Sie sind auf Seite 1von 22

Journal of R isk R esearch 1 (2), 165–185 (1998)

Imaginable su rprise in glo bal change scie nce


ST E P H E N H . SCH N E ID E R
D epartm ent of B iological Sciences and Institu te for International Stu dies, Stanford University, Stanford,
CA 94305–5020, USA

B . L . T U R N E R II , H O L L Y M O R E H O U SE G A R R IG A
G eorge Perk ins M arsh Institu te and Gradu ate School of G eography, Clark University, W orcester,
M A 01610, USA

A bstract

D ecisionmakers at all scales (individuals, Ž rms, and local, national, and international
gove rnmental organizations) are concerned abou t reducing their vulnerability to (or the
likelihood of) unexpected events, ‘surprises.’ A fter brie y and selectively reviewing the
literature on uncertainty and surprise, we ado pt a deŽ nition of ‘surprise’ that does not
include the strict requirement that it apply to a wholly unexpected outcome, but rather
recognizes that many events are often anticipated by some, even if not most observers.
Thus, we deŽ ne ‘imaginable su rprise’ as events or processes that depart from the expec-
tations of some deŽ nable community. Therefore, what gets labelled as ‘surprise’ depends
on the extent to which what happens departs from community expectations and on the
salience of the problem. We offer a typology of su rprise that distinguishes imaginable
su rprises from risk and uncertainty, and develops several kinds of impediments to over-
coming ignorances. These range from the need for more ‘normal science’ to
phenomenological impediments (e.g., inherent unpredictability in some chaotic systems)
to epistemological ignorance (e.g., ideological blocks to reducing ignorance). Based on
the input of some two dozen scholars at an A spen Global Change Institute Su mmer
Workshop in 1994 *, we construct two tables in which participants offer many possible
‘imaginable surprises’ in the global change context, as well as their potential salience
for creating unexpectedly high or low carbon dioxide emissions. Improving the antici-
patio n of surprises is an interdisciplinary enterprise that should offer a sceptical
welcoming of outlier ideas and methods.

1. Surprise! Ozone hole discovered over A ntarctica in 1985


G iven all the modern satellite technology in the possession of NA SA and other U S
agencies, it is ironic that the opening of a hole in the ozone over the Sou th Pole in the
late 1970s went u ndetected for years. The satellite instru mentation did not fail u s; rather,
the compu ter progra ms written to diagnose the vast volu mes of satellite data were
speciŽ cally instru cted to reject measu rements that diverged sharply from expected
*The au thors wish to thank the members of 1994 A spen G lobal Change Institu te Workshop for two stimu lating
weeks of discu ssions on which mu ch of this article is based. In particu lar , we appreciate the extensive comments,
reference su ggestions and other assistance we received from J.X. Kasperson, and the help in modifying Fig. 1 from
D avid Victor.

1366-9877 © 1998 E & FN Spon


166 Schneider et al.

normal conditions. For example, every time a very high or low valu e came in, the
compu ter progra ms omitted it. The rejected valu es were called to no one’s attention.
Incredibly, the phenomenon overhead went u ndetected by ou r high- technology for
nearly a decade. R ather, incredu lou s British scientists, plotting by hand their own
grou nd- based records of how mu ch u ltraviolet (U V) radiation was reaching the earth’s
su rface at their station on the coast of A ntarctica (Farman et al., 1985) , detected a
steady decrease in the springtime ozone in the sou thern hemisphere from the mid-1970s
to the mid-1980s. This u nexpected phenomenon immediately triggered a reprogram-
ming of the U S data to allow all valu es, and there in beau tifu l living (false) colou rs for
all to see (ideal for television) were maps showing a deep hole in the ozone over the
A ntarctic continent growin g in intensity over time and drifting over nearby oceans and
continents. This example shows that, sometimes, the knowable remains u ndetected
because of the assu mptions or views that frame the qu estion or methods of analysis.

