You are on page 1of 22

Extending private-collective

innovation: a case study


Matthias Stuermer, Sebastian Spaeth and
Georg von Krogh
ETH Zurich, Department of Management, Technology and Economics, Zurich, Switzerland.
sspaeth@ethz.ch

The private-collective innovation model proposes incentives for individuals and firms to
privately invest resources to create public goods innovations. Such innovations are character-
ized by non-rivalry 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 research 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-collective innovation model by analyzing the hidden costs
for the company involved. We examine the development of the Nokia Internet Tablet, which
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 the actions taken by the management to mitigate these costs throughout the
development period.

1. Introduction forfeited and the resulting innovation is offered to


the public for free. Examples of the private-

I n a private investment model of innovation,


firms use internal processes to create ideas,
knowledge, and technologies and commercialize
collective innovation model include collaborative
composing of music on the Internet by many
musicians, the open and collective development
these in the market place. Firms appropriate of a drug formula for treating malaria, or the
returns from private investment in innovation creation and sharing of new designs for sporting
through intellectual property rights (Granstrand, equipment among sports enthusiasts. An often-
1999). This model is contrasted with the private- cited example of the model is open source soft-
collective model of innovation, where firms and ware development resulting in products such as
individuals expend private resources to create Linux, MySQL, or Apache. Open source software
public goods innovations. Such innovations are comes with licenses that make it non-exclusive:
characterized by non-rivalry and non-exclusivity the software is free for all to download, use,
in consumption (von Hippel and von Krogh, modify, and redistribute. Open source software
2003, 2006). It is similar to ‘open innovation,’ is also characterized by non-rivalry as one per-
pertaining to models of innovation where firms son’s use of the product does not diminish anyone
frequently exchange ideas, knowledge, and tech- else’s benefits from using it.
nology with outside firms and individuals (Ches- Although researchers have examined indivi-
brough, 2003). However, open innovation does duals’ motivations to participate in open source
not assume that intellectual property rights are software development, to date, there has been

170 R&D Management 39, 2, 2009. r 2009 The Authors. Journal compilation r 2009 Blackwell Publishing Ltd,
9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
Development of the Nokia internet tablet

limited empirical examination of a firm’s incen- novations are made freely available to all as
tives for private-collective innovation. Moreover, public goods. Science is often cited as an example
the literature has emphasized the benefits the of this model. However, companies have the
model brings to the innovator rather than the option to free-ride on public goods innovations,
costs and has not discussed how the latter could such as, for example, a biotechnology company
be mitigated. Research has shown that the im- commercializing scientific knowledge on genetics
plementation of new models of innovation often without contributing research back to the
has unintended consequences, including ‘hidden scientific community. Therefore, society elects
costs’ (e.g., Crawford, 1992), and there is a need to subsidize the activity of innovators, e.g.,
for more empirical work on the benefits and costs university-based research on the human genome
of implementing private-collective innovation. funded by the government.
In this paper, we advance empirical research on Recently, a third model, called the private-
the incentives and costs of the private-collective collective model of innovation incentives, has
model of innovation. A case study design allows been suggested where public subsidy is absent
an in-depth investigation of benefits, costs, and and where the innovator expends private re-
actions only partly discussed in prior work (von sources for public goods innovation (von Hippel
Hippel and von Krogh, 2006). Our case draws on and von Krogh, 2003). The model is counter-
quantitative and qualitative data from the crea- intuitive: why should I make my innovations
tion of the Nokia Internet Tablet. Nokia based available to all and why pay for something that
the development of this product mainly on open anyone else can use for free? Generally speaking,
source software and made a large part of the in the case of private-collective innovations, the
research and product development transparent innovator receives higher benefits when contribut-
and accessible as a ‘public goods innovation.’ ing to the public goods creation than by only free-
Outside contributors involving firms and indivi- riding on its production by others. One aspect is
duals, unpaid by Nokia, expended a significant the privately retained tacit knowledge that inno-
amount of private resources on its development. vators receive through the production of freely
The paper is structured as follows: in the next available knowledge that distinguishes them from
section, we discuss the private-collective model pure users of the explicit knowledge. This implies
for innovation incentives. The third section de- that firms receive certain benefits during the
scribes the research design, and the fourth section process of creating publicly available innovations,
contains the case description. The fifth section while the mere application of such knowledge
presents the findings organized along the topics of bears less incentives. (see also Grand et al.,
benefits and costs incurred in the implementation 2004; Gächter et al., 2006)
of the model and strategic actions to mitigate
these costs. Finally, we conclude the paper and
outline implications for management practice and
2.1. Benefits when applying
future research.
private-collective innovation
A closer examination of the model outlined in von
2. The private-collective innovation model Hippel and von Krogh (2006) and other literature
reveals six complementary benefits for firms to
There are two predominant models of innovation innovate in this manner: the cost of controlling
incentives in the technology and innovation man- knowledge, learning, reputation gains, and fast
agement literature. The private investment model and widespread diffusion of innovations, as well
assumes that innovators step forward and invest as lower cost of innovation and manufacturing.
in innovation if and when they can appropriate First, in the long run, the cost of protecting
returns from these investments. Intellectual prop- knowledge needed to innovate (Liebeskind,
erty rights is a necessary condition for the model 1996) might outweigh the benefits of doing so.
because it safeguards returns appropriation (Ar- Often, extensive investments in knowledge man-
row, 1962; Dam, 1995). In contrast, the collective agement systems are needed to protect informa-
action model assumes that innovators, provided tion that ultimately and inevitably spills over to
with the right public subsidy, contribute to public the public (Alavi and Leidner, 1999; Foray, 2004).
goods innovations (David, 1992, 1998; Stiglitz, For example, although the source code for Sony’s
2006). Public goods are characterized by non- robot dog Aibo was protected, it was ultimately
rivalry and non-exclusivity in consumption. In- hacked and published by Sony’s customers.

r 2009 The Authors R&D Management 39, 2, 2009 171


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

Second, innovators who contribute to collective software, which should have a positive impact on
goods innovations benefit from learning from the cost of innovation (see Haefliger et al., 2008).
their own and others’ contributions. In addition However, much of this software comes with
to benefits garnered from the public goods itself restrictive open source licenses, which requires
(products and services), innovators also benefit the firm to make any combination between this
from learning in the process of creating it (Allen, and other software adhere to this license. Hence,
1983; Nuvolari, 2004; Baldwin and Clark, 2006). while the reuse of such products may reduce the
Thus, it should come as no surprise that many costs of innovation, it also ‘forces’ the firm to
contributors to open source software projects are contribute to public goods innovations.
computer science students. By providing open Sixth, the supply of public goods innovations
source software to the public, contributors may to the market makes it possible for manufacturers
get others to use it, test it, and provide feedback to learn about these innovations and reduce
on how to improve it (Lakhani et al., 2002). Some manufacturing-incurred fixed costs related to re-
authors have even referred to open source soft- search and development. Free access to innova-
ware projects as ‘epistemic communities,’ where tions may incentivize manufacturers to ramp up
people create shared knowledge of software de- manufacturing capacity, pursue economies of
velopment (Edwards, 2003). scale, and reduce the price of manufactured
Third, innovators may gain a positive reputa- products. Additional benefits to the customers
tion by privately expending resources for public of the manufactured product may include en-
goods innovations (Allen, 1983; Lerner and hanced product quality and product warranties
Tirole, 2002; Muller and Pénin, 2006). For ex- (Kotha, 1995; Harhoff et al., 2003).
ample, PricewaterhouseCoopers gains a positive
reputation among regulators, peers, and custo-
mers when they provide research to the public on
practices of corporate governance. The firm’s
2.2. Empirical evidence and hidden costs
reputation is further enhanced when regulators Empirical research on the private-collective inno-
actively use and reference the research during vation model is mainly found in the field of open
standard setting in principles of auditing and source software development, where the focus has
corporate governance (PricewaterhouseCoopers, been on individual contributors and projects (e.g.,
2005). Baldwin and Clark, 2006; Gambardella and Hall,
Fourth, being the first to contribute a public 2006; Roberts et al., 2006; Shah, 2006). Research
goods innovation increases the likelihood of ben- on the application of the model by firms is rare,
efiting from fast and widespread adoption of the with some exceptions: Dahlander (2004) explored
innovation (Allen, 1983). As a consequence, firms the network effects available to firms that provide
may gain an advantage over competitors stem- open source software to the public. Jeppesen and
ming from network effects (von Hippel and von Frederiksen (2006) explored users’ motives to
Krogh, 2003). Establishing a ‘dominant design’ or contribute voluntarily to the development of
an ‘open standard’ onto which the firm can fit media products by firms. Henkel (2006) investi-
other technologies and even preempt the intro- gated firms that revealed open source software
duction of competing technology may provide the embedded in their devices to other firms and
firm with additional advantages (Economides, found incentives for them to do so. Given the
1996; see also Economides and Katsamakas, focus on public goods innovations in these works,
2006). the authors have tended to focus on the cost of
Fifth, by contributing to public goods innova- forfeiting intellectual property rights in the pri-
tions, the firm may lower the cost of innovation. vate-collective innovation model.
Chesbrough (2003) argued that involving outside Innovation research has pointed to several
firms, organizations, and individuals in the devel- types of ‘hidden costs’ in the implementation of
opment of products reduces the direct labor cost new innovation models that are not inherent to or
in innovation. In addition, when the firm con- captured in the models themselves. For example,
tributes to public goods innovations, such as open Crawford (1992) and Smith (2004) pointed to the
source software, it can also effectively reuse costs of implementing the accelerated product
existing technology found in the public sphere. development model in US manufacturing indus-
Research has shown that, in software develop- tries. Firms often find that the rapid launching of
ment, the reuse of open source software is con- products that have not been properly tested leads
siderably higher than the reuse of firm-internal to costly recalls, or that fast product development

