotional bonds, commitment, and social
identification with a collaborative digital innovation In unsponsored communities, people become ac-
project is challenging, as many traditional ways of tive participants and rise through an informal hierar-
experiencing social connection are hampered by spa- chy through discussing and demonstrating their skills
tial distance and predominant electronic communica- in coding (von Krogh et al., 2003). In sponsored
tion. One way to foster social identification and com- projects, firms retain some level of control, for exam-
mitment is by experienced similarity of values and ple, deciding which developers have the right to alter
norms among participants (Fiol and O’Connor, the project’s source code. Dahlander and Magnusson
2005), facilitating feelings of sympathy and common (2005) refer to a “balancing act of control” by the
identity, as well as reduced uncertainty as major firm in sponsored OSS projects. Firms need to open
mechanisms of identification processes (e.g., Hogg the software development process to participants
and Terry, 2000). Research on OSS development pro- from the outside to generate their energy, interest, and
cesses indicates that participants may expect sponsor- creativity, while maintaining a certain level of control
ing firms to be “similar” and to adhere to the general to realize appropriate economic returns (West, 2003).
norms of the community. For instance, Lee and Cole To capture this tension between control and openness,
(2003) observed strong norms of knowledge sharing West and O’Mahony (2008) suggest that firm-spon-
within an early study of the Linux developer commu- sored OSS projects not only reveal knowledge to

How Firms Make Friends: Communities in Private-Collective Innovation 54


members of the participant community (“transparen- distinctive qualities of the group (access) are per-
cy”) but also realize “accessibility,” by providing par- ceived to be valuable and attractive, leading to
ticipants with access to the project’s core resources, stronger identification with the group (Fiol and O’-
such as the roadmap, discussion lists, or the official Connor, 2005). We suggest:
version of the software code. The accessibility con- Hypothesis 4: The perceived accessibility of a
struct captures the extent to which participants per- firm-sponsored open source project affects positively
ceive that they can influence the development participants’ commitment to the project and identifi-
process. When the firm is perceived to restrict acces- cation with it.
sibility—for example, by denying participants the
possibility to adapt the source code to their needs— Corporate Credibility
the intrinsic motivation of participants should de-
cline, due to lack of autonomy and control. This is One important reason for firms to sponsor OSS
consistent with conceptual models in work psycholo- projects is to enhance their reputation within the
gy that explain perceived autonomy as a major pre- community of participants and beyond (Gruber and
dictor of intrinsic motivation (e.g., Hackman and Henkel, 2006). Therefore, the firm would need to be
Oldham, 1980; Hertel, 2007); and also consistent perceived as “credible” (Herbig and Milewicz, 1993)
with self-determination theory, which postulates neg- and trustworthy (e.g., Kramer and Tyler, 1996; Mayer
ative effects of perceived external determination on et al., 1995) by community participants. So far, this
intrinsic interest in a task (Deci and Ryan, 1985; Frey aspect has not been considered in either the literature
and Jegen, 2001; Osterloh and Rota, 2007). Indeed, on participant motivation or the literature on innova-
empirical research demonstrates that voluntary partic- tion process. Studies of sponsorship in the field of
ipants in collaborative digital innovation dislike ex- marketing (Cornwell and Maignan, 1998; Cornwell et
ternal control (Shah, 2006; Schroer and Hertel, al., 2005) suggest that corporate credibility relates to
2009). Thus, we suggest: how consumers perceive the “goodness” of a compa-
Hypothesis 3: Perceived autonomy within a firm- ny’s actions and how it manages its activities and
sponsored open source project, and possible influence supports good initiatives (Javalgi et al., 1994). While
over it, affects positively participants’ intrinsic moti- the approaches in firm sponsorship towards partici-
vation to contribute to the project. pants and consumers differ, the basic effects of spon-
In addition to these direct effects of autonomy on sorship interventions are similar, because individuals
intrinsic motivation, perception of the sponsoring fir- make causal inferences about events they observe and
m’s accessibility might also affect voluntary engage- experience (Rifon et al., 2004). For example, in cases
ment by increasing commitment and identification where the firm sponsors charitable initiatives, indi-
processes. By giving them access to OSS projects and viduals often mistrust the firm because they believe
their governance, sponsoring firms may offer partici- the firm’s main objective is to maximize corporate
pants additional opportunities to develop a sense of profits (Bae and Cameron, 2006). Likewise, authors
belonging to the community. For example, by giving have argued that when a firm sponsors an OSS
participants access to the source code the sponsoring project, community participants may mistrust this ef-
firm grants them the right to influence decisions on fort and worry that the firm might exploit and secre-
the future direction of the project (West and O’Maho- tively sell community-based innovations (Varadara-
ny, 2008). This kind of reinforcement of outsiders’ jan and Menon, 1988; Bae and Cameron, 2006; Dean,
opportunities to influence decision-making in the 2002).
project reduces participants’ uncertainty about the Newell and Goldsmith (2001) define perceived
project’s future, and should lead to higher identifica- corporate credibility as “personal judgment of a fir-
tion with the community (Fiol and O’Connor, 2005). m’s reputation.” The authors review and test credibil-
Access to the project’s core resources also signifies ity measures and cluster them statistically in two
that a participant has entered a “core team” of soft- components. “Expertise” measures the extent to
ware developers in the community (von Krogh et al., which the observer feels the firm is capable of fulfill-
2003). Participants rewarded in this way are likely to ing its claims, while “trustworthiness” measures
experience increased feelings of self-worth because whether or not the firm can be trusted.6,7 Consistent
of a sharper distinction between “us” (the core-team with more general approaches on organizational trust
members) and “them” (regular participants with no
access rights) (Hertel et al., 2003; Kelly, 1993). This 6 Reputation itself is a much broader concept and includes,
sort of enhancement of self-worth occurs when the but is not limited to, expertise and trustworthiness (Gold-
smith et al., 2000).

How Firms Make Friends: Communities in Private-Collective Innovation 55


(e.g., Kramer and Tyler, 1995; Mayer et al., 1995), re- Cameron, 2006). Such suspicions and mistrust will
search on virtual collaboration (Jarvenpaa et al., also lower the willingness of participants in collabo-
1998; Jarvenpaa and Leidner, 1999) and open source rative digital projects to identify with the project and
communities (Osterloh and Rota, 2004; Stewart and to feel committed to it. In a similar way, psychologi-
Gosain, 2006) have shown that mutual trust among cal theories assume that the perceived trustworthi-
participants is important for the coordination of work ness, support, and fairness of an organization are pre-
in distributed work contexts (see also de Laat, 2005). conditions to developing a sense of belonging and
However, while theory has addressed the interplay identification (e.g., Meyer et al., 2006; Rhoades et al.,
between incentives and trust in motivating agents in 2001). This line of argument leads us to propose:
employment relationship intrinsically and extrinsical- Hypothesis 6: The perceived corporate credibility
ly (Bénabou and Tirole, 2003), no work has exam- of a firm affects positively the commitment and iden-
ined the extent to which a sponsoring firm’s corpo- tification of participants with the firm-sponsored
rate credibility leads participants to be more intrinsi- open source project.
cally motivated to contribute.
There are several reasons for assuming that firms’ Interrelation between Extrinsic and Intrinsic Motivation
corporate credibility (in terms of perceived expertise
and trust) affects developers’ intrinsic motivation in A model based on the distinction between intrinsic
OSS projects. One main aspect that software devel- and extrinsic motivation needs to address the ques-
opers value in OSS projects is that they can extend tion of how these two constructs are interrelated. In
their own programming skills by learning from others the context of a self-determination framework (Deci
(e.g., Hars and Ou, 2001; Hertel et al., 2003). The and Ryan, 1985), various scholars have argued for
more competent and “top-notch” a sponsoring firm is “crowding-out effects,” describing a decrease in in-
perceived to be, the more participants will be driven trinsic motivation due to external incentives (e.g.
by such intrinsic learning needs. In a similar way, if Deci, 1975; Frey, 1997; Frey and Oberholzer-Gee,
developers perceive the sponsoring firm as untrust- 1997; Weibel et al., 2007). The argument is based on
worthy, participants will reduce their efforts to avoid the idea that extrinsic incentives might be perceived
being a “sucker” and being exploited (Kerr, 1983). as constraining self-determination and diminishing
Preliminary empirical work on firm involvement in self-esteem, which in turn reduce the interest and en-
OSS communities demonstrates that a firm’s reputa- joyment of a task (Frey and Jegen, 2001). Apart from
tion and trustworthiness may impact on developers’ initial laboratory studies with school children, empiri-
reported willingness to contribute, justifying the role cal evidence for such crowding-out effects have been
for perceived corporate credibility in a model of par- reported in the context of the positioning of a nuclear
ticipant motivation (Dahlander and Magnusson, waste facility (Frey, 1997), the collection of monetary
2005; 2008; Dahlander and Wallin, 2006). Thus, we donations (Gneezy and Rustichini, 2000a), or inter-
propose: ventions in daycare centers (Gneezy and Rustichini,
Hypothesis 5: A firm’s perceived corporate credi- 2000b).
bility in a firm-sponsored open source project affects However, the overall empirical support for crowd-
positively the intrinsic motivation of participants to ing-out effects in ongoing work settings is mixed at
contribute to the project. best (see Cameron et al., 2001; Rynes et al., 2005 for
Again, we expect that the perceived credibility of overviews). For instance, a meta-analysis by Eisen-
a firm will not only directly affect the intrinsic moti- berger and Cameron (1996) based on 83 empirical
vation of community members, but will also have an studies found that, overall, extrinsic rewards and in-
impact on effort and performance through commit- trinsic motivation were not negatively but actually
ment and identification processes. Marketing scholars slightly positively related. Crowding-out effects were
have argued that a decrease in a sponsoring firm’s mainly demonstrated for children in laboratory re-
credibility might raise suspicions about hidden mo- search with low monetary incentives. However, there
tives behind its sponsoring initiatives, and the its ac- is no clear evidence that this can be generalized to
tions will be perceived negatively (Goldsmith et al., adults in more meaningful long-term settings where
2000; Newell and Goldsmith, 2001; Bae and they work for substantial monetary rewards. In daily
work settings, the informational functions of external
rewards might often outweigh the controlling aspects,
so that money or reputation increase rather than re-
7 This dual construct of corporate credibility stems from the
idea of “source credibility” (Shimp and Delozier, 1986; duce intrinsic motivation.
Belch et al., 1987; see also Ohanian, 1990).