2. Not so surprising: climatic surprises are imaginable


In the wake of the heat waves, Ž res, and drou ghts of 1988, media coverage of global
warming exploded. Not su rprisingly, so did congressional hearings on climate. R obert
Watson, then NA SA scientist and later head of the U S delegation to the plenary session
of the 1995 Intergove rnmental Panel on Climate Change (IPCC) testiŽ ed to the Senate
E nergy Committee 1 that ‘scientiŽ c u ncertainty is a thing that scares me more than
absolu te knowledge. The fact that we have said that climate cou ld possibly change
dramatically and qu ickly has me concerned. The A ntarctic ozone thing came as a
su rprise.’ To that remark, New Jersey Senator Bill Bradley pressed the panel of scien-
tist witnesses to specu late ou t-lou d on what kinds of su rprises they had in mind. O ne
of u s (SH S), after qu ipping that su rprises are things you don’t know abou t, nonethe-
less followed with a list of highly u ncertain, bu t plausible ou tcomes, like a  ip- op in
North A tlantic ocean cu rrents, a su rge in polar ice masses, or u nexp ected regional
patterns of climate response to bu ild-u ps of greenhou se gases. O ther witnesses also had
their hu nches as to a nu mber of su ch possible ‘su rprises’. The willingness of the witnesses
to specu late on speciŽ c ‘su rprises’ illustrates that scientists, as well as members of the
pu blic, deal often with the anticipation of the ‘u nexpected’!
Strictly speaking, a su rprise cannot be anticipated; by deŽ nition it is an u nexpected
event. Potential climate change, and more broadly global environm ental change, is
replete with this kind of su rprise – the tru ly u nexpected – because of the enormou s
complexities of processes and interrelationships involved and ou r insu fŽ cient u nder-
standing of them (su ch as cou pling ocean, atmosphere, and terrestrial systems) (e.g.,
D armstadter and Toman, 1993a ; Broecker, 1994; Casti, 1994; Chapter 1 and Su mmary
for Policymakers, IPCC, 1996) as well as ou r improved u nderstanding of the existence
of chaos in su ch complexity (e.g., Cohen and Stewart, 1994; R obinson, 1982). The IPCC
(1996) Su mmary for Policymakers conclu des:
‘Fu tu re u nexpected, large and rapid climate system changes (as have occurred in the past)
are, by their nature difŽ cult to predict. This implies that fu tu re climate changes may also
involve “surprises.” In particular these arise from the non- linear natu re of the climate
system. When rapidly forced, non- linear systems are especially su bject to unexpected
behavior.’

1 See chapter 2 of Schneider, 1990, for details on this hearing.


G lobal change science 167

Yet, by focu sing on the perceptions of and responses to environm ental events by the
pu blic and expert commu nities alike, as indicated in the Senate E nergy Committee
exchange noted, the existence of a different kind of ‘su rprise’ emerges, one that can
be anticipated (Kates and Clark, 1996) . R isk-hazard and related research demonstrates
repeatedly that the event, process, or ou tcome registered as a su rprise by the commu-
nity in qu estion was frequ ently known or forecast by others or the same event was
knowable within the competing frameworks of u nderstanding (D armstadter and Toman,
1993b) . This message also emerged from the 1994 A spen G lobal Change Institute
(A G CI) session on ‘A nticipating G lobal Change Su rprise,’ (A ppendix 1) the resu lts of
which we, the co-chairs (Schneider and Tu rner, 1995) of that session, report here. We
sketch the rationale u sed to reach the conclu sion that the search for anticipated su rprises
constitutes a fru itful avenu e of research and then present the resu lts of exercises to
identify what some of these su rprises might be, how they cou ld affect the impacts of
climate and other global changes, and how we might enlarge or focu s ou r vision of this
kind of su rprise.

3. From uncertainty to surprise: a brief review


Su rprise is an attribu te of events or interpretations of them that is closely akin to the
related attribu tes of risk and u ncertainty. Variou s meanings of these attribu tes may be
fu sed at times, owing to the different commu nities and cu ltures employing them. The
interdisciplinary risk-hazard research commu nity, however, has devoted considerable
attention to this and related concepts, a brief review of which informs this work.
Much of the cu rrent work on su rprise has grown ou t of an extensive body of research
on u ncertainty.2 Yet, althou gh widely acknowledged and studied, u ncertainty remains
a difŽ cu lt concept to deŽ ne or codify. D ifferent conceptu alizations and approaches to
u ncertainty abou nd in the literature, crossing nu merou s Ž elds of study and tou ching a
wide range of prob lem types. Two basic options are invariably followed in the face of
u ncertainty, however. The Ž rst is to redu ce the u ncertainties throu gh data collection,
research, modelling, simulation techniqu es, and so forth. Following this option, the
objective is to overcome u ncertainty, to make the u nknown known. But the dau nting
nature of u ncertainties su rrou nding global environmental change, as well as the need
to make decisions before the ‘normal’ science option can provide resolu tion, force a
second option – that of managing or integrating u ncertainty directly into the decision-
making or policy-making process. Before u ncertainty can be so integrated, however,
the nature and extent of the u ncertainty mu st be clariŽ ed. This u nderstanding is
approached in several ways, which we brie y review here.
The Ž elds of mathematics, statistics, and more recently physics, provide the ‘science
of u ncertainty’ with many powerfu l means and techniqu es to conceptu alize, qu antify,
and manage u ncertainty, ranging from the frequ ency distribu tions of probability theory,
to the possibility and belief statements of Bayesian statistics, and even to a method for
qu antifying ignorance (between belief and disbelief) fou nd in D empster-Shafer theory
(for examples see Tonn, 1991; A yyu b et al., 1992; Yager, 1992). A ddressing other aspects
of u ncertainty, fu zzy set logic offers an alternative to classical set theory for situa-
tions where the deŽ nitions of set membership are vagu e, ambigu ou s, or non- exclu sive
2
For early discu ssions on this su bject, the reader is directed to the literatu re cited, as well as the following: Clar k
and Mu nn, 1986; Kasperson et al., 1988; Kates, 1985; Svedin and A niansson, 1987; Timmermann, 1986; and Toth
et al., 1989.
168 Schneider et al.