172 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

leads to significant delays in pilot- and full-scale operating system Linux, its strategy of a conspic-
manufacturing. With respect to open innovation, uous commitment to open source software and its
Kessler et al. (2000) found some indication that devotion to building a community of volunteers
the cost of product development increased with are unique for a multinational corporation in the
increasing dependency on external sources of consumer electronics industry. In order to deter-
technology in the innovation process. Empirical mine the distinctiveness of the case, we compared
research is needed that validates the specific it with a wider sample including Sharp, Motorola,
benefits in the implementation of private- Hewlett-Packard, and Sony.
collective innovation, examines the costs incurred In the 1990s, Sharp introduced a Personal
in such innovation, and identifies firms’ strategies Digital Assistant called Zaurus, which ran on an
to mitigate these costs. We contribute to this adapted Linux operating system. Soon after,
research by investigating the implementation of voluntary developers programmed a variety of
private-collective innovation in the case of the applications for the device and started a vivid
Nokia Internet Tablet development. community of developers (among them one of the
interviewees). However, in contrast to Nokia,
Sharp did not reveal much of its own source
3. Research design code, scarcely supported the community and its
interests, and eventually stopped selling the device
The research on the development of Nokia’s in Europe and the United States.
Internet Tablet focuses on the process of imple- Motorola brought Linux-based cell phones to
menting private-collective innovation. We investi- the market in 2003. The development of their
gate benefits in implementation and extend the operating system was carried out by MontaVista,
private-collective innovation model by identifying a vendor of embedded Linux software, and Troll-
implementation costs and strategies to mitigate Tech, the provider of the graphical user interface
these. Research on implementation processes typi- ‘Qt.’ In contrast to Motorola, Nokia did not rely
cally requires longitudinal observation (Pettigrew, on a few service providers to implement the soft-
1990), prompting a case study design. In order to ware but collaborated with many, mainly small,
obtain insights into the development process, we software firms in an open fashion. By using open
gathered different types of data and performed source software, Motorola expected to cut costs
quantitative and qualitative analysis. Done prop- and speed up software development, because they
erly, such combined analysis offers a valuable did not pay per-unit royalties and built applica-
insight, as Shah and Corley (2006) have recently tion software on the existing open source software
argued. In the following sections, we describe the (Shankland and Charny, 2003). Similar to Sharp,
sampling, data sources, and data analysis. however, Motorola neither revealed source code
beyond legal requirements nor did the firm pro-
vide extensive developer documentation of the
software on its devices. Moreover, their Linux
3.1. Sampling appliances granted no administrator access to the
Our research design is a single-case study focuss- user, inhibiting the installation or modification of
ing particular attention on sampling (see Eisen- native applications. Other mobile devices running
hardt, 1989). There are three reasons for selecting embedded Linux included HP’s iPAQ and the
a particular case: fit, distinctiveness, and its re- Sony Mylo. However, as we discovered in inter-
velatory nature (see also Yin, 2003; Siggelkow, views and in press articles, these companies re-
2007). First, Nokia’s Internet Tablet development tained the software’s source code and we could
represents a case that both serves to explore and find no evidence that these firms attempted to
extend the private-collective model. The Internet build up a community of outside developers as
Tablet was created by private investments by proposed by the private-collective innovation
Nokia, building on existing open source products model (von Hippel and von Krogh, 2003).
as well as releasing a substantial amount of Third, given the research gap on the implemen-
knowledge (source code). tation of private-collective innovation, we also
Second, the distinctiveness of the case is pro- searched for a revelatory case. The main criterion
vided through Nokia’s unique approach in pro- for selecting a revelatory case is the researchers’
ducing a product based on open source software. access to a previously inaccessible setting for
While Nokia is not the first manufacturer to scientific observation. Establishing ties to Nokia
create a mobile device based on the open source and the developer community surrounding the

r 2009 The Authors R&D Management 39, 2, 2009 173


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

Internet Tablet, the researchers gained access to a contractors, and eight with unpaid volunteers
variety of data including documents, interviews, (see Table 1). In order to protect their privacy,
prototypes, and online conversations. Shedding interviewees were anonymized. The interviews
light on the reasons for and the effects of this lasted on average 75 min. The interview guidelines
innovation project going open and abstracting included questions on the firm–community rela-
these underlying intentions into a model that can tionship, the strategies Nokia used to reveal
be used in future research, as well as raising the knowledge and technology to the community,
attention of practitioners to this mode of product motivation, and other issues (see the Appendix
development, motivated the selection of the case. A for examples of two distinct interview guide-
lines). The initial interview guidelines were up-
dated and enhanced over time, integrating and
building upon the results of interviews already
3.2. Data sources analyzed. All interviews were transcribed verba-
This study relies on several sources of data. First tim and, using the software Max.QDA, codified
and most importantly, we conducted semi-struc- using an open coding technique (for a discussion,
tured interviews, allowing participants the oppor- see also Strauss and Corbin, 1998). This led to the
tunity to narrate stories, provide anecdotes, and creation of 80 codes, which were subsequently
state opinions. Through an initial reading of the merged and reduced to 12 categories, including
mailing lists, relevant stakeholder groups and seven incentive and five cost categories.
data sources in the development project were The second data source consisted of the pro-
identified as ‘Nokia employees,’ ‘Nokia-paid con- ject’s user and developer mailing list. The
tractors,’ and ‘independent individuals.’ Inter- monthly archives were downloaded from their
views were conducted with participants from all inception in May 2005 until the end of December
the stakeholders identified. The initial partici- 2006. The archival data were examined using the
pants were selected from the developer and user statistical software ‘R’ in order to indicate the size
mailing lists, and subsequent interviewees were and activity of the community. The resulting
identified through snowball sampling (Hecka- statistics are presented later in the text.
thorn, 1997). In total, 23 interviews were con- Third, one co-author followed the developer
ducted; 10 with Nokia employees, five with mailing list over the course of several months

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
# Key Role Date Duration Contribution, function

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

174 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

and observed the project’s chat channel on IRC tute (ETSI) in 1990. During this period, the
(Internet Relay Chat), which is not publicly company also divested its other businesses. In
archived. Often, informal discussion takes place 1998, Nokia overtook Motorola as the world’s
on IRC, giving the researchers a feeling for what largest mobile phone manufacturer (ICFAI,
‘really happens’ in the community. Such online 2005). R&D operations at Nokia have always
participant observation is rather uncommon in been scattered across the world, working in a
the research on open source software develop- dispersed, non-hierarchical structure, allegedly to
ment but is deemed both necessary and helpful in prevent the development of ‘tunnel vision.’ In
understanding the unfolding dynamics of the 2003 and 2004, Nokia suffered a decline in the
Internet Tablet development. No formal analysis market share of its mobile phone business. The
was performed with the data but being immersed company had misinterpreted the market demand
in the community helped to interpret mails and for ‘clamshell’ devices and camera phones and
understand the issues raised in the interviews. had failed to adapt fast enough to these new
Fourth, secondary sources, such as news re- developments (ICFAI, 2005). However, in 2007,
ports, blogs of Nokia members and volunteers, Nokia posted EUR 51.1 bn of net sales and an
and corporate web sites of Nokia and contracting operating profit of EUR 8.0 bn, spending EUR
firms were included in the case study database. 5.6 bn on research and development.
For example, these sources provided additional
information on the extent to which Nokia spon-
sors other open source projects. Some Web arti- 4.1. Internet Tablet History
cles were used to get background information on
the Nokia device and potential competitors. We According to N1, Nokia started to experiment
also searched an independent web forum for with incorporating open source products, specifi-
discussions and opinions of users of the Internet cally based on the Linux kernel, into their devices
Tablet. in 2000. At the same time, the company sought to
develop a device that would take advantage of the
increasing availability of wireless access points
4. Nokia and the development of the and provide access to Internet appliances every-
Internet Tablet where. N1, the head of the software development
team, summarized the vision for this new type of
In this section, we present a short overview of the mobile device as follows:
history of Nokia and the Internet Tablet devel-
opment, and we provide a descriptive analysis of At the same time, totally independently, we had
Maemo, the community for the Internet Tablet another stream of thought which was this kind
software platform. The purpose of this analysis is of category of Internet Tablets. The big idea
to confirm the correctness of the case to examine behind that was really the same way mobile
the implementation of the private-collective phones liberated voice. Not only houses or
model for innovation incentives. offices have phone numbers, but people have
Nokia was originally set up in 1865, producing numbers. So you can take the phone wherever
pulp and paper. It underwent a series of remark- you go. We have the same vision that we want
able transformations in its business model. In to do the same thing with the Internet and
1967, it merged with the Finnish Rubber Works Internet use cases. You don’t need to fire up a
Ltd. and the Finnish Cable Works, forming the PC and you don’t need to go to your desk. You
Nokia Corporation with four major businesses: have this very light portable device that gives
forestry, rubber, cable, and electronics. A diversi- you access to the Internet. Whether that is
fied company, with a product portfolio ranging browsing, email, chat, VoIP. (N1)
from tires to television sets, it first started produ-
cing mobile phones in 1981, manufacturing car Inspired by this vision, Nokia designed an overall
phones for the first international cellular mobile software architecture of the operating system
phone network. The first hand-portable phone based on open source components and partly
sets were introduced in 1987. During the 1990s, adapted these components themselves and partly
Nokia focused on telecommunications, especially contracted developers for specific implementation
on mobile phones based on the then emerging tasks. In 2002, individuals who were active in
GSM standard, which had been published by the architecture-crucial open source projects were
European Telecommunications Standards Insti- approached by Nokia and asked to perform