How Firms Make Friends: Communities in Private-Collective Innovation 56


Indeed, research on open source software provides shown that strong identification of OSS developers
no clear evidence for crowding-out effects. On the with a “hacker community” is a crucial cause for
one hand, Osterloh and Rota (2007) propose that firm obligation- or community-based intrinsic motivation.
sponsorship involves extrinsic incentives, such as For mundane but necessary tasks, such as user sup-
payment and careers, for some (but not all) partici- port, identification with the community’s cause also
pants, and that these incentives crowd out the intrin- proved to be relevant for motivation (Lakhani and
sic motivation of participants in the community. von Hippel, 2003). Hars and Ou (2002) suggest that
However, empirical data for this conjecture is still social identification is a special sort of altruism and
pending. On the other hand, Alexy and Leitner (2008) they find strong evidence of such “kinship” motiva-
suggest that introducing extrinsic rewards might be a tion. In their survey of Linux developers, Hertel et al.
feasible way to acknowledge voluntary participants (2003) found that motivational processes in OSS
in OSS projects. Moreover, Roberts et al. (2006) communities resemble voluntary actions in other
measured intrinsic and extrinsic motivation of OSS groups, like civil rights, labor, or peace movements
developers in a longitudinal study of the Apache (see also Kelly, 1993; Simon et al., 1998). Open
project and did not find negative effects of extrinsic source software participants were strongly motivated
rewards on intrinsic motivation. In a compound mod- by identification with their project and showed high
el, negative and positive effects of extrinsic rewards motivation by investing many hours per week for the
seem to outweigh each other in collaborative digital progress of the Linux project. While the causality of
innovation, leading us to propose: this process might work in both directions—strong
Hypothesis 7: Extrinsic rewards are unrelated to social identification motivates higher levels of contri-
the intrinsic motivation of participants in firm-spon- butions and vice versa (Kelly, 1993)—commitment
sored open source software projects. and identification are thought to impact mostly on the
motivation for collective action (Simon et al., 1998).
Commitment and Identification We propose:
Hypothesis 8: Commitment to the community in
Feeling connected to social groups and communi- a firm-sponsored open source project, and identifica-
ties is a basic need of humans, because we are social tion with it, affect positively the intrinsic motivation
animals. This basic need is also relevant in work con- of participants.
texts, acknowledged in constructs such as affective
organizational commitment (e.g., Allen and Meyer,
Consequences of Extrinsic and Intrinsic Motivation
1996; Meyer et al., 2006) or organizational identifica-
tion (e.g., van Dick et al., 2006). Although these con- Finally, both extrinsic and intrinsic motivation of
structs differ in various aspects, the motivational con- participants in collaborative digital innovation should
sequences of commitment and identification for the be positively related to behavior and performance, as
focused group, organization, or community are gener- indicated by voluntary contributions (cf. Roberts et
ally positive. People with high rather than low com- al., 2006). For extrinsic rewards, we base our predic-
mitment usually show higher motivation and engage- tion on exchange processes (e.g., Blau, 1964) and
ment with the group, organization, or community be- economic contracts (Rynes et al., 2005) rather than
cause they feel strong emotional bonds with it. More- on basic reinforcement effects as explicated in classic
over, high commitment is also linked with feelings of learning theories (e.g., Skinner, 1953). Participants in
obligation to support the group, organization, or com- digital collaborative innovation projects face complex
munity. Likewise, people with high rather than low tasks that afford thorough planning and persistent
identification with a group, organization, or commu- work as well as evaluation and self-regulation pro-
nity exert higher support as they experience the group cesses that cannot be shaped by simple reinforcement
as part of their social self; the success of the group plans. Instead, participants fulfill work contracts and
also increases their personal self-esteem and feelings accomplish tasks to reach personal goals (career,
of security. make money, etc.). Moreover, external incentives
Even though spatial distance and reduced oppor- also have a feedback function relevant for intrinsic
tunities of face-to-face communication make the de- motivation (cf. Deci and Ryan, 1985; Hackman and
velopment of high commitment and identification in Oldham, 1980). Thus, we propose:
digital collaborative innovation challenging, these Hypothesis 9: Extrinsic rewards in a firm-spon-
positive effects of commitment and identification sored open source project affect positively the contri-
should hold. Indeed, Lakhani and Wolf (2005) have bution rate of participants.