(e.g., Z adeh, 1965, 1990). More recently, researchers have proposed chaos theory and
complexiŽ cation theory to focu s on expecting the u nexpected in models and theory
(e.g., Casti, 1994).
The practical application of many of these techniqu es was originally pioneered by
researchers in decision analysis (see R aiffa, 1968) . In the Ž elds of economics and
decision theory, researchers continu e to study rational decision making u nder u ncer-
tainty and how to assess the valu e of collecting additional information (e.g., Clemen,
1991). Methods for modelling risk attitudes, leading to the terms risk-prone and risk-
adverse, attempt to capture how different people faced with making a decision react
to the u ncertainty su rrou nding the expected ou tcomes. Looking towards fu ture
ou tcomes in projecting climate change impacts and estimating marginal costs and bene-
Ž ts of mitigation effor ts, economists su ch as Nordhau s (1991, 1993) and Yohe (1991,
1993) explicitly address u ncertainties, and discu ss methods for better measu ring nonlin-
earities leading to su rprises. In the context of energy, tables of possible, bu t u ncertain,
risks of alternative energy systems have been prepared (e.g., Schneider, 1979).
U ncertainty and, in its related context, su rprise, are treated largely as the realization
that events, cu rrently u nknown, will occu r affecting the Ž nal ou tcome of a decision.
This acknowledgment of u ncertainty has fou nd a prominent place in many other Ž elds
of study, each one speaking its own langu age of u ncertainty. For example, researchers
making risk assessments and setting safety standards Ž nd it most u sefu l to distingu ish
between risk (the probability of a certain negative effect resu lting from a hazard occu r-
rence, given the speciŽ ed level of exposu re), variability (inter-individu al differences in
vu lnerability and su sceptibility), and u ncertainty (model parameter variability and any
u nexplained residu al) (e.g., Bogen and Spear, 1987). This three-pronged distinction
provides information on u ncertainty tailored to the needs and research qu estions of the
risk analyst. R esearchers in other Ž elds of study, for instance compu ter science, may
be most interested in addressing other aspects of u ncertainty, su ch as how it affects the
decision  ow or logic of the compu ter system. Thu s, in writing for other compu ter scien-
tists, Bonissone and Tong (1985) su ggest a conceptu alization that identiŽ es fou r sou rces
of u ncertainty in an expert system: (i) inaccu racy in the set of facts making u p the
knowledge base; (ii) imprecision in the decision ru le representation langu age; (iii)
incomplete information; and (iv) aggregation of ru les from different knowledge sou rces.
The above examples are not meant to imply that separate u ncertainty analyses follow
strict disciplinary bou nds. While the last two examples offer approaches to u ncertainty
targeted speciŽ cally to their respective disciplines, some of the concepts are clearly
transferable to other disciplines and to other types of decision making processes,
including those concerning global environmental change. Consequ ently, research on
u ncertainty cross-cu ts a nu mber of different disciplines. In work related to hazards and
risk, sociologists, anthropologists, psychologists, and geograph ers have made important
contribu tions to the discu ssions on risk perception, risk commu nication, and the social
ampliŽ cation of risk (D ou glas and Wildavsky, 1982; Fischhoff et al., 1982; Kahneman
et al., 1982; Kasperson et al., 1988). Similarly, work on visu alizing or graphically
conveying u ncertainty also crosses a diverse set of disciplines including psychology,
compu ter science, and geographic information systems (e.g., ButtenŽ eld and Beard,
1991; MacE achren, 1992).
Du e to its pervasive nature, a desire to u nderstand u ncertainty encou rages researchers
to join forces in interdisciplinary projects, conferences, and worksho ps drawing on the
wider breadth of knowledge, techniqu es, and experiences. R ecent interdisciplinary
G lobal change science 169