r 2009 The Authors R&D Management 39, 2, 2009 175


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

some tasks as contractors for the company (e.g., 4.2. The Maemo Community
C1, C2, and C3). They had to sign a non-dis-
The operating system and the software of the
closure agreement (NDA) that prevented the
Nokia Internet Tablet are based on the Maemo
leaking of much information until the Internet
software platform, an effort led by Nokia and
Tablet went public.
announced the same day the device was launched.
The prototype device was first publicized on
The intention behind the platform was to provide
May 25, 2005. At the GNOME User and Devel-
open source components usually deployed on
oper European Conference on May 31, 2005,
Linux desktop distributions and to adapt and
Nokia announced that it would give away 500
enhance these for the environment of handheld
devices for about a third of the regular sales price
devices (Maemo.org, 2006). Nokia had down-
to selected software developers and donate the
loaded the open source GIMP Toolkit (GTK)
sales to the GNOME foundation, a not-for-profit
graphical toolkit and other components such as
organization dedicated to supporting the
the GStreamer framework for multimedia and
GNOME graphical desktop environment (Nokia,
modified these to fit the needs of an embedded
2005). The sales of the product did not commence
device with restricted hardware resources. Nokia
until November 2005, when the 770 Internet
also added their own software developments and
Tablet was officially released with a price of
parts from independent software vendors as pro-
USD 439. Customers could order the product
prietary software, protected by commercial soft-
via a Nokia Internet Tablet-dedicated Web site or
ware licenses and released as binary code only.
through the official Nokia Web shop. However,
Figure 1 provides an overview of the software
little effort went into promoting the Internet
architecture distinguishing between software pub-
Tablet to the wider public. The product was not
lished under an open source software license,
available through other distribution channels
commercial software components published by
such as local Nokia mobile phone shops1.
third-party vendors, and Nokia’s proprietary
While the Internet Tablet 770 was still being
components.
sold throughout 2006, its successor was an-
The Maemo platform uses its own infrastruc-
nounced by Nokia’s CEO Olli-Pekka Kallasvuo
ture, such as a revision control system, a software
on January 8, 2007, at the International Consu-
bug tracker, and mailing lists that allow commu-
mer Electronics Association show in Las Vegas.
nication between developers and users. In No-
The successor product was added to the N-series
vember 2006, the Maemo.org site had 54,000
of Nokia devices, a popular brand for Nokia’s
unique visitors. While the operating system con-
major communication products, and marketed to
tains proprietary software (some hardware dri-
the ‘mainstream’ public as the N800. On October
vers and applications such as the Opera Web
17, 2007, the third generation of the Internet
browser), the main software platform, Maemo,
Tablet, the N810, was presented in San Francisco
was open source and developed by the Maemo
at the Web 2.0 Summit. Nokia executive VP and
community.
head of the company’s multimedia business unit,
A source code repository allows developers to
Anssi Vanjoki, stated ‘The N810 is the first of
add new software components to the product and
these devices targeted at a ‘normal’ consumer
group, beyond 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 4.1300 display with – given its size – an
unusually high resolution of 800  480 pixels,
which, using a stylus, can be utilized as a touch
screen. It connects wireless through Bluetooth
(connecting to a mobile phone) or through a
common WiFi to the Internet. It is also possible
to connect to another PC through the integrated
USB port. The device has no hard disk, but flash-
based storage is included that can be extended
with external flash storage media. Since the N800,
a VGA webcam has been integrated and in the
N810 a GPS receiver is also included. Figure 1. Maemo software stack.

176 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

upload improved software that resolves problems solutions and system administration tools to an
and bugs in previous versions. The Maemo source Electronic Flight Information System for use with
code has its own repository of source code. As small aircrafts.
of January 2007, our descriptive analysis shows Most communication occurs through mailing
that 33 developers added more than 7.2 million lists. For discussion purposes, Maemo has a
lines of code to this repository, forming the developer list where actual technical development
core of the operating system (although much of issues are discussed and a user mailing list for
this is unmodified code from other open source discussing issues related to the operation and use
projects). Figure 2 visualizes growth in the of the product. We examined the Maemo devel-
source code over time. It should be noted that oper mailing list from its inception in June 2005
the developers were exclusively employees of until December 2006. In total, 832 participants
Nokia or formally affiliated with the Maemo identified by their email address contributed to
project, indicating that Nokia retains tight con- the list, with 79 participants (9.5%) posting from
trol over the actual changes that occur to the core an official @Nokia.com or @maemo.org email
system in the software architecture (see also Kuk, address. The monthly number of postings to the
2006). list ranged from 98 to 548, with a median of 328
In order to help identify and remove software (mean 339.8). 19.6% of all mails were sent by
bugs, Maemo also has a so-called ‘bug tracker’ of Nokia addresses. Out of the top 15 posters, only
its own. This is used to enter software errors, or four had an official email address from Nokia/
bugs, and keep track of the bug-fixing process. Maemo. Communication on mailing lists is orga-
Our analysis shows that in June 2008, this tracker nized in so-called ‘threads.’ Often, a thread is
contained 3228 bugs, of which 1133 bugs were started by someone asking a technical question,
marked as ‘open,’ meaning that they were waiting followed by replies by others who attempt to
to be fixed. discuss and answer the question. One hundred
The Maemo project infrastructure also offers a and eighty three threads were started by Nokia
repository, garage.maemo.org, where people in- affiliates, while 931 mails were replies to an
dependently of Nokia can register their projects existing thread. Non-Nokia participants, on the
and use the developer infrastructure for free. Our other hand, started 1670 threads and posted 4010
analysis shows that in June 2008, Garage con- replies. A w2 test of the frequency contingency
sisted of 615 active projects and 12,446 registered table confirms that Nokia participants differ sig-
users. Projects led and contributed to by indepen- nificantly in the thread start/reply frequencies
dent individuals ranged from GPS navigation from non-Nokia participants: Nokia members

Figure 2. Lines of code in the Maemo Subversion code repository.

r 2009 The Authors R&D Management 39, 2, 2009 177


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

were more likely to post a reply to an existing related to the implementation of private-collective
thread rather than start a new one. innovation and strategies to mitigate these costs.
Besides discussing development, there is also a The benefits and costs of this extended model are
Maemo user mailing list that serves to support summarized in Tables 2 and 3.
general user discussion. This list follows similar
patterns although it is less frequented than the
developer mailing list. Its number of postings per
5.1. Benefits in implementation
month ranged from three to 253, with a median of
133 and a mean of 127 mails. As can be expected, First, considering the reduced cost of knowledge
user-related discussions started to pick up after protection, product development managers at
the device was introduced to the market in No- Nokia were conscious that costs stemming from
vember 2005, whereas development discussions protecting proprietary knowledge and costs of
preceded this point. Our analysis shows that 511 voluntarily revealed knowledge imply a trade-
participants contributed to the list, 6.46% of them off. For the Internet Tablet platform, Nokia
from Nokia. Only two out of the top 15 posters decided to keep some components proprietary,
identified themselves as belonging to Nokia. while developing other components as open
10.9% of all mails were sent from Nokia employ- source software. According to the interview with
ees. These started 29 new threads and replied to N1, about 25% of the software is unmodified
220 existing threads. Non-Nokia participants, on open source software. Another 50% consists of
the other hand, started 775 new threads and sent existing open source code to which Nokia made
1510 mails to existing ones. Also on the user improvements or adaptations and that was again
mailing list, a w2 test confirmed that Nokia released under an open source license. Only 25%
affiliates were significantly more likely than non- comprises closed code that Nokia either imple-
Nokia participants to reply to existing threads mented from scratch or that is closed commercial
rather than start new ones. (proprietary) software from a commercial vendor.
Our brief examination of the Maemo commu- N1 stated that one reason for not releasing source
nity, including source code development and code under an open source license was that Nokia
technical and user discussions, supports the state- wanted to keep control of the look and behavior
ment that the development of the Nokia Internet of end-user applications, rather than fearing a loss
Tablet is an implementation of private-collective of valuable knowledge. Disclosing the knowledge
innovation. Large parts of the product are a
public goods innovation and Nokia employees
as well as external individuals and organizations Table 2. Benefits in the implementation of Private-
expend considerable private resources (time, Collective Innovation and findings in the Nokia case
knowledge, and technology) to contribute to the Benefits Findings in the Nokia case
innovation. In the mailing list geared towards
Low knowledge Revealing source code rather
development, more than 80% of all emails were protection costs than protecting it; however,
sent by non-Nokia affiliates. In the next section, undetermined costs for revealing.
we present the findings from the case study. Learning effects Collaboration with external firms
and individuals
Reputation gain Increased attraction of Nokia as
an employer and for building
5. Findings their own developer community
Adoption of Standard setting of the platform
The following section presents findings on the innovation configuration
implementation of private-collective innovation Increased innovation Reuse of open source software,
in the case of the Nokia Internet Tablet develop- at lower costs outsourcing of software testing
and bug fixing and maintenance
ment. The aim of this section is twofold: first, we to open source communities.
illustrate a case of private-collective innovation Experimentation and
with empirical data, providing specific benefits for contributions of new applications
the company involved. The findings confirm the by lead users
six conjectures on the benefits derived above from Lower No licensing fees for software
manufacturing costs platform
the existing literature and additionally identify a NEW: Faster time- Tapping of distributed
benefit in the case: faster time-to-market. In to-market technology expertise and high
addition, we extend the model of private-collec- flexibility of software platform
tive innovation, highlighting the hidden costs