How Firms Make Friends: Communities in Private-Collective Innovation 57


Positive effects of intrinsic motivation on contri- was set up, also provided by Nokia, which allowed
bution rate are assumed to be largely independent of community participants to develop and offer their
conscious strategic reasoning. Instead, intrinsic moti- own OSS programs to customers. While Nokia sup-
vation leads to high persistence and effort for its own ported this initiative, the company did not control it
sake, and should therefore increase contribution rate directly.
because participants simply enjoy what they are do- Second, OpenMoko Inc. was publicly announced
ing. This conjecture also holds in collaborative digital as a project by Taiwanese mobile phone manufacturer
innovation, as has been demonstrated by initial em- FIC in November 2006, and became an independent
pirical studies (Lakhani and von Hippel, 2003; company in January 2008. The company is develop-
Roberts et al., 2006; Wu et al., 2007). Thus, we pro- ing a new mobile phone that is based entirely on open
pose: source software. Care was taken to ensure that OSS
Hypothesis 10: Participants’ intrinsic motivation developers had access to all hardware functionality
in a firm-sponsored open source project affects posi- (for example, the GPS chip was changed when it be-
tively their contribution rate to the project. came clear that there was no open source driver for
it). The hardware design of the mobile phone was
Sample and Research Design published under a liberal license whereby developers
could inspect how the hardware was built up or man-
In the following section, we describe our sample ufacture their own “OpenMoko,” should they wish to
of firm-sponsored OSS projects and the methodology do so. Like Nokia, OpenMoko provided and con-
we used to test the model. trolled project infrastructure and the source code
repository. The company also used subcontractors to
Sample fulfill specific tasks. A community of voluntary par-
ticipants could give feedback via various mailing lists
As our empirical research setting, we wanted to
and a bug-tracker. OpenMoko also adopted Nokia’s
examine two OSS projects that were heavily spon-
approach, providing infrastructure that allowed inde-
sored and dominated by a commercial firm but were
pendent programmers to host and develop OSS pro-
still seeking and inviting voluntary contributions
grams related to the mobile phone.
from outside participants. Both projects needed to be
similar enough to provide a comparable setting, yet
different enough to observe potentially diverse be- Research Design
havior and motivation of participants. The core In this section, we describe the data sources, mea-
source code repository needed to be controlled by the sures, and analysis we conducted. All the available
firm, so that it could grant outside writers access to archives of an email list relating to the development
the source code base. Finally, we intended to look at and community discussion of both the Maemo and
areas where developer resources were scarce and OpenMoko projects were downloaded and harvested
multiple projects were competing for participants. for email addresses. Obviously fake and duplicate
For this study, we sample two projects. The first, (e.g., work and private) addresses were removed, re-
Maemo, was initiated by Nokia in 2005 as the operat- sulting in 2,151 unique addresses for the Maemo
ing system of their newly developed Internet Tablet. community and 2,593 unique addresses for the Open-
Maemo is an open source platform, partly using ex- Moko community. We developed and distributed a pi-
isting software (Linux, GTK, Debian package man- lot version of the survey to about 30 community
agement) and partly developing new software. In or- members at the Maemo conference in Berlin on Sep-
der to achieve this, Nokia hired independent subcon- tember 19, 2008. Based on their feedback, the ques-
tractors to solve specific tasks. The technical infra- tionnaire was refined and adapted to make it as clear
structure and source code administration were pro- as possible to community participants. Two versions
vided and controlled by Nokia. While releasing much of the survey, one geared toward Maemo-Nokia and
of their underlying software as open source, some one toward OpenMoko-OpenMoko Inc., were created
end-user applications and hardware drivers remained and put online in November 2008. Both Nokia and
closed-source software. Nevertheless, a lively com- OpenMoko Inc. were asked to endorse the survey.
munity of participants, supported by Nokia employ- However, while both firms encouraged us to perform
ees, sprang up. Participants were able to give feed- the survey, they were not in any way affiliated to it.
back to Nokia developers and to enter failures in the Invitations, complete with the appropriate version of
bug-tracking system (or indeed provide the piece of the questionnaire, were sent to all identified partici-
source code that would fix them). An infrastructure

How Firms Make Friends: Communities in Private-Collective Innovation 58


pants of the Maemo and OpenMoko communities. As fy with its members. Although organizational com-
an additional incentive, we offered to raffle off four mitment and social identification cover slightly dif-
Internet Tablet/OpenMoko devices. Within three ferent aspects, they both refer to individuals’ feeling
weeks (and following two reminders), 322 email ad- of connection with a community, crucial to their mo-
dresses were reported as failing and 1,233 usable re- tivated actions toward it (e.g., Meyer et al., 2006; van
sponses were received (Maemo: 429, OpenMoko: Dick et al., 2006). Four questionnaire items were
804), corresponding to an acceptable response rate of used from Allen and Meyer’s measurement of the
27.9%. “affective commitment” construct (1990), which in-
cludes social identification (see Appendix).
Measures Extrinsic Motivation. The extrinsic motivation
of participants includes pay and career orientation
The dependent variable in our model “number of and whether participants receive non-monetary gifts
hours spent” is the self-reported number of hours that from the company, such as travel or conference subsi-
a participant dedicates to the community using and dies. Three items are part of this latent variable, such
producing digital technology. The types of effort col- as “I contribute to the project because it increases my
lected include writing source code, communicating career opportunities.”
through electronic means, as well as testing and im- Intrinsic Motivation. Despite the long tradition
proving the resulting software. Validity checks of of research on intrinsic motivation, there is no unam-
these self-reports are described in the results section. biguous agreement about the definition and measure-
Measures, questionnaire items, and sources in the lit- ment of this variable. Given the central role of this
erature are provided in the appendix. Unless other- construct in the current study, we decided to adopt a
wise indicated, all measures used a response scale broader approach that includes both enjoyment of
from 1 “strongly disagree” through 5 “strongly various subtasks in the community (coding, bug re-
agree.” porting, marketing, etc.), feedback from others, and
Knowledge Revealing. This variable refers to the learning (cf. Deci and Ryan, 1985). Please note that,
company’s perceived level of information and knowl- as part of intrinsic motivation (e.g., Hackman and
edge revelation. The measure captures whether par- Oldham, 1980; see Hertel, 2007, for a recent discus-
ticipants perceive that the company reveals sufficient sion in the OSS context), feedback from others is dis-
documentation and source code, and whether it is se- tinct from mere status motivation, which can be cate-
cretive about future plans for the project. Three items gorized as extrinsic motivation. Accordingly, in a
are part of this latent variable, for example, “The confirmatory factor analysis, feedback from others
company reveals sufficient proprietary source code as showed a loading of 0.48 when assigned to intrinsic
open source software.” motivation but no fit to extrinsic motivation (see also
Accessibility. This variable refers to the level to Roberts et al., 2006). A total of 13 items measure this
which community participants perceive that they can latent variable.
access relevant databases belonging to the firm and We modeled the constructs according to our hy-
can influence its actions. Item examples are “I can get potheses and tested them in a structural equation
commit access for the PROJECT source code reposi- model (SEM). Such models have been created in the
tory if I want to” or “My opinion is sufficiently taken context of OSS development to identify the link be-
into account when COMPANY makes decisions re- tween developer motives and performance (i.e.,
garding PROJECT.” In total, four items measured “rank” changes in the formal hierarchy) (Roberts et
this latent variable. al., 2006). PLS models have been used to test the im-
Credibility. This latent variable refers to the per- pact of ideology on trust and contribution efforts
ceived credibility of the firm, based on the perceived (Stewart and Gosain, 2006).
expertise of the company’s employees and the com-
pany’s trustworthiness. Questions asked here includ- Results
ed, for example, the perceived skill of company em-
ployees and the attractiveness of the firm as an em- We present the empirical results of our study in
ployer. In total, five items were used to measure this this section. We start by discussing the findings on
variable. the dependent variable, then the results on the full
Commitment and Identification. This latent model. We measured the self-reported dependent
variable indicates whether participants feel emotion- variable “number of hours invested” in the model and
ally attached to the respective community and identi- decided on additional verification of its validity. The

How Firms Make Friends: Communities in Private-Collective Innovation 59


questionnaire asked for a detailed split of the time one person might need more time to report a bug due
spent on different tasks (coding, bug reporting, bug to a very thorough debugging process, while others
fixing, mailing lists, etc.), so we could calculate the simply enter a new feature request without any re-
hours a participant would spent on each activity per search in the database. This might diminish the corre-
week. We collected data on the number of bugs re- lation between the time spent and the number of bugs
ported and the number of bugs fixed from the respec- reported. Third, in Maemo and OpenMoko, there is a
tive projects’ bug databases for all participants and plethora of surrounding sub-projects and related
matched survey respondents with participants in the projects, many of them with their own bug database.
bug database. Of the 429 Maemo respondents, 118 If participants reported bugs there, we would not
were found to have reported bugs to the Maemo bug have been able to measure accurately the number of
tracker, with 62 participants claiming to have spent bugs reported. Altogether, this analysis provides con-
time on bug reporting (57 claimed to have reported fidence that the number of self-reported hours spent
bugs and were found to have done so). Since the total per week is a valid measure on both projects.
number of self-reported hours spent was roughly esti- A structural equation model was set up explaining
mated, and since we cannot necessarily assume equal the number of hours participants dedicated to the
productivity for all participants (some might take project in question. It was fitted using the SEM pack-
longer to report a bug than others), we calculated age in the R software by the method of minimizing
Spearman’s Rank correlation coefficient, which as- the negative log-likelihood (Fox, 2006). The model
sumes only a monotonic relationship between two Chi Square value of 3612 (d.f. = 521) is significant,
variables. The correlation of the number of self-re- which is not surprising given the comparably large
ported hours of bug reporting versus the actually sample size. The goodness-of-fit index of 0.84 (AGFI
measured number of filed bugs was 0.63, which is = 0.82) indicates a somewhat non-optimal model fit,
sufficient to confirm that community participants re- whereas the RMSEA index of 0.07 is below the gen-
ported a valid number of hours per week. eral acceptance value of 0.08, indicating an accept-
In the OpenMoko project, 90 people claimed to be able but not optimal fit (Browne and Cudeck, 1993).
spending time on bug reporting and 85 were found to Further measures are the standardized root mean
have been reporting bugs, with an overlap of 44 peo- square residual (SRMR) of 0.18, which is quite high,
ple both claiming to and actually having reported and the Bayesian information criterion (BIC) of
bugs. The rank correlation for self-reported hours of -52.95. Including a dummy variable that controls for
bug reporting versus actually filed bugs is r = 0.44, the project in question (Maemo/OpenMoko) did not
which also confirms the validity of the measure. lead to significant model improvements. Thus, we
There are three possible reasons for this result. First, conclude the same model holds for both projects in
participants may have used different email addresses the sample.
for mailing lists and bug reporting and did not reveal Hypothesis 1 was rejected: the amount of knowl-
their bug database ID in the survey. In this case, some edge the firm was perceived to reveal did not have
participants could not be allocated to the data. Sec- any significant direct impact on the intrinsic motiva-
ond, productivity among participants might differ, as tion of community participants. There is an indirect

Extrinsic
Motivation 0.385***
Knowledge
0.005 Contribution
Revealing -0.022
0.121*** Intrinsic
Motivation 0.318***
0.03*
Accessibility 0.403***
0.053***
Identification
Corporate
0.581***
Credibility

0.489***

Figure 2: SEM model and standardized path coefficients

How Firms Make Friends: Communities in Private-Collective Innovation 60


mediated link, however, as knowledge revealing cor- innovation. However, if and how participants per-
relates with affective commitment (H2) which, in ceive, accept, and use technology is dependent not
turn, correlates with intrinsic motivation (H8). While only on the technologies per se, but also on the social
community participants may have been complaining context (Orlikowski, 2000). This study makes two
about the lack of source code and specifications in contributions to the literature. First, we advance the
their email communication, it does not mean that a research on participant motivation in virtual collabo-
lack of knowledge revealing reduces their intrinsic rative digital innovation projects, showing that the
motivation. Yet, the firm that reveals information and distinction between extrinsic and intrinsic motivation
knowledge seems to increase participants’ attachment is useful to predict the level of contributions by par-
to the community and identification with it. ticipants. This lends extensive empirical support to
The effect of company accessibility is significant- prior work by Hertel et al. (2003), Shah (2006),
ly positive on both intrinsic motivation (H3) and Lakhani and Wolf (2003), and Roberts et al. (2006).
commitment to/identification with the community In both projects examined, participants’ collective
(H4). Although significant, the effect is small in the contributions to development of the mobile devices’
direct case (standardized path coefficients of 0.03) as software were quite large, however thanks to using
well as in its effect on commitment/identification electronic platforms the required efforts of making
(standardized path coefficients of 0.05). Participants such contributions were low. Going beyond the moti-
do not appear to expect companies to listen to their vation of community participants per se, our study
opinions or pick up their contributed code. It seems shows that the context of innovation affects partici-
participants’ perception of accessibility provided by pants’ commitment, identification, and motivation. To
the firm does not largely affect their intrinsic motiva- our knowledge, this study is the first to demonstrate
tion. that participants’ level of contributions to collabora-
Firm credibility, indicated by the perceived exper- tive digital innovation is significantly related to per-
tise of employees and the trustworthiness of the firm, ceived firm attributes, and that the effects of this are
had by far the largest effect on both intrinsic motiva- mediated by intrinsic motivation as well as commit-
tion (H5) and commitment/identification (H6). These ment and social identification processes.
perceived firm attributes matter more than the Somewhat surprisingly, perceived firm accessibil-
amount of information revealed and the degree to ity, i.e., autonomy within the project and impact on
which company participants can influence the com- firm decisions about it, had only a small (albeit sig-
pany and its actions. Moreover, Hypothesis 8 was nificant) effect on community identification or intrin-
confirmed, as commitment and identification were sic motivation. This result might be that, while the
significantly related to intrinsic motivation, showing projects investigated do not grant community partici-
that the more participants identified with their com- pants direct access to the code repository, they do of-
munity, the more they were intrinsically motivated to fer a means to contribute to and influence the devel-
participate. Finally, the extrinsic (H9) and intrinsic opment process. For example, on the project’s online
motivation of community participants correlates posi- collaboration platform, participants can create inde-
tively with the number of hours that a participant pendent subprojects that aim to enhance the device’s
spends on the project. functionality without interfering with the official soft-
Consistent with Hypothesis 7, no evidence for the ware distribution. Bug fixes may also be submitted
so called “crowding-out” effect was found, as extrin- by anyone, leaving integration to the firm. Some sim-
sic and intrinsic motivation were close to zero and ilar anecdotal evidence was found in qualitative inter-
not statistically significant. Thus, in accordance with views by West and O’Mahony (2008) who conclude
a skeptical view on the generalizability of the that “people can be phenomenally valuable contribu-
“crowding-out” effect (e.g., Cameron et al., 2001; tors without having [code repository commit] access
Rynes et al., 2005), no evidence was found for detri- for a long time” (p. 158).
mental effects of external rewards on intrinsic moti- Another surprising result is that the firm’s per-
vation in the two projects (see also Roberts et al., ceived level of knowledge and information revelation
2006, for similar results). has only a moderate impact on participant motiva-
tion, and only through community identification. The
Discussion and Conclusion result is unexpected, since we initially believed this
to be a major driver in the model. Information and
Firm boundaries become more permeable when knowledge provided by the firm were expected to in-
they embrace and participate in collaborative digital crease the available resources for participants to com-

How Firms Make Friends: Communities in Private-Collective Innovation 61


plete their tasks and, thus, should also have increased they expect from firms like Nokia and Apple, which
their motivation. Yet, in hindsight, looking at enthusi- are known to protect most of their software through
astic communities of participants who reverse-engi- traditional intellectual property rights, including com-
neer proprietary applications, or create applications mercial software licenses. Determining an
for the more closed Apple iPhone, it is plausible that “objective” degree of openness and information re-
information and knowledge revelation in isolation vealing by comparing the protection mechanisms and
does not increase the intrinsic motivation of partici- activities of each firm is beyond the scope of this
pants. This lends additional credibility to the result contribution. Future research should extend this ques-
that context matters for the motivation of participants tion to a large sample of firms that conduct collabora-
in collaborative digital innovation. tive digital innovation and examine how openness
Second, we contribute to the literature on process- and information revealing shape perceived openness.
es of collaborative digital innovation (e.g. Franck and Moreover, the data provide for a cross-sectional anal-
Jungwirth, 2003; von Krogh et al. 2003; Shah, 2006; ysis of collaborative digital innovation with data
Kuk, 2006; MacCormack et al., 2006) by exploring mainly drawn from the same source (self-ratings).
how firms and developers form a symbiotic relation- Apart from common method biases that might over-
ship in creating motivational structures and produc- estimate the interrelations (Podsakoff et al., 2003),
ing an innovation outcome. Based on the results, we causal processes could not be tested. Longitudinal re-
can conclude that the perceived attributes of firms search designs are desirable in future studies, allow-
strongly affect the extent to which participants identi- ing scholars to explore the extent to which knowl-
fy with the community and contribute to collabora- edge and information revealing will lead to a subse-
tive digital innovation, in particular that firms need to quent increase in corporate credibility. Such temporal
be perceived as “credible sponsors.” We provide evi- consequences of changes in the model’s constructs
dence, of interest to innovation process scholars, that are relevant for innovation process research.
in order to succeed in attracting voluntary partici- Practitioners interested in supporting the digital
pants to collaborative digital innovation, firms need a collaboration between a firm and a surrounding com-
positive “image” (Dutton et al., 1994) along the three munity of voluntary contributors may observe that
dimensions of knowledge revealing, accessibility, and “openness” per se does not increase the intrinsic mo-
credibility. tivation of community of participants and their
Some potential limitations apply to our study. Par- propensity to contribute. Voluntary participants in our
ticipants received a personal invitation to participate study did not expect to have access to company deci-
in the online survey and, although the response rate sions or impact on them. Yet, this did not reduce their
of 28% is acceptable, there might be a self-selection motivation. However, both Nokia and OpenMoko of-
bias among respondents. Given that community par- fered a repository for autonomous, community-man-
ticipants are defined by informal and voluntary pres- aged applications that mattered to participants. The
ence in the project, it was impossible to exert addi- mere provision of knowledge and information by the
tional pressure on potential survey respondents to re- firm, even if perceived by the participants, may not
veal their identity. The incentive of winning a con- directly boost participants’ motivation. However, in
sumer device might also have introduced a self-selec- the sample, if participants perceive that the firm re-
tion bias, although the data were screened manually veals knowledge, they identify somewhat more
to detect and remove any non-serious survey submis- strongly with their community. Thus, making partici-
sions. However, because this research investigated in- pants feel “at home” in their community is an impor-
terrelationships between the variables instead of main tant part of raising voluntary engagement. Above all,
scores, potential selection biases lead to a more con- practitioners should recognize that corporate credibil-
servative test of the hypotheses. ity (the perceived expertise and trustworthiness of the
Our research focuses on individually perceived firm and its employees) proves to be the most impor-
firm attributes. This approach is appropriate because tant factor for motivation. While corporate credibility
of the initial conjecture that participants’ motivation is not easy to control (increasing the employees’ ex-
will be shaped by their perception of the sponsoring pertise), the perceived characteristics can be im-
firm in collaborative digital innovation. However, the proved to a certain extent, e.g., by sending developers
chosen design does create the difficulty that some in- to conferences or by having them engage in discus-
dividuals might expect higher standards of knowl- sion with community participants.
edge revealing and accessibility from firms dedicated This study has contributed to the increasingly im-
to open source software, such as OpenMoko, than portant topic of voluntary contributions in collabora-