178 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

Table 3. Costs and possible mitigation strategies in the new possibilities with open source components in
implementation of Private-Collective Innovation order to build new competences. Although the
Cost Findings in the Mitigation strat- company could have chosen a commercial open
Nokia case egy source software integrator such as Red Hat or
Difficulty to Released source Partial revealing
Montavista, delivering ready-made solutions in
differentiate code can be reused of source code order to release a new product quickly, Nokia
by competitors to retain control decided to learn about the novel technologies
of look and feel internally in order to gain the potential for future
Guarding Plans for new pro- Selective reveal- innovations. N3 reported:
business ducts ing of future
secrets plans and pro-
tection of infor- [We wanted to make] something that is more
mation through than just putting a product into the market
NDAs
through learning about the possibilities of
Reducing Investments for Sharing the
community Software Develop- costs with other leveraging open source in deeper and more
entry ment Kit, preview actors in the substantial ways. As a result, we didn’t take
barriers version of platform, community the quick and simple approach, we went a little
device program, bit deeper and we learned more.
staff for community
management, and
increased commu- Building upon technologies developed mainly by
nication effort external contractors and volunteers implied that
Giving up Development direc- Hiring of key many outsiders, rather than the internal research
control tion such as scope developers and and development staff, had the most intimate
of functionality of participation in
knowledge of certain software components. The
open source pro- upstream com-
jects is controlled munities. No interview with contractor C2 revealed that in the
by external parties single vendor beginning, Nokia attempted to solve technical
controls plat- issues internally. However, when the company
form faced severe time pressure on a product launch
Organizational Required internal Adapting and
date, it would outsource the request to a trusted
inertia restructuring of opening up pro-
processes cesses external firm that specialized in the area of
technical development. For specific parts of the
Internet Tablet software, Nokia contracted sev-
eral small enterprises with experience in open
source components (see Table 4). Working with
was not free either; all source codes had to be these experts and listening to feedback from the
checked by Nokia’s department to ensure that Maemo community helped Nokia to rapidly learn
Nokia did not reveal intellectual property or new open source technologies and how those were
patents they did not have the right to. Altogether, produced by their communities.
it remains unclear whether Nokia saved costs by Nokia’s strategy was deliberately focused on
revealing knowledge rather than spending effort small companies, rather than on a few, large
to protect it. Nokia seems to see it as beneficial: contractors. Collaborating with small firms all
over Europe enabled Nokia to fill gaps in the
We have evidence that some of our competi- company-internal technical knowledge needed for
tors are now looking at our code and they are development, while still retaining control of the
investigating if they could use our code in their
products. You might say that we help them
now to get their products out fast. [. . .] But if Table 4. Some contractors to Nokia in the software
we had not put it out then we could not have development process
used the OSS communities who have already Company Expertise Country
helped us to develop that code. (N1)
KernelConcepts GPE and Embedded Linux Germany
Second, Nokia obtained learning effects by ex- OpenedHand Matchbox UK
Collabora Telepathy UK
perimenting with open source software technolo- Imendio GNOME and D-BUS Sweden
gies and by collaborating with contractors, firms, Fluendo GStreamer Spain
and volunteers. The platform had started as a Movial Scratchbox Finland
research project within Nokia, seeking to explore

r 2009 The Authors R&D Management 39, 2, 2009 179


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

core software architecture by having internal handling the open source interaction quite
experts in these areas (C2). well. I think they are quite a good open source
In addition to learning about the technologies citizen.
in use and acquiring the skills to work with them,
Nokia learned to cooperate with a diverse com- For Nokia, recruiting active open source contri-
munity of employees, volunteers, and contractors butors enabled them to select the best individuals
rather than relying only on contractors who could based on their prior contribution to the various
be forced to keep schedules and timelines: projects that constituted the development of the
Internet Tablet. In some cases, Nokia directly
It’s all about the process . . . You develop this contacted skilled open source software developers
openly within the communities and you try to and invited them to apply for jobs. Thus, most of
synchronize your own work with the heartbeat the Nokia employees working on the Internet
of the communities. Some companies now Tablet were previously active participants in
understand this better than others. We cer- open source projects and known to the Nokia
tainly have done our learning. We have made managers through prior collaboration. This fact
some mistakes too on this front. (N1) reflects that for many, Nokia became an interest-
ing company to work for, as C1 testified:
For example, one of Nokia’s major learnings
concerned the early decision to independently At that time, it was a dream come true. I did
continue development of the GTK, a collection not have a good job. I was spending all my
of software constituting the core of the graphical spare time hacking Matchbox [window man-
interface. Among others, C3 stated ‘the disadvan- ager of the Internet Tablet]. I was really
tage of doing this of having a forked or large enjoying it. And then you’re given a chance
patch was experienced by Nokia.’ By following a to get paid to do that full time. It was pretty
separate development stream, Nokia became dis- fantastic and an amazing piece of luck.
connected from the code maintenance effort by
the community and thus realized they had to According to interviews with Nokia development
move their changes back into the main open managers, Nokia employees selected through the
source project as architecture team leader, N3, community were highly motivated to continue to
reported: work on technologies they already knew. Possibly
intrinsic motivation, such as fun – often a primary
It was also a learning field that we had to go cause of contributions to open source develop-
through. If we had known then what we know ment – played a role in their continued high-level
today, we would have been able to do it with- efforts (see e.g., Torvalds and Diamond, 2001;
out such a large patch. We would have been Luthiger Stoll, 2006).
able to do more directly upstream in techni- Fourth, the private-collective innovation model
cally better ways with less effort for changing proposes that being the first to contribute a public
code. goods innovation increases the likelihood of fast
and widespread adoption of the innovation. Ac-
Third, the growing commitment to open source cording to the interviews, because the community
software development led to a reputation gain for was already familiar with underlying technolo-
Nokia, and interviewees V2, V3, and V4 sug- gies, adapting existing applications from other
gested that this led to the increasing attachment projects to the Internet Tablet platform proved
of volunteers and recruiting benefits. Particularly, to be a relatively easy task. Nokia also invited
the openness of the Maemo platform encouraged competitors to participate in the creation and use
volunteer users and developers to buy such a of their platform, citing a ‘the more the merrier
device, to improve the operating system, and to approach.’ By initiating a vendor-independent
create applications on top of it. Nokia was embedded software platform intended for use in
accredited with pioneering development of an other mobile devices, Nokia made it easier for
open embedded platform by users who enjoyed volunteers, contractors, and competitors to con-
running applications from third parties as for tribute. Spreading the innovation and inviting
example V6 stated: others to participate was seen as crucial:

Yeah, I definitely have a more positive view of We believe the world is changing and the
Nokia this way. Especially, I think they are competitive advantage comes from how many

180 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

others you can get to participate in this net- As such, both the costs of innovation were kept
work. (N1) low and ideas that would not have been devel-
oped otherwise could be tested and integrated:
In fact, in July 2007, Intel announced that they
were adopting the Internet Tablet’s user interface I can develop, say, twenty ideas a day and this
framework Hildon into their new product cate- community can develop a hundred ideas a day.
gory called Mobile Internet Devices (Paul, 2007), So it’s more important to be part of the
which will eventually lead to a higher developer community with a hundred ideas than by
and user basis of Nokia’s Maemo platform. yourself with twenty ideas. (N1)
Fifth, by contributing to public goods innova-
tions, firms can lower the cost of innovation. Investigating the software architecture, we found
Building on existing and mature technologies that most contributions by volunteers were sepa-
that could be integrated into the new hardware, rate applications that could be installed indepen-
Nokia enabled the development of a solid, yet dently from the core operating system. The
cheap operating system for embedded platforms. interviews showed that volunteers who made
Collaborating within existing open source pro- such contributions showed high commitment to
jects allowed Nokia to benefit from the collective and responsibility for their work, listening to user
programming efforts: feedback and, in some cases, when others de-
manded it, even enhancing their software against
So what is happing in the D-BUS, in the GTK, their personal belief of the usefulness of the
in the GStreamer, in the Linux kernel is that I features (explained by V2). Through their con-
put two guys there, IBM puts two guys, tributions and feedback from users, volunteers
Motorola maybe puts one guy, maybe Novell slowly gravitated towards more development
puts a couple of guys. So for the price of two work in the community.
guys, I get four or six guys working on the Sixth, supply by anyone of public goods in-
same problem. (N1) novations to the market enables manufacturers to
learn about innovations and thereby reduce costs
Additionally, the company benefited from volun- in manufacturing. This conjecture indicates parti-
tary contributions leading to enhancements of the cular benefits that private-collective innovation in
device. The volunteers contributed several inno- software offers to computer hardware manufac-
vations including applications, user interface im- turers (von Hippel and von Krogh, 2003).
provements, translations, bug reports and fixes, Through choosing a software platform that is
testing different peripherals, and making feature available under an open source license as indi-
requests. Allowing volunteers experiment with the cated, Nokia reduced fixed costs related to re-
software also created ‘proof of concepts’ which search and product development. Nokia
enabled innovations previously seen as unrealistic manufactured and sold the Internet Tablet, but
by Nokia engineers (V3). One example of this is the product’s functionality and, thus, ability to
the swap memory enhancement (using the flash fulfill user needs were, to a large extent, shaped by
memory as extended virtual memory) which was the users themselves. User-developed applica-
initiated by volunteer V4 and included in a sub- tions, such as mapping and navigation software,
sequent official version of the Internet Tablet’s could easily be installed by the end users them-
operating system. Thus, the likelihood of finding selves for free, keeping Nokia’s costs down.
a ‘killer application’ (see e.g., Downes and Mui, Interestingly, in addition to fixed cost reduction,
1998) in the process (by evolution or sheer luck) Maemo also has a positive impact on variable
increased. An open source software developer costs in manufacturing because Nokia did not
observed: have to pay a per-device license fee to an intellec-
tual property owner. For example, in the begin-
I think from my point, if you let people change ning of 2006, a comparable proprietary operating
things [. . .] and document them and open them system, Symbian, demanded USD 7.5 per device
up so people can hack their own stuff, you for the first 2 million units.
never know what is going to happen, what In addition to confirming theoretical conjec-
kind of things people are going to write for tures on benefits, one more benefit emerged in the
your device which ultimately could make it sell study that should be considered crucial for pri-
millions and millions if someone writes the vate-collective innovation: faster time-to-market.
killer application for it. (C1) Using external, modular technologies not only

r 2009 The Authors R&D Management 39, 2, 2009 181


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

impacted on costs, but it also led to the creation The OpenMoko actually bases on our stuff.
of a new operating system working on a new [. . .] the Moko window is a kind of base
hardware platform in a short time. According to window for their applications I think. There
interviews with N1 and N3, this fast development in the comments of the source file you can find
created flexibility, which, combined with user that this is based on the Hildon window by
feedback of pre-releases, allowed for a quick Nokia. They even have this copyright ‘Nokia
time-to-market compared with other devices the corporation.’ So they are based on our stuff.
company had launched. But, on the other hand, our stuff is not that
However, integration of the numerous open revolutionary. I mean that’s how it goes. We
source components into an executable environ- base on somebody else’s stuff too. (N4)
ment is very challenging as, for example, C3
explained. One way to rapidly tap into this Nokia’s strategy to mitigate this cost was to
knowledge of open source communities was selectively open up its software development.
achieved by Nokia’s strategy of contracting Figure 1 visualized how Nokia revealed the mid-
open source developers and small firms – the dle layer of software, the basic infrastructure,
‘bridges between Nokia and the communities’ under an open source license, while keeping parts
(N3): of the bottom layer (hardware specific software)
and much of the user-visible applications under
their own proprietary license. N1 argued that it
It’ll get done quicker and probably better if wanted to ensure a unique ‘Nokia look’ by
they pay us to do it. [. . .] Over the years we retaining control over applications, for example
have been involved, we have so much experi- of the device’s email program. In doing so, they
ence and knowledge, we know all the tricks. kept crucial parts closed (e.g., power management
We know how to get things like X and Match- and other hardware drivers) and prevented repli-
box up and running quickly on hardware. We cation of these parts by competitors. In all the
are just basically selling that knowledge as interviews conducted with volunteers, the respon-
well. Although they could very well likely dents demonstrated understanding for Nokia’s
figure it out themselves, we just can get them decision to selectively reveal source code,
there a lot quicker. (C1) although some stated their strong preference for
releasing all software under an open source
license.
Second, Nokia was concerned with potential
costs stemming from losing business secrets, such
5.2. Costs of implementation and as plans for future devices. In order to mitigate
strategies to mitigate these
these costs, Nokia revealed knowledge in terms of
While the benefits of private-collective innovation software but kept future key product innovations
have been spelled out previously, the hidden costs and business figures, such as devices sold, number
of implementing the model have been neglected in of employees, or investments expended, confiden-
previous work or remain unknown. In this sec- tial. For the development of new software fea-
tion, we present the findings from the case study tures, they contracted several small enterprises
along five categories of costs, together with No- or even motivated individual volunteers with
kia’s strategies (where applicable) to mitigate unique knowledge critical to the development
these. First, when software is freely available of the Internet Tablet to sign an NDA. The
even to direct competitors, it is possible for NDA protected Nokia against the leakage of
current and future competitors to design clones information about planned new product develop-
that look like and behave in a manner very similar ments. The agreements effectively created a
to the original product. Competitors are in the three-tiered community of people ‘who knew,’
position not only to imitate but to replicate the which meant Nokia insiders, ‘those with a clue,’
product (see Kogut and Zander, 1992). This which meant contractors, and the ‘regulars’ (C3).
potential lack of differentiation of products repre- However, this information imbalance led to
sents a cost to the firm as it forfeits an opportu- strong tensions for Nokia with its external com-
nity to gain competitive advantage (Granstrand, munity members. Thus, Nokia hired N9 e.g., in
1999). Such reuse of open source software is order to update the technical roadmap of the
completely legal and, in fact, took place as ex- Internet Tablet with all the information necessary
plained in this illustrative example: for software developers.

182 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

Third, in order to facilitate the increasing provements and modifications, combined with a
involvement of volunteers, Nokia needed to carry meritocratic organization of the projects in-
the costs of reducing community entry barriers. volved, gave Nokia enough influence on the
The company invested in the creation of a Soft- direction of software development. Yet, the prac-
ware Development Kit that enabled new volun- tice of contracting developers from incumbent
teers to easily start development for the Internet communities such as GNOME in order to gain
Tablets. It is a common practice for software reputation and control raised concerns from some
manufacturers to provide such an SDK at high Maemo community members as to how Nokia
costs (Jacobson et al., 1999). However, in the case would influence the future of the projects. For
of the Internet Tablet, Nokia offered the devel- example:
opment tools for free. In order to allow volunteers
to adapt their software for upcoming platform Obviously when they are sponsoring a project,
releases, Nokia also offered a development snap- then they are going to have some control over
shot of their work-in-progress (often including the direction and what gets into it. [. . .] I think,
software for yet unannounced features) which they truly want to work with the community
could be used to ensure that an application would and want to keep them happy as well, so it’s all
also run on future releases. Employees were sent a bit of give and take, I suppose. (C1)
to related conferences in order to increase aware-
ness of the platform and answer questions from The situation was different for projects that were
current and future volunteers and contractors. initiated and controlled by Nokia itself. Accord-
Additional staff, such as ‘community representa- ing to C3, Nokia granted only write-access to its
tive’ N9, were hired in order to communicate software repository to its employees, thus retain-
between Nokia internal developers and the ex- ing control of the actual published source code. It
ternal community members. In order to mobilize is worth noting that this strategy created some
more volunteers to join the Maemo community, tensions in the community, with the non-Nokia
Nokia sold 1500 heavily subsidized devices to members complaining that they could not help or
active open source developers. While such direct contribute if they did not know about the future
costs by Nokia cannot be mitigated easily, invest- direction of the software development or about
ments in community building, knowledge diffu- the estimated timeframes of future software re-
sion, and marketing may be lower in the future leases. The Maemo community attempted to
through sharing the effort with other community influence Nokia’s behavior, mostly through pro-
members. Some interviewees explained that, for viding intense feedback on mailing lists. For
example, with increasing popularity, well-inte- instance, in 2000, Nokia released a new version
grated Maemo community members started to of the Maemo platform as a binary download as
support new volunteers who were becoming in- V4 remembers. However, Nokia did not publish
volved in the development process. the source code at the same time but staved off
Fourth, by contributing source code to open the community by arguing that the legal depart-
source projects that were not managed by the ment needed three more weeks to clear the
company, Nokia gave up control of the future code for reasons of intellectual property protec-
development direction of core technologies de- tion. A strong, negative reaction from the com-
ployed in the company’s hardware. According to munity taught Nokia to proceed differently next
N1, the company traded having full control of the time, releasing both binary and source code
technology for participation in joint development, simultaneously. Sanctions on the part of the
thus benefiting from sharing the cost of innova- community were mainly ‘withdrawal of love’
tion with outsiders. For example, GTK was and the ‘threat of forking.2’ When attempting to
originally intended for use on desktop PCs, and balance control and openness, Nokia considered
according to interviewees N1, C1, V2, V4, and the threat of defection of volunteers to other open
N3, it was necessary to adapt GTK to the low- platforms used by emerging competing ‘open
resource environment of the Nokia Internet Ta- devices.’
blets by decreasing memory consumption. In Fifth, because the private-collective model of
order to regain some control of these critical innovation incentives breaks with the traditional
software components, Nokia hired key developers private-investment model that is prevalent in
from the GTK community and contracted small industry, it is reasonable to expect that the
enterprises with in-depth knowledge in this area. implementation of the model in an established
According to N3, the contribution of code im- firm incurs costs of organizational inertia