How Firms Make Friends: Communities in Private-Collective Innovation 62


tive digital innovation sponsored by firms. Our find- Cornwell, T. B., I. Maignan. 1998. An international review of
ings confirm that perceived firm attributes, in particu- sponsorship research. Journal of Advertising 27(1) 1-21.
lar corporate credibility, affect participants’ intrinsic Cornwell, T. B., C. S. Weeks, D. P. Roy. 2005. Sponsorship-
motivation to contribute. This result is relevant to in- linked marketing: opening the black box. Journal of Advertis-
novation research and practice and opens up new re- ing 34(2) 21-42.
search questions and a call for extended testing. Coyle-Shapiro, J. A. 2002. A psychological contract perspec-
tive on organizational citizenship behavior. Journal of Organi-
zational Behavior 23(8) 927-946.

Dahlander, L., M. G. Magnusson. 2005. Relationships be-


tween open source software companies and communities: ob-
References servations from Nordic firms. Research Policy 34(4) 481-493.

Alexy, O., M. Leitner. 2008. Norms, rewards, and their effect Dahlander, L., M. Wallin. 2006. A man on the inside: unlock-
on the motivation of open source software developers. Work- ing communities as complementary assets. Research Policy
ing paper, Technical University of Munich. 35(8) 1243-1259.

Allen, N. J., J. P. Meyer. 1996. Affective, continuance, and Dean, D. H. 2002. Associating the corporation with a charita-
normative commitment to the organization: an examination of ble event through sponsorship: measuring the effects on corpo-
construct validity. Journal of Vocational Behavior 49(3) 252- rate community relations. Journal of Advertising 31(4) 77-87.
276.
Deci, E. 1975. Intrinsic Motivation. New York, London:
Allen, N. J., J. P. Meyer. 1990. The measurement and the an- Plenum Press.
tecedents of affective, continuance and normative commitment
to the organization. Journal of Occupational Psychology 63(1) Deci, E. L., R. M. Ryan. 1985. Intrinsic Motivation and Self-
1-18. Determination in Human Behavior. New York: Plenum.

Bae, J., G. T. Cameron. 2006. Conditioning effect of prior rep- de Laat, P. 2005. Trusting virtual trust. Ethics and Information
utation on perception of corporate giving. Public Relations Technology 7(3) 167-180.
Review 32 144-150.
Dutton, J. E., J. M. Dukerich. 2006. The relational foundation
Belch, G., M. Belch, A. Villareal. 1987. Effects of advertising of research: an underappreciated dimension of interesting re-
communications: review of research. Research in Marketing 9 search. Academy of Management Journal 49(1) 21-26.
59-117.
Dutton, J. E., J. M. Dukerich, C. V. Harquail. 1994. Organiza-
Bénabou, R., J. Tirole. 2003. Intrinsic and extrinsic motiva- tionaI images and member identification. Administrative Sci-
tion. Review of Economic Studies 70(3) 489-520. ence Quarterly 39(2) 239-263.

Bitzer, J., W. Schrettl, P. J. Schröder. 2007. Intrinsic motiva- Eisenberger, R., J. Cameron. 1996. Detrimental effects of re-
tion in open source software development. Journal Of Com- wards: reality or myth?. American Psychologist 51(11) 1153-
parative Economics 35(1) 160-169. 66.

Blau, P. M. 1964. Exchange and Power in Social Life. John Feller, J., B. Fitzgerald. 2002. Understanding Open Source
Wiley and Sons, New York. Software Development. London, UK: Addison-Wesley.

Bonaccorsi, A., C. Rossi. 2006. Comparing motivations of in- Fiol, C. M., E. J. O’Connor. 2005. Identification in face-to-
dividual programmers and firms to take part in the open face, hybrid, and pure virtual teams: Untangling the contradic-
source movement: from community to business. Knowledge, tions. Organization Science 16(1) 19-32.
Technology, and Policy 18(4) 40-64.
Fleming, L. 2001. Recombinant uncertainty in technological
Bonaccorsi, A., C. Rossi. 2003. Why open source software can search. Management Science 47(1) 117-132.
succeed. Research Policy 32(7) 1243-1258.
Fox, J. 2006. Structural equation modeling with the sem pack-
Browne, M. W., R. Cudeck. 1993. Alternative ways of assess- age in R. Structural Equation Modeling 13(3) 465-486.
ing model fit. In: Bollen, K. A., J. S. Long (eds.). Testing
structural equation models. Sage, Newbury Park. Franck, E., C. Jungwirth. 2003. Reconciling rent-seekers and
donators – the governance structure of open source. Journal of
Cameron, J., K. M. Banko, W. D. Pierce. 2001. Pervasive neg- Management and Governance 7 401-421.
ative effects of rewards on intrinsic motivation: the myth con-
tinues. The Behavior Analyst 24(1) 1-44. Frey, B. S. 1997. Not Just For The Money: An Economic The-
ory Of Personal Motivation. Cheltenham, UK: Edward Elgar.
Cohen-Charah, Y., P. Spector. 2001. The role of justice in or-
ganizations: a meta-analysis. Organizational Behavior and Frey, B. S., R. Jegen. 2001. Motivation crowding theory. Jour-
Human Decision Processes 86(2) 278-321. nal of Economic Surveys 15(5) 589-611.

How Firms Make Friends: Communities in Private-Collective Innovation 63


Frey, B. S., F. Oberholzer-Gee. 1997. The cost of price incen- von Hippel, E. 2005. Democratizing Innovation. MIT Press.
tives: an empirical analysis of motivation crowding-out.
American Economic Review 87(4) 746-755. Hogg, M. A., D. J. Terry. 2000. Social identity and self-catego-
rization processes in organizational contexts. Academy of
Ghosh, R. A. (2005).Understanding free software developers: Management Review 25(1) 121-140.
findings from the FLOSS study. Feller, J., B. Fitzgerald, S.
Hissam, K. R. Lakhani, eds.Perspectives on Free and Open Jarvenpaa, S., K. Knoll, D. Leidner. 1998. Is there anybody
Source Software, Cambridge MA: MIT Press. out there? Antecedents of trust in global virtual teams. Journal
of Management Information Systems 14(4) 29-64.
Gneezy, U., A. Rustichini. 2000a. Pay enough or don’t pay at
all. The Quarterly Journal of Economics 115 791-810. Jarvenpaa, S. L., D. E. Leidner. 1999. Communication and
trust in global virtual teams. Organization Science 10 (6, Spe-
Gneezy, U., A. Rustichini. 2000b. A fine is a price. Journal of cial Issue: Communication Processes for Virtual Organiza-
Legal Studies 29(1) 1-17. tions) 791-815.