r 2009 The Authors R&D Management 39, 2, 2009 183


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

(see Sorenson and Stuart, 2000). Because the 6. Discussion and conclusion
Internet Tablets also include software written
by third-party vendors, Nokia needed to ensure In this paper, we identified a research gap in the
that their software revealed to the Maemo com- literature on private-collective innovation (von
munity did not infringe on intellectual property Hippel and von Krogh, 2003, 2006): little is
rights. As mentioned above, interviewees com- known about the implementation by firms of the
mented that Nokia’s internal approval of open private-collective model of innovation incentives.
source software was slow and bureaucratic. In We argued that the implementation of the model
addition, the complex internal processes of a will be associated with benefits, ‘hidden’ costs,
large multinational organization often made it and strategies to mitigate these. In order to
difficult for Nokia’s developers to collaborate examine and extend the model through empirical
with external open source projects. One intervie- work, we used a case study design. Using quanti-
wee commented: tative and qualitative data, we demonstrated that
the development of the Internet Tablet is a case of
The biggest problem is that Nokia is a very big private-collective innovation. Next, we analyzed
company and that Maemo is a very small data from several sources in order to identify the
group. I think it’s like ten or twenty people in benefits and costs incurred in the implementation
total working on it. A lot of software which of private-collective innovation and the strategies
they use on the 770 is developed in some other by Nokia to mitigate these costs.
groups in Nokia. For example, the movie plug- Nokia launched the Internet Tablet as a pri-
ins and MP3 plug-ins. Those are the same I vate-collective innovation project and as a low-
think as they use on their phones in Symbian. cost probe (Brown and Eisenhardt, 1997). At the
So that’s a Symbian decision. You can’t tell the time of product launch, neither a product cate-
Symbian people to use that bug system instead gory nor a market for these devices existed.
of the internal bug system. They would say, Rather than following existing market demand,
‘Why?’ (V1) Nokia targeted technology pioneers to find out
who would use the Internet Tablet and how it
Nokia employees also commented that the inter- would be used in real-life applications (similar to
nal Nokia firewall would not allow connection to what Zander and Zander, 2005, called ‘exploiting
the official developer chat room of the Maemo the inside track’). Nokia opened up the product’s
community, which was located outside Nokia’s software using externally developed open source
network. In order to mitigate the costs of orga- technologies, allowed for and encouraged contri-
nizational inertia, Nokia employees stated that butions by outsiders, and in the process created a
they would work on a case-by-case basis to new market for a product it had envisioned.
remove obstacles. Employees were, for example, When the product proved successful, Nokia
assigned to make sure that entries in the external, moved from targeting technology pioneers to-
publicly accessible bug tracking system would be wards the mainstream market with the subse-
paid attention to. quent release of the Internet Tablet N800 and
Altogether, Nokia managers were conscious of N810.
the trade-offs between revealing knowledge and This study confirmed most incentives to inno-
technology and the benefits from participating in vate identified in previous literature (von Hippel
private-collective innovation. The following state- and von Krogh, 2003, 2006). It remains incon-
ment summarizes well the experiences regarding clusive whether Nokia saved knowledge protec-
the trade-offs between cost and benefits in the tion costs by revealing most of their software.
model: However, the company gained skills and knowl-
edge through collaborating with outside volun-
Some people might say that one of the pro- teers and contractors and thus also acted as a
blems is that you are leaking and giving out system integrator coordinating a loosely coupled
your secrets and so forth, but it’s more like a network of component providers (see also Bru-
trade-off. What is more important to you: to soni et al., 2001). As the main contributor in the
give some of your secrets an internal work-out project, Nokia enjoyed reputation benefits both
or how much help in creating these products as an attractive employer and as an ‘open source-
you get for free. I think, if you calculate, you friendly company’ amongst open source software
are far more on the positive side when you developers who contributed software to the
decide to share. (N1) platform. By creating the vendor-independent

184 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

GNOME embedded platform and inviting com- software for the product, Nokia regained some
petitors to contribute and use the software plat- influence and control. Internal processes some-
form, Nokia facilitated the adoption of the times proved inadequate to enable a transparent
software as a common platform for embedded and open development process, incurring some
devices. Moreover, Nokia built upon existing delays and costs as well as frustration in the
technology and took advantage of users’ previous community. Nokia acknowledged this challenge
contributions to open source software projects. that the interviewees described as a ‘learning
By taking advantage of existing open source process.’
components, Nokia managed to create a complete The extended model of private-collective inno-
operating system with only a handful of devel- vation provides additional insights for research-
opers and was able to integrate ideas and im- ers. First, implementing private-collective
provements from other Maemo community innovation may enhance organizational learning
members. In terms of costs, Nokia’s manufactur- and renewal, in addition to being a form of ‘open
ing could benefit from low-cost software devel- product development’ (Chesbrough, 2003). Dur-
opment and avoided paying the common ing the development of the Internet Tablets,
per-device license fees. Finally, our study found Nokia adapted and learned to work with a com-
that increased flexibility and a faster time- munity of volunteers. An open research question
to-market is a benefit in implementing private- is to what extent such learning can enable firms to
collective innovation. work with outside volunteers across generations
Thus, although the six benefits at first glance and categories of products. Second, Harhoff et al.
make it rational for the firm to choose private- (2003), modeling the payoff for innovators to
collective innovation among alternative models, freely reveal their innovations find that one of
previous work also raised the awareness of unin- four conditions, ‘greater generality of the knowl-
tended consequences or ‘hidden costs’ resulting edge’ reduces the likelihood of free-revealing by
from implementing private-collective innovation. the innovator (see also Muller and Pénin, 2006).
For example, in order to obtain outside contribu- However, this case study showed that much
tions, a firm may need substantial investments in knowledge revealed proved to be generic frame-
documenting the released software, training po- works: it laid the foundation for a generic em-
tential contributors, and developing online tutor- bedded Linux desktop environment. More
ials. These costs of implementation may offset the specific components, such as power management
benefits to private-collective innovation.3 and some end-user applications, were kept pro-
We briefly reviewed literature that indicated prietary. This contradicts Harhoff’s et al. find-
‘hidden costs’ associated with the implementation ings, and the issue of what kind of knowledge is
of novel innovation models (Crawford, 1992; likely to be revealed under what conditions re-
Kessler et al., 2000; Smith, 2004). The study quires more attention in future research.
found that the implementation of the private- Limited by its design, the current study can
collective model of innovation incentives in No- only generalize the findings to theory. The ex-
kia’s development of the Internet Tablet incurred tended private-collective innovation model pro-
costs and that the company found strategies to vides a set of benefits, costs, and mitigation
mitigate these. In particular, the potential lack of strategies that must be tested on a larger sample
product differentiation as well as revealed busi- in future research using cross-sectional as well as
ness secrets incurred costs to the company. Nokia longitudinal designs. Thus, it will be important to
mitigated these costs by selectively revealing garner insights on technological, industry, and
knowledge and technology. Another cost con- market conditions that provide different levels of
cerns the lowering of entry barriers to the Maemo benefits and costs of innovation. For example, in
community. The company invested in several industries of non-virtual goods or where product
measures to reduce such barriers, including dis- development constitutes a minor share of fixed
counted devices and a free SDK, as well as costs in production (e.g., cement manufacturing),
allocating employees responsible for community companies may find it more attractive to pursue
communication in order to attract further volun- private-investment innovation. Moreover, if in-
teers. Moreover, using technologies that are novation is largely based on tacit knowledge
partly maintained externally has the advantage acquired through extensive and costly apprentice-
of shared innovation costs but implies giving up ship (e.g., luxury goods), volunteers who join
full control of the future development of that product development products may be rare.
technology. Through hiring key developers of Future research will also have to investigate the

r 2009 The Authors R&D Management 39, 2, 2009 185


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

impact of company age and size on the innova- Crawford, C. (1992) The hidden costs of accelerated
tion incentives in the private-collective model. As product development. Journal of Product Innovation
we found in the case of an established company, Management, 9, 3, 188–199.
the process represented costs of organizational Dahlander, L. (2004), Appropriating the commons:
inertia. firms in open source software. Technical report,
Managers who want to experiment with flexible Chalmers University of Technology.
Dam, K. (1995) Some economic considerations in the
solutions, while keeping their own product devel-
intellectual property protection of software. Journal
opment costs low, should investigate ways to
of Legal Studies, 24, 2, 321–377.
implement private-collective innovation. Sharing David, P. (1992) Knowledge, property, and the system
development costs and enabling contributions dynamics of technological change. Proceedings of the
from third parties, as well as boosting organiza- World Bank Annual Conference on Development Eco-
tional learning, are powerful reasons why the nomics, 215–247.
model is attractive in practice. However, there David, P. (1998), Knowledge spillovers, technology trans-
are potential ‘hidden costs’ in implementing the fers, and the economic rationale for public support of
model. Learning from Nokia’s successful ap- exploratory research in science. Background Paper for
proach, managers should think ahead about pos- the European Committee for Future Accelerators.
sible costs and create strategies to mitigate them. Downes, L. and Mui, C. (1998) Unleashing the Killer
The experience from the development of the Inter- App: Digital Strategies for Market Dominance.
net Tablet provides possible mitigation strategies. Boston, MA: Harvard Business School Press.
Economides, N. (1996) The economics of networks.
International Journal of Industrial Organization, 14,
6, 673–699.
Acknowledgements Economides, N. and Katsamakas, E. (2006) Two-sided
competition of proprietary vs. open source technol-
The research was supported by the Swiss National ogy platforms and the implications for the software
Foundation (grants 100012-101805 & 105512- industry. Management Science, 52, 7, 1057–1071.
106932). The authors wish to thank all interview Edwards, K. (2003), Epistemic communities, situated
partners for their time and comments, as well as learning and open source software development.
Eric von Hippel and Gideon Markman for pro- Technical report, Technical University of Denmark.
viding us with insightful comments and reviews. Eisenhardt, K.M. (1989) Building theories from case
study research. Academy of Management Review, 14,
4, 532–550.
References Foray, D. (2004) Economics of Knowledge. Cambridge,
MA: MIT Press.
Alavi, M. and Leidner, D.E. (1999) Knowledge man- Gächter, S., von Krogh, G. and Haefliger, S. (2006)
agement systems: issues, challenges, and benefits. Private-collective innovation and the fragility of
Communications of the AIS, 1, 1, 1–37. knowledge sharing. Working paper available at:
Allen, R. (1983) Collective invention. Journal of Eco- http://nottinghamnetlearning.com/economics/cedex/
nomic Behavior and Organization, 4, 1, 1–24. papers12006-24.pdf
Arrow, K.J. (1962) The economic implications of Gambardella, A. and Hall, B.H. (2006) Proprietary
learning by doing. The Review of Economic Studies, versus public domain licensing of software and
29, 3, 155–173. research products. Research Policy, 35, 6, 875–892.
Baldwin, C.Y. and Clark, K.B. (2006) The architecture Grand, S., Georg von Krogh, D.L. and Swap, W.
of participation: does code architecture mitigate free (2004) Resource allocation beyond firm boundaries:
riding in the open source development model? Man- a multi-level model for open source innovation. Long
agement Science, 52, 7, 1116–1127. Range Planning, 37, 591–610.
Brown, S.L. and Eisenhardt, K.M. (1997) The art of Granstrand, O. (1999) The Economics and Management
continuous change: linking complexity theory and of Intellectual Property, Edward Elgar, Cheltham.
time-paced evolution in relentlessly shifting organi- Haefliger, S., von Krogh, G. and Spaeth, S. (2008)
zations. Administrative Science Quarterly, 42, 1–34. Code reuse in open source software. Management
Brusoni, S., Prencipe, A. and Pavitt, K. (2001) Knowl- Science, 54, 1, 180–193.
edge specialization, organizational coupling, and the Harhoff, D., Henkel, J. and von Hippel, E. (2003)
boundaries of the firm: why do firms know more than Profiting from voluntary information spillovers:
they make? Administrative Science Quarterly, 46, 4, how users benefit by freely revealing their innova-
597–621. tions. Research Policy, 32, 1753–1769.
Chesbrough, H. (2003) Open Innovation. The New Heckathorn, D.D. (1997) Respondent-driven sampling:
Imperative for Creating and Profiting from Technol- a new approach to the study of hidden populations.
ogy. Boston, MA: Harvard Business School Press. Social Problems, 44, 174–199.