Goldsmith, R. E., B. A. Lafferty, S. J. Newell. 2000. The im- Javalgi, R. G., M. B. Traylor, A. C. Gross, E. Lampman. 1994.
pact of corporate credibility and celebrity credibility on con- Awareness of sponsorship and corporate image: an empirical
sumer reaction to advertisements and brands. Journal of Ad- investigation. Joumal of Advertising 13(4) 47-58.
vertising 29(3) 43-54.
Kelly, C. 1993. Group identification, intergroup perceptions
Gruber, M., J. Henkel. 2006. New ventures based on open in- and collective action. European review of social psychology
novation – an empirical analysis of start-up firms in embedded 4(1) 59-83.
Linux. International Journal of Technology Management
33(4) 356-372. Kerr, N. 1983. Motivation losses in small groups: A social
dilemma analysis. Journal of Personality and Social Psychol-
Hackman, J., G. Oldham. 1980. Work Redesign. Reading: Ad- ogy 45(4) 819-828.
dison-Wesley.
Klandermans, B. 1997. The Social Psychology of Protest.
Harhoff, D. 1996. Strategic Spillovers And Incentives For Re- Basil Blackwell.
search And Development. Management Science 42(6) 907-
925. Kogut, B. 2003. The Global Internet Economy. Cambridge
MA.
Harhoff, D., J. Henkel, E. von Hippel. 2003. Profiting from
voluntary information spillovers: how users benefit by freely Kogut, B., A. Metiu. 2001. Open source software development
revealing their innovations. Research Policy 32 1753-1769. and distributed innovation. Oxford Rev. Econom. Policy 17(2)
248-264.
Hars, A., S. Ou. 2002. Working for free? Motivations for par-
ticipating in open-source projects. International Journal of Kogut, B., U. Zander. 1992. Knowledge of the firm, combina-
Electronic Commerce 6 25-39. tive capabilities, and the replication of technology. Organiza-
tion Science 3 383-397.
Hars, A., S. Ou. 2001. Working for Free? Motivations for par-
ticipating in open-source projects. Proceedings of the 34th An- Kramer, R., T. Tyler. 1996. Trust in Organizations. Frontiers
nual Hawaii International Conference on System Sciences of Theory and Research. Thousand Oakes: Sage Publications.
(HICSS-34) 7.
von Krogh, G., E. von Hippel. 2006. The Promise of Research
Henkel, J. 2006. Selective revealing in open innovation pro- on Open Source Software. Management Science, Special Issue
cesses: the case of embedded Linux. Research Policy 35(7) on open source software 52 975-983.
953-969.
von Krogh, G., S. Spaeth, K. R. Lakhani. 2003. Community,
Hemetsberger, A. 2004. When Consumers Produce on the In- Joining, And Specialization In Open Source Software Innova-
ternet: The Relationship between Cognitive-affective, Social- tion: A Case Study. Research Policy 32(7) 1217-1241.
ly-based, and Behavioral Involvement of Prosumers.
http://opensource.mit.edu/papers/hemetsberger1.pdf, accessed Kuk, G. 2006. Strategic interaction and knowledge sharing in
July 2008. the kde developer mailing list. Management Science 52(7)
1031-1042.
Herbig, P., J. Milewicz. 1993. The relationship of reputation
and credibility to brand success. The Journal of Consumer Lakhani, K. R., E. von Hippel. 2003. How open source soft-
Marketing 10(3). ware works: “free” user-to-user assistance. Research Policy
32(6) 923-943.
Hertel, G. 2007. Motivating job design as a factor in open
source governance. Journal of Management and Governance Lakhani, K. R., R. G. Wolf. (2005).Why hackers do what they
11(2) 129-137. do: understanding motivation and effort in free/open source
software projects. Feller, J., B. Fitzgerald, S. Hissam, K. R.
Hertel, G., S. Niedner, S. Herrmann. 2003. Motivation of soft- Lakhani, eds.Perspectives on Free and Open Source Software,
ware developers in open source projects: an Internet-based MIT Press.
survey of contributors to the Linux kernel. Research Policy
32(7) 1159-1177.

How Firms Make Friends: Communities in Private-Collective Innovation 64


Lattemann, C., S. Stieglitz. 2005. Framework for governance Podsakoff, P., S. M. J. L. N. Podsakoff. 2003. Common
in open source communities. Proceedings of the 38th Annual method biases in behavioral research: A critical review of the
Hawaii International Conference on System Sciences (HICSS literature and recommended remedies. Journal of Applied Psy-
’05) - Track 7. chology 88(5) 879-903.

Lee, G. K., R. E. Cole. 2003. From a firm-based to a commu- Pénin, J. 2007. Open knowledge disclosure: an overview of
nity-based model of knowledge creation: the case of the Linux the evidence and economic motivations. Journal of Economic
kernel development. Organization Science 14 633-649. Surveys 21(2) 326-347.

Lerner, J., J. Tirole. 2005. The economics of technology shar- Raymond, E. S. 1999. The Cathedral and the Bazaar. 1st Edi-
ing: open source and beyond. The Journal of Economic Per- tion. O’Reilly, Sebastopol, CA.
spectives 19(2) 99-120.
Rhoades, L., R. Eisenberger, S. Armeli. 2001. Affective com-
Lerner, J., J. Tirole. 2002. Some simple economics of open mitment to the organization: the contribution of perceived or-
source. Journal of Industrial Economics 50(2) 197-234. ganizational support. Journal of Applied Psychology 86(5)
825-836.
Luthiger, B., C. Jungwirth. 2007. Pervasive fun. First Monday
12(1). Rifon, N. J., S. M. Choi, C. S. Trimble, H. Li. 2004. Congru-
ence effects in sponsorship: the mediating role of sponsor
MacCormack, A., J. Rusnak, C. Y. Baldwin. 2006. Exploring credibility and consumer attributions of sponsor motive. Jour-
the structure of complex software designs: an empirical study nal of Advertising 33(1) 29-52.
of open source and proprietary code. Management Science
52(7) 1015-1030. Roberts, J., H. Hann, S. Slaughter. 2006. Understanding the
motivations, participation, and performance of open source
Mayer, R., J. Davis, F. Schoorman. 1995. An integrative model software developers: a longitudinal study of the Apache
of organizational trust. Academy of Management Review 20(3) projects. Management Science 52(7) 984-999.
709-734.
Robinson, S., E. Morrison. 1995. Psychological contracts and
Meyer, J., T. Becker, R. van Dick. 2006. Social identities and OCB: the effect of unfulfilled obligations on civic virtue be-
commitments at work: toward an integrative model. Journal havior. Journal of Organizational Behavior 16(3) 289-298.
of Organizational Behavior 27(5) 665-683.
Rynes, S., B. Gerhart, L. Parks. 2005. Personnel Psychology:
Miller, D., P. H. Friesen. 1982. Innovation in conservative and Performance evaluation and pay for performance. Annual Re-
entrepreneurial firms: two models of strategic momentum. view of Psychology 56 571-600.
Strategic Management Journal 3 (1) 1 - 25.
Schroer, J., G. Hertel. 2009. Engagement in an open web-
Muller, P., J. Pénin. 2006. Why do firms disclose knowledge based encyclopedia: Wikipedians and why they do it. Media
and how does it matter? Journal of Evolutionary Economics Psychology 12(1) 96-120.
16(1-2) 85-108.
Shah, S. 2006. Motivation, governance, and the viability of
Newell, S. J., R. E. Goldsmith. 2001. The development of a hybrid forms in open source software development. Manage-
scale to measure perceived corporate credibility. Journal of ment Science 52(7) 1000-1014.
Business Research 52(3) 235-247.
Shimp, T., M. Delozier. 1986. Promotion Management and
Ohanian, R. 1990. Construction and validation of a scale to Marketing Communications. Chicago, IL: Dryden Press.
measure celebrity endorser's perceived expertise, trustworthi-
ness, and attractiveness. Joumal of Advertising 19 39-52. Simon, B., M. Loewy, S. Sturmer, U. Weber, P. Freytag, C.
Habig, C. Kampmeier, P. Spahlinger. 1998. Collective identifi-
Organ, D., R. Moorman. 1993. Fairness and organizational cit- cation and social movement participation. Journal of Person-
izenship behavior. What are the connections?. Social Justice ality and Social Psychology 74(3) 646-658.
Research 6(1) 5-18.
Skinner, B. F. 1953. Science and Human Behavior. New York:
Organ, D., P. Podsakoff, S. MacKenzie. 2006. Organizational Macmillan.
citizenship behavior. Its nature, antecedents, and conse-
quences. Thousand Oakes: Sage Publications. Spaeth, S., S. Haefliger, G. von Krogh, B. Renzl. 2008. Com-
munal resources in open source software development. Infor-
Orlikowski, W. J. 2000. Using technology and constituting mation Research 13 (1).
structures: a practice lens for studying technology in organiza-
tions. Organization Science 11(4) 404-428. Stewart, K. J., S. Gosain. 2006. The impact of ideology on ef-
fectiveness in open source software development teams. MIS
Osterloh, M., S. Rota. 2007. Open source software develop- Quarterly 30 (2) 291-314.
ment—Just another case of collective invention?. Research
Policy 36(2) 157-171. Stuermer, M., S. Spaeth, G. von Krogh. 2009. Extending pri-
vate-collective innovation: a case study. R&D Management 39
Osterloh, M., S. Rota. (2004).Trust and community in open (2) 170-191.
source software production. Trust and Community on the In-
ternet, M. Baumann and U. Matzat.