186 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

Henkel, J. (2006) Selective revealing in open innovation Osterloh, M. and Rota, S. (2004) Trust and Commu-
processes: the case of embedded linux. Research nity in Open Source Software Production. In Lahno,
Policy, 35, 7, 953–969. B. and Matzat, U. (eds.), Trust and Community on the
von Hippel, E. and von Krogh, G. (2003) Open source Internet. Lucius&Lucius, Stuttgart, Germany.
software and the ‘‘private-collective’’ innovation Paul, R. (2007) Intel launches site for open source
model: issues for organization science. Organization mobile Linux development. Ars Technica. Available
Science, 14, 2, 209–223. at: http://arstechnica.com 19 July 2007. Accessed 15
von Hippel, E. and von Krogh, G. (2006) Free reveal- Jan 2008.
ing and the private-collective model for innovation Pettigrew, A. (1990) Longitudinal field research on
incentives. R&D Management, 36, 3, 295–306. change: theory and practice. Organization Science,
ICFAI (2005) Nokia and the Global Phone Industry, 1, 3, 267–292.
ICFAI Center for Management Research, Hydera- PricewaterhouseCoopers (2005) Governance, risk, and
bad, India. compliance series: global best practices, Available
Jacobson, I., Booch, G. and Rumbaugh, J. (1999) The from: http://www.globalbestpractices.com.
Unified Software Development Process, Addison- Roberts, J., Hann, H. and Slaughter, S. (2006) Under-
Wesley, Boston. standing the motivations, participation, and perfor-
Jeppesen, L.B. and Frederiksen, L. (2006) Why do mance of open source software developers: a
users contribute to firm-hosted user communities? longitudinal study of the Apache projects. Manage-
The case of computer-controlled music instruments. ment Science, 52, 7, 984–999.
Organization Science, 17, 1, 45–63. Shah, S. (2006) Motivation, governance, and the
Kessler, E., Bierly, P. and Gopalakrishnan, S. (2000) viability of hybrid forms in open source software
Internal vs. external learning in new product devel- development. Management Science, 52, 7, 1000–1014.
opment: effects on speed, cost, and competitive Shah, S.K. and Corley, K.G. (2006) Building better
advantage. R&D Management, 30, 3, 213–224. theory by bridging the quantitative–qualitative divide.
Kogut, B. and Zander, U. (1992) Knowledge of the Journal of Management Studies, 43, 8, 1821–1835.
firm, combinative capabilities, and the replication of Shankland, S. and Charny, B. (2003), Linux to
power most Motorola phones. Available from:
technology. Organization Science, 3, 383–397.
http:// http://news.com.com/2100-1001-984424.html
Kotha, S. (1995) Mass customization: implementing
accessed Jan 24th, 2007.
the emerging paradigm for competitive advantage.
Siggelkow, N. (2007) Persuasion with case studies.
Strategic Management Journal, 16, 21–42.
Academy of Management Journal, 50, 1, 20–24.
Kuk, G. (2006) Strategic interaction and knowledge
Smith, P. (2004) Accelerated Product Development:
sharing in the KDE developer mailing list. Manage-
Techniques and Traps. New Jersey: Wiley.
ment Science, 52, 7, 1031–1042.
Sorenson, O. and Stuart, T. (2000) Aging, obsolesence,
Lakhani, K., Wolf, B., Bates, J. and DiBona, C. (2002),
and organizational innovation. Administrative Science
The Boston Consulting Group hacker survey. available
Quarterly, 45, 1, 81–112.
from: http://www.bcg.com/opensource/BCGHacker-
Stiglitz, J.E. (2006) Scrooge and intellectual property
SurveyOSCON24July02v073.pdf. Accessed 1 July 2004. rights. British Medical Journal, 333, 7582, 1279–1280.
Lerner, J. and Tirole, J. (2002) Some simple economics Strauss, A.L. and Corbin, J. (1998) Basics of Qualitative
of open source. Journal of Industrial Economics, 50, Research: Techniques and Procedures for Developing
2, 197–234. Grounded Theory. Thousand Oaks, CA: Sage.
Liebeskind, J.P. (1996) Knowledge, strategy, and the Torvalds, L. and Diamond, D. (2001) Just for Fun.
theory of the firm. Strategic Management Journal, London, UK: Texere.
Special Issue: Knowledge and the Firm, 17, 93–107. Yin, R.K. (2003) Case Study Research: Design and
Luthiger Stoll, B. (2006), Spass und software-entwick- Methods. Thousand Oaks, CA: Sage.
lung: zur motivation von open-source-programmier- Zander, I. and Zander, U. (2005) The inside track: on
ern. PhD thesis, University of Zurich. the important (but neglected) role of customers in the
Maemo.org (2006), Maemo website. Available from: resource-based view of strategy and firm growth.
http://maemo.org, accessed December 2006. Journal of Management Studies, 42, 8, 1519–1548.
Martin, R. (2007), Nokia launches ‘Context-Aware’
internet tablet. InformationWeek. Notes
Muller, P. and Pénin, J. (2006) Why do firms disclose
knowledge and how does it matter? Journal of 1. A brief anecdote: in 2006, when one of the co-
Evolutionary Economics, 16, 1–2, 85–108. authors of this paper asked for accessories in such
Nokia (2005), Nokia makes donation to GNOME a shop, none of the shopkeepers were even aware of
Foundation. Nokia press release. the product’s existence.
Nuvolari, A. (2004) Collective invention during the 2. A so-called ‘‘fork’’ results when ‘dissatisfied pro-
British industrial revolution: the case of the Cornish grammers’ copy the original source code from a
pumping engine. Cambridge Journal of Economics, project and continue its development in an alter-
28, 3, 99–119. native, competing project.

r 2009 The Authors R&D Management 39, 2, 2009 187


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

3. In addition, Osterloh and Rota (2004) pointed out certainly good. Do you really want to merge
that the mere presence of a firm in private-collective these things? So, corporations didn’t really
innovation may ‘crowd out’ intrinsic motivation by understand and still not all of them do, that
voluntary contributors (e.g., fun and enjoyment). OSS is not only about software and what kind
of software components you have, but it’s all
Appendix A about the process that you develop this openly
within the communities and you try to syn-
Gaining insight from interview codings chronize your own work with the heartbeat of
the communities. Some companies now under-
As described in the method section, we conducted
stand better than others. We certainly have
interviews with a continuously updated, semi-
done our learning. We have made some mis-
structured questionnaire investigating different
takes too on this front.
aspects of our research question. After transcrip-
tion of the interviews, we categorized the quotes
in similar statements, eventually leading to 80 What kind of mistakes?
different codes and 1,026 text codings. Iterative
regrouping and recoding of the statements even- For example, we did exactly what I described
tually led to three main themes with a total of 12 with our email client. We took a suite for -mail
patterns. Based on these, we created our extended from OSS and took a snapshot of that and
model of the implementation of private-collective continued developing that over a year. We
innovation. didn’t synchronize it with the open source
Since space restrictions prevent us from quoting project. We are now totally in our own branch
too many statements within the article, we explain and the maintenance costs are increasing. We
in the following sections how we extracted a general have two options: we either make a huge effort
finding from analyzing the interview statements. in merging these two branches or then we just
For example, in the case of the subsequently forget what we have done and jump back to
merged category ‘Network Collaboration,’ the where the community is today.
screenshot of the text analysis software Max.QDA
shows that the pattern consisted of three additional Quotes leading to our abstracted finding,
codes: ‘Maintenance/Upstream Project,’ ‘Nokia as written in the text:
collaborating with Third-Party Vendors,’ and
‘How Nokia Manages OSS Development.’ In ‘Interviewees N1 and N3 pointed out that Nokia
sum, 38 codings from the interviews covered state- learned over time that this is indeed the case and
ments on ‘Network Collaboration.’ Triangulating that it was inefficient in the long run to maintain
these statements and interpreting them eventually their own software version while the upstream
led to the case study findings in Section 5. project is continuously improved by a much
larger group of developers.’

Excerpt of interview with N1:


[. . .] Suppose that I am a company X and I Interview guideline for a volunteer member
want to build a Linux-based product. Now I of the Maemo community
go to the community and I take a snapshot of
the Linux operating system today. I think a
About the interviewee and his involvement
copy of the version 2.6.18. Then I go into my
1. What is your education?
lab. I work two years with that, doing all kinds 2. What and for whom do you work at the
of fancy things. And then I again come back to moment?
the community and say I took a snapshot two 3. Since when have you been involved in open
years ago, look what I have done. And the source software (OSS) development and what
community says we don’t care. You should did you do?
have been participating in the community 4. In what way is your paid work related to
work all the time so we could have seen what OSS?
you do to integrate our great things with what 5. Do you have commited access to OSS
you have done. Now we have got two totally projects? If yes, which ones?
incompatible branches. Your branch may be 6. What mailing lists are you subscribed to and
good, but there is our branch and that is follow most of the discussion?