How Firms Make Friends: Communities in Private-Collective Innovation 65


van Dick, R., M. Grojean, O. Christ, J. Wieseke. 2006. Identi- West, J. 2003. How open is open enough? Melding proprietary
ty and the extra-mile: Relationships between organizational and open source platform strategies. Research Policy 32(7)
identification and organizational citizenship behaviour. British 1259-1285.
Journal of Management 17 (4) 283-301.
West, J., S. O'Mahony. 2008. The role of participation archi-
tecture in growing sponsored open source communities. In-
Varadarajan, P. R., A. Menon. 1988. Cause-related marketing: dustry & Innovation 15(2) 145-168.
a coalignment of marketing strategy and corporate philan-
thropy. Journal of Marketing 52 58-74. Wu, C., J. H. Gerlach, C. E. Young. 2007. An empirical analy-
sis of open source software developers’ motivations and con-
tinuance intentions. Information and Management 44(3) 253-
Weibel, A., K. Rost, M. Osterloh. 2007. Crowding-out of in- 262.
trinsic motivation – opening the black box. Working paper,
University of Zurich. Zenger, T. R., W. S. Hesterly. 1997. The disaggregation of cor-
porations: selective intervention, high-powered incentives, and
molecular units. Organization Science 8(3) 209-222.

How Firms Make Friends: Communities in Private-Collective Innovation 66


CV Matthias Stürmer

Name Matthias Stürmer, Dr. sc. ETH Zürich, lic.rer.pol.


Private Address Steinhölzliweg 77
3007 Bern
Switzerland
matthias@stuermer.ch
+41 31 371 80 87 (home)
+41 76 368 81 65 (mobile)
Website www.stuermer.ch
Working Address Liip AG
Kornhausplatz 14
3011 Bern
Switzerland
Birthday February 20th 1980
Nationalities Swiss and German
Civil Status Married to Anita Stürmer-Müller
Children Lionel Semion (2006) and Kai Timon (2008)

Education

January 2006 – June 2009 PhD Student, ETH Zürich


In the Department of Management, Technology, and Economics MTEC
at the Chair of Strategic Management and Innovation of Prof. Georg von Krogh
Thesis title: “How Firms Make Friends: Communities in Private-Collective Innovation”
PhD thesis in the area of strategy, technology and innovation management

August 2008 – Sept. 2008 Visiting Scholar at University of Washington, Seattle (Foster School of Business)
Joint research project with Prof. Sonali Shah on reciprocity in the Eclipse community

October 2000 – February 2005 Studies of Business Administration and Computer Science, University of Bern
Specialization in management of information systems and organizations
Title of licentiate (master thesis): “Open Source Community Building”

February 2002 – July 2002 Erasmus Semester at University of Oviedo, Spain

August 1996 – July 2000 Gymnasium Thun-Schadau, Switzerland (C-Matur)

August 1997 – July 1998 High School Year in Belleville, Michigan, USA

How Firms Make Friends: Communities in Private-Collective Innovation 67


Work

August 2009 – today Project Manager and Business Development at Liip AG


Acquisition and management of software development projects, open source technol-
ogy consulting

August 2009 – today Lecturer of Strategic Management at ETH Zürich

January 2006 – June 2009 Research and Teaching Assistant, ETH Zürich
In the Department of Management, Technology, and Economics MTEC
at the Chair of Strategic Management and Innovation of Prof. Georg von Krogh
Lectures: Corporate Strategy, Basics of Scientific Work, PhD Course on Innovation
Theory

March 2009 – today Secretary of Parliamentarian Group Digital Sustainability


Nonpartisan group of several national politicians promoting open source software,
open standards, open content and open access in Switzerland

September 2006 – today Initiation and Organization of OpenExpo


Largest Swiss conference and exhibition on open source software, taking place twice
a year in Bern and Zürich, 1200 visitors, budget over CHF 120'000 each event

June 2007 – May 2009 Initiation and Organization of Open Source Software Educational Conference
Annual conference and workshops on OSS in education

January 2005 – today Foundation and Management of nice, a Company specialized in Web Services
Implementation of TYPO3 and other web applications for customers, consulting of
institutions and firms on community building strategies

August 2005 – December 2005 Civil Service at TearFund Switzerland


Support in a campaign with focus on the UN Millennium Development goals

March 2005 – July 2005 Civil Service at Diego Thomson, Lima, Peru
Introduction of Linux and other open source software in a teacher seminary

December 2002 – December 2004 Co-Foundation of foresite Systems, an Internet Company

2000 – 2005 Development of foresite CMS (PHP & MySQL)

Languages
Level Experience
German Strong writing and speaking skills Native language
English Strong writing and speaking skills High school year in the US, master and doctoral thesis
Spanish Intermediate writing and speaking skills Erasmus semester in Spain, 4 months civil service in Peru
French Intermediate writing and speaking skills 8 years education in school, internships in Romandy

How Firms Make Friends: Communities in Private-Collective Innovation 68


Programming Skills
Level Experience
HTML/CSS High Programming of approx. 80 websites for companies, NGOs etc.
PHP/MySQL Intermediate Programming of content management system and customized web applications
Java Basic Student projects at university, participation in workshops

Voluntary Work

Member of the Board Swiss Open Systems User Group /ch/open (a Swiss open source software association)
AELL (Swiss association for scholarships at a teacher's training college in Lima, Peru)
EVP Stadt Bern (Swiss evangelical people's party)
Active Member StopArmut 2015 (leader of campaign promoting the UN Millennium Development goals)
TearFund Schweiz (international aid organization)
Former Activities 1998 – 2000 Amnesty International (local group leader)
1986 – 1998 Boy scouts (group leader, organizing camps and other events)

Hobbies
Writing and reading blogs, working with Linux, cooking for family and friends, jogging, bicycle tours, hiking,
building igloos

Scientific Publications
Matthias Stuermer 2009 “How Firms Make Friends: Communities in Private-Collective Innovation” doctoral the-
sis under supervision of Prof. Georg von Krogh, Chair of Strategic Management and Innovation, ETH Zürich.
Matthias Stuermer, Sebastian Spaeth, Georg von Krogh 2009 "Extending private-collective innovation: a case
study " R&D Management 39(2), pp. 170-191.
Sebastian Spaeth, Matthias Stuermer, Georg von Krogh 2009 "The Push Model of Open Innovation" International
Journal of Technology Management, Special Issue on Open Innovation, forthcoming.
Georg von Krogh, Matthias Stuermer, Markus Geipel, Sebastian Spaeth, Stefan Haefliger 2009 “How component
dependencies predict change in complex technologies” working paper.
Georg von Krogh, Sebastian Spaeth, Guido Hertel, Matthias Stuermer 2009 “Developer motivation in collabora-
tive innovation: The impact of perceived firm characteristics“ working paper.
Sebastian Spaeth, Georg von Krogh, Matthias Stuermer, Stefan Haefliger 2008 "A Lightweight Model of Compo-
nent Reuse: A Study of Software Packages in Debian GNU/Linux" working paper.
Sebastian Spaeth, Matthias Stuermer, Stefan Haefliger, Georg von Krogh 2007 “Sampling in Open Source Soft-
ware Development: The case for using the Debian GNU/Linux Distribution” Proceedings of the 40th Annual
Hawaii International Conference on System Sciences (HICSS).
Matthias Stürmer 2005 “Open Source Community Building” licentiate (master thesis), University of Bern.