188 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

7. When and how did you get involved in the somebody is employed by Nokia, working for
Nokia Internet Tablet project? Why? one of Nokia’s contractors or if they are a
8. What are your contributions within the voluntary contributor . . .
Maemo project? 1. . . . regarding contributions?
9. How many hours per week do you spend con- 2. . . . regarding technical knowledge?
tributing to Maemo? Has this changed over time? 3. . . . regarding form of communication?
10. Have you received financial or other mate- 4. . . . regarding helpfulness?
rial benefits from Nokia for your participation 5. . . . other particular issues?
in the Maemo community? Knowledge: Guarding Business Secrets
11. What is the main reason for your contribu- 1. Why do external developers demand more
tions within the Maemo community? knowledge on plans of future Maemo develop-
Knowledge Revealing Strategy ments?
1. In your opinion: Why did Nokia not just 2. What kind of information could Nokia
integrate OSS components but create a com- publish in a technical roadmap while guarding
munity portal and actively participate in estab- business secrets?
lished OSS projects? Organization: Network Collaboration
2. How would external participation have 1. Do you as a voluntary contributor feel part
evolved if Nokia had not published their own of the Maemo community? In what way or why
software developments as OSS? not?
3. Would you have participated in the Maemo 2. What are the benefits for Nokia when their
community if Nokia had used a proprietary software developments are integrated into es-
operating system? tablished upstream OSS projects such as Linux
4. From your perception: What type of soft- Kernel, GTK, GStreamer, D-BUS?
ware does Nokia release as OSS and what as 3. In which OSS projects is Nokia’s software
proprietary binaries? code successfully ‘integrated upstream’? Why?
Knowledge: Knowledge Reuse 4. In which OSS projects is Nokia not able to
1. What do you think, why did Nokia not chose integrate their developments? Why?
a ‘commercial’ Linux distribution such as Red- 5. Are there certain examples how Nokia
Hat or SUSE but Debian? learned about the importance of this? If yes,
2. Expert’s estimation (in %): How many lines which ones?
of code of the software in the N800 (out of the 6. How does Nokia gain trust in these projects?
box) is Organization: Reducing Network Entry Barriers
1. unmodified, preexisting OSS (e.g., parts of 1. What is your benefit in participating in the
GTK) development?
2. modified, preexisting OSS by Nokia or 2. How does Nokia encourage external pro-
contractors (e.g., parts of GTK) grammers to voluntarily contribute to the In-
3. newly created OSS by Nokia or contrac- ternet Tablet software?
tors (e.g., Hildon) 3. What should they improve so you’d spend
4. newly created proprietary software by more time programming for Maemo?
Nokia or contractors (e.g., Canola) 4. How relevant are voluntary contributions so
5. unmodified or modified preexisting pro- far for the Maemo platform?
prietary software by Nokia or contractors (e.g., 5. In which communities does Nokia still need
Opera) to improve its acceptance?
3. What tasks are usually required in order Organization: Recruiting benefits
to integrate preexisting OSS for the Internet 1. What is your image of Nokia?
Tablet? 2. Would you accept a job offer by Nokia if you
Knowledge: Distributed Technology Expertise could work at the Maemo project?
1. Why is it not difficult for external program- Organization: Balancing Control
mers to develop software for Maemo? 1. How can Nokia maintain control over
2. Why isn’t more software written for the source code they integrated in incumbent OSS
platform? projects?
3. What kind of questions do Nokia developers 2. What type of control is important and which
mostly ask? one can be neglected?
4. In your opinion, what technical knowledge 3. How does Nokia influence future develop-
do you possess which Nokia employees don’t? ments in established OSS projects?
5. What are the differences in Maemo commu- 4. What made volunteer developers leave the
nity participation (mailing list, IRC . . .) if Maemo community?

r 2009 The Authors R&D Management 39, 2, 2009 189


Journal compilation r 2009 Blackwell Publishing Ltd
Matthias Stuermer, Sebastian Spaeth and Georg von Krogh

5. What are other similar OSS projects volun- What’s your role within the project?
teer developers could be attracted to? What are the goals of the Nokia 770?
6. Is it probable that certain components Is there a difference for N800?
of the Maemo software will get forked by the Choices
community or by other software or hardware Why did you choose Linux instead of Symbian/
vendors? Why? Why not? Windows CE? (What would have been different?)
7. How transparent is the development process Why did you choose Debian instead of e.g.,
of the Maemo platform? Suse or RedHat?
8. Would you like to have more influence in the Why did you choose the GNOME/GTK stack
development process of the Maemo platform? and not KDE?
Organization: Organizational Inertia Why didn’t you contract an Embedded Linux
1. What organizational aspects make it difficult company such as MontaVista to build the
for Nokia to participate in OSS communities? operating system for you? (such as e.g., Motor-
2. How is collaboration with volunteers af- ola)
fected by Nokia’s organizational structure? What were the technical and business reasons
Organization: Organizational Learning to break backwards-compatibility of the OS
1. From your perspective, what have the key 2007 Edition? What are the benefits and risks
learnings been since the release of the Nokia of this decision?
770 Internet Tablet? Contractors
2. What organizational improvements should Why do you contract small firms such as
Nokia make in the future? KernelConcepts? (What are their strengths?)
Competitive Situation: Time-to-market Which external companies did you contract for
1. Did the volunteer community help to speed up the Internet Tablet development?
the development process of the Maemo platform? When do you do things internally? (For exam-
Competitive Situation: Countering Uncertainty ple . . .)
with Flexibility What type of knowledge is transferred to Nokia
1. How flexible is the hardware/software plat- through contractors? (How?)
form to implement completely new use cases for How long would it have taken you to create
the Internet Tablet? Nokia 770 without the help of contractors?
2. How does uncertainty of the future inhibit Why didn’t you do it all by yourself?
innovation by Nokia resp. by users? What was the advantage of choosing OSS?
Competitive Situation: User Contributions What are the challenges contracting small firms?
1. What are key contributions from the volun- Volunteers
tary developer community? How do you motivate volunteer contributions?
2. How well are the users aligned with Nokia’s (For example . . .)
plans for the Internet Tablet? What is the influence of revealing source code
3. Could volunteers develop a killer application in the context of motivation?
for the Tablet? Why? Why not? What kind of contributions are made volunta-
Competitive Situation: Difficulty to Differentiate rily and for what do you have to pay money?
1. In what way does Nokia lose when it reveals (For example . . .)
the source code of their own developments? How business-critical are voluntary contribu-
2. What do you think, for what reasons are tions really for the functioning of the Internet
certain parts (hardware drivers, graphical user Tablet? (Looking at the interviews, it seems as
interface, applications) not reveale if the most difficult, central things are done by
3. Do competing projects (e.g., Openmoko, Nokia itself or by contracted firms. Armin
OLPC etc.) using code from Maemo? What parts? Warda said his contribution concerning swap
Conclusions memory was negligible.)
1. In your opinion: What is the greatest benefit On what occasions do you meet volunteer
and what is the greatest disadvantage of ‘open- contributors physically? (How often?)
ing up’ software and hardware products? What is the difference between physical vs.
virtual collaboration?
Interview guideline for a Nokia manager When do you decide to pay somebody for a
certain development effort? (For example . . .)
Intro On what criteria of a developer do you decide
Since when do you work for Nokia? to hire somebody? (For example . . .)
Since when have you been working for the What’s different when recruiting somebody
Internet tablet project? from the community than somebody from

190 R&D Management 39, 2, 2009 r 2009 The Authors


Journal compilation r 2009 Blackwell Publishing Ltd
Development of the Nokia internet tablet

within Nokia? (e.g., in the case of the Maemo integrate a standard and develop their own
product manager) products faster?)
What changed in terms of employer attractive-
ness when Nokia started OSS development?
Do Nokia employees ask technical questions to Matthias Stuermer studied Business Administration
the volunteer community? (For example . . .) and Computer Science at the University of Berne
Communities and graduated in 2005 with his thesis on open source
Why do you prefer collaborating with the up- community building. He currently pursues his doc-
stream project? (instead of forking your own toral dissertation on firm involvement in open source
version) communities at the Chair of Strategic Management
In what cases is it better to fork an established and Innovation at ETH Zurich, Switzerland.
project? (Why?)
How do you make Nokia accepted within the Sebastian Spaeth graduated as Master of Science
OSS communities? (For example . . .) at Linköping Technical University, Sweden. After
In what kind of tasks is Nokia strong compared
receiving his Doctor in Business Administration
to the strengths of the volunteer community?
from the University of St.Gallen, Switzerland, in
(For example . . .)
2005, he conducts research in the areas of net-
In what kind of tasks is the volunteer commu-
nity strong compared to the strengths of No- work innovation and open business models as a
kia? (for example . . .) researcher at the ETH Zurich, as well as teaching
What are the challenges being a commercial Master courses on Strategic management. He has
company within OSS communities? (for exam- published in Research Policy and Management
ple . . .) Science among other journals.
Competitors
Are there competitors in the mobile device Georg von Krogh is Professor of Strategic Man-
industry pursuing the same OSS strategy? agement and Innovation at the ETH Zurich’s
Do you know of cases where competitors used Department of Management, Technology, and
your revealed source code? Economics. He received his MSc and PhD from
How do you collaborate with competitors? (for the Norwegian University of Technology and
example . . .) Natural Science. He has published articles and
What is the benefit of setting standards such as books on strategic management, knowledge crea-
the GNOME Embedded Platform? (Don’t tion, and innovation, as well as organization and
competitors catch up more easily if they can management theory.

r 2009 The Authors R&D Management 39, 2, 2009 191


Journal compilation r 2009 Blackwell Publishing Ltd