How Firms Make Friends: Communities in Private-Collective Innovation 69


Other Publications
Matthias Stürmer “Kostensparende und nachhaltige ICT-Lösungen mit Open Source Software” InfoWeek 01/2009
Matthias Stürmer “Open Source Software und die öffentliche Hand - Ein nachhaltiger Beitrag zur Wissensge-
sellschaft und zum Werkplatz Schweiz” Publikation der Swiss Open Systems User Group /ch/open, 2009
Matthias Stürmer “Warum das Open Source-Innovationsmodell nachhaltig ist” Goldwyn Report 05/2007
Matthias Stürmer “Soll Office Open XML ein ISO Standard werden?" Computerworld Schweiz, #36-2007
Matthias Stürmer “Community Building” TYPO3 Magazin 1/2006
Matthias Stürmer, Thomas Myrach “Open Source Community Building” Open Source Jahrbuch 2009
Matthias Stürmer “Business Opportunities of Companies Active in Open Source Projects” unpublished working
paper, 2005

Supervised Theses
“Open Source Strategie für eine Versicherung”, Master Thesis, August 2009
“Einfluss von Warentests auf Konsumenten und Anbieter”, Master Thesis, August 2009
“Entering a foreign country: Case Study of an Open Innovation-based company”, Master Thesis, August 2009
“Sustaining Paradise – Developing Indicators for a Sustainable Tourism Strategy ”, Master Thesis, July 2009
“Entwicklungen von Ernährungsgewohnheiten der Trends für Convenience-, Gesundheits- und Nachhaltigkeit-
sprodukte”, Master Thesis, July 2009
“Bewertungsmodell von Immobilien”, Master Thesis, July 2009
“Strategie und strategisches Management in der Stadtentwicklung”, Master Thesis, March 2009
“The theoretical evaluation of the ‘Telescopic Observations’ framework as a new tool in strategy development for
its incorporation into an existing firms overall planning methodology”, Master Thesis, November 2008
“The Fundamentals of Strategic Logic and Integration for Merger and Acquisition Projects”, Master Thesis, Au-
gust 2008
“Businessplan für das Unternehmen Redwood Fessler”, Semester Thesis, July 2008
“Business Models for Open Source Software in the Mobile Device Market”, Master Thesis, May 2008
“Targetmarketing von Mikrofinanz-Investmentprodukten”, Master Thesis, May 2008
“Bugzilla Analysis”, Semester Thesis, April 2008
“Wachstumsoptionen eines Schweizer Logistikdienstleisters”, Master Thesis, April 2008
“Evaluation eines neuen Handelsystems für Strukturierte Produkte auf Aktien zur Optimierung des Kundenser-
vice”, Master Thesis, March 2008
“Open Source Software: An Insight to FLOSS Business Models and Profitability ”, Master Thesis, Sept. 2007
“Adaption eines sachgutbetrieblichen Produktionsstrategie-Modells für den Bankbetrieb”, Master Thesis, May
2007
“Demografische Untersuchung der Eclipse Core-Teams”, Master Thesis, March 2007

How Firms Make Friends: Communities in Private-Collective Innovation 70


Presentations
“Nachhaltige, innovative eGovernment-Lösungen basierend auf dem Open Source Community-Prinzip” Novem-
ber 17th, 2009 in Fachsession of eGovernment-Symposium 2009, Bern, Switzerland
“Digitale Nachhaltigkeit: Auswirkungen für Business und Gesellschaft” November 12th, 2009 at World Usability
Day, Technopark, Zürich, Switzerland
“Digitale Nachhaltigkeit: Zusammenhang zwischen Open Source Software, Offenen Standards und freiem Wis-
sen” October 13th, 2009 at Fachtagung „InterOperabilität - vom Stand der Technik zu Zukunftsstrategien auf eu-
ropäischer Ebene“ in Vienna, Austria
“Digitale Nachhaltigkeit in der Schweiz – ein Bericht zur Lage der Nation” September 23th, 2009 at OpenExpo,
Switzerland
Participant of panel discussion "Proprietäre Software versus Opensource – Erfahrungen im Dokumenten Manage-
ment" September 18th, 2009 at eGov Fokus “Dokumenten-Management und Langzeitarchivierung”, Kompe-
tenzzentrum Public Management und E-Government, Berner Fachhochschule Wirtschaft und Verwaltung,
Switzerland
“Netzzunft-Treffen: Politik 2.0 Teil 1 – Digitale Nachhaltigkeit” July 29th, 2009 at ETH Zürich, Switzerland
“How firms make friends: Communities in Private-Collective Innovation” July 16th, 2009 at LIIP in Zürich,
Switzerland
“Open Source Lobbying im Bundeshaus” May 6th and 20th, 2009 at Open Business Lunch in Zürich and Bern,
Switzerland
“Open Source Software für die öffentliche Hand: Ein nachhaltiger Beitrag für die Wissensgesellschaft Schweiz”
March 4th, 2009 at the InfoSocietyDays 2009 in Bern, Switzerland
“Incentives and costs in implementing Private-Collective Innovation: A case study” October 16th, 2008 in the
cege-Forschungskolloquium, Faculty of Economic Science at Georg-August-Universität Göttingen, Germany
“The Push Model of Open Innovation“ August 13th, 2008 in the Strategy, Technology, and Innovation Track at the
Academy of Management Annual Meeting 2008, Anaheim, USA
“Incentives and costs in implementing Private-Collective Innovation: A case study” August 12th, 2008 in the New
Product Development Track at the Academy of Management Annual Meeting 2008, Anaheim, USA
“Code Reuse, Motivation, Koordination, Kollaboration: Was Entwicklerteams von Open Source Communities ler-
nen können”, May 28th, 2008 at Internet Briefing Software Development Conference, Zürich, Switzerland
“Enabling Knowledge Creation through Outsiders: Towards a Push Model of Open Innovation”, May 27th, 2008
at ETH Zurich, MTEC Research Seminar, Zürich, Switzerland
“Incentives and Costs in Implementing Private-Collective Innovation: A Case Study”, May 15th, 2006 at Euro-
pean Academy of Management 2008, Ljubljana, Slovenia
"Open Source Software und die öffentliche Hand: Chancen und Herausforderungen für die Schweiz", November
6th, 2007 at eGovernment Symposium, Bern, Switzerland
"Chancen und Herausforderungen von Open Source Software in Schulen", September 5th, 2007 at ICT conference
of PH Bern
"Warum das Open Source-Innovationsmodell nachhaltig ist", September 2nd, 2007 at DIGICOMP Open Tuesday,
Zürich
"Community Building als Wettbewerbsvorteil am Beispiel von Nokia", June 14th, 2007 at DIGICOMP OpenDay,
Zürich
"Open Source Community Building: Perspektiven auf Makro- und Mikro-Ebene" June 6th, 2007 at Arbeitsgruppe
Open Source der Schweizerischen Informatikkonferenz (SIK), Bern
"Linux-Einstieg leicht gemacht mit Ubuntu" June 2nd, 2007 at Open Source Software im Unterricht, Zürich

How Firms Make Friends: Communities in Private-Collective Innovation 71


"Crowding Effects: How Money Influences Open Source Projects and its Contributors" June 1st, 2007 at Linux-
Tag in Berlin
"Community Building als Wettbewerbsvorteil: Fallstudie zum Nokia Internet Tablet" March 8th, 2007 at OpenEx-
po in Bern
"Issues of Company Involvement in OSS Projects: The Example of the Nokia Internet Tablet" February 25th,
2007 at FOSDEM in Brussels
"Linux umsteigen, jetzt: Ubuntu" January 16th, 2007 at EB Zurich
"Sampling from the Debian GNU/Linux Distribution: Software Reuse in Open Source Software Development"
January 4th, 2007 at the 40th Hawaii International Conference on System Sciences (HICSS)
“Open Source Software Research: Communities, Companies and the Future” March 15th, 2006 at Puzzle, Bern
"Begrüssung zur 3. OSS Tagung für die öffentliche Verwaltung" November 20th, 2006 at the 3rd Open Source
Conference of Public Administration, Bern
“Open Source Community Building” November 11th, 2005 at the /ch/open Open Business Lunch, Zürich

How Firms Make Friends: Communities in Private-Collective Innovation 72

Related Interests