efforts in the area of global environmental change, including the Intergove rnmental
Panel on Climate Change A ssessment, IPCC, (1996) and the R esou rces for the Fu ture
work shop (D armstadter and Toman, 1993a ,b), have made u ncertainty an explicit and
central research concern, as has a recent A spen G lobal Change Institute Workshop
(Moss and Schneider, 1997). G iven the interdisciplinary natu re of the study of global
environmental change, it is not difŽ cu lt to see how research in this Ž eld has drawn on
a wide range of techniqu es.
These activities and accomplishments notwithstanding, the answer to the qu estion of
how to deŽ ne or codify u ncertainty remains u nresolved. Commu nities of researchers
tend to develop deŽ nitions and taxonom ies of u ncertainty speciŽ cally related to the
types and range of policy qu estions that they address in their research. A lthou gh no
all-encompassing typology of u ncertainty exists, the stru ggle to deŽ ne the u nknown
aspects of a Ž eld of science serves a very real pu rpose. The typologies make explicit
what is not known, and break it down into distinct components. D ecomposing u ncer-
tainty into different dimensions provides insight into the natu re of the u ncertainty. The
dimensions also offer some stru cture on how to approach or redu ce the u ncertainties,
make the most of available information, and u tilize both what is known and what is
u nknown in decision making.
O ne of the most extensive and comprehensive conceptu alizations of u ncertainty is
that developed by Smithson (1988). Smithson’s taxon omy is based on the sociology of
knowledge, in which he moves the applicability of the social constru ctivists’ dictum that
knowledge is socially constru cted and negotiated to ignorance; it too is a social constru c-
tion. Thu s, u ncertainty is deŽ ned by Smithson as an incompleteness of knowledge or
information that is caused by vagu eness, probability, and ambigu ity. To provide some
perspective on how this deŽ nition relates to others in his taxonom y, consider a few
more of his categorizations: ignorance is encou ntered when a person fails to agree or
show awareness of ideas that another person deŽ nes as actually or potentially valid;
error refers to being in an erroneou s cognitive state; taboo is socially enforced irrele-
vance; and irrelevance is the act of ignoring something.
O ther typologies of u ncertainty exist, each making their own contribu tions to the
wider u nderstanding of u ncertainty (Brooks, 1986; Fu ntowicz and R avetz, 1990; Faber
et al., 1992; Wynne, 1992; Casti, 1994; R owe, 1994; Moser, 1997) . H ere, we will limit
the discu ssion to three of these – Fu nctowicz and R avetz (1990), Wynne (1992), and
R owe (1994). Selected because they are widely cited and recognized in the global envi-
ronmental change literature: they provide interesting insights into the role of u ncertainty
in science for policy making, and they provide the most approp riate context for the
concept of u ncertainty in ou r u sage below.
Fu nctowicz and R avetz rank situations on an incremental scale from those possessing
low to high levels of u ncertainty. They show how different management approaches
correspond with the level of system u ncertainty matched with varying degrees of deci-
sion stakes. Where decision stakes and the level of systems u ncertainty are low
constitutes, the realm of applied science and the type of u ncertainty encou ntered is
technical. A s the decision stakes and the level of u ncertainty increase, the realm of
professional consu ltancy and methodological u ncertainty exists. Fu nctowicz and R avetz
u se the term ‘professional consu ltancy’ to describe a ‘learned art’ involving personal
ju dgment that depends on higher-level, interpretive skills. Science in this realm requ ires
the scientist to make qu alitative assessments or ju dgments, setting conŽ dence limits
(form al or informal), arou nd methods and estimates. When both the decision stakes
170 Schneider et al.

and the level of systems u ncertainty are high, epistemological u ncertainty is encou n-
tered in the realm of ‘second order’ science. Fu nctowicz and R avetz u se the term ‘second
order’ science to deŽ ne an area ou tside the traditional qu antitative methods of applied
science. This term takes into account the postmodernist argu ment that all knowledge
is socially constru cted and that science is only one way of produ cing knowledge. When
they extend u ncertainty into this realm, Fu nctowicz and R avetz are acknowledging that
the u ncertainties identiŽ ed by science are in part deŽ ned by society as a whole. There
may be ignorance abou t ignorance. Fu nctowicz and R avetz call for ‘post-normal’
methods of evalu ation to address these issu es, althou gh other terms have also been
su ggested (Weinberg (1972) u ses the term ‘trans-scientiŽ c’ and O ’R iordan and Cameron
(1994) propose ‘civic-scientiŽ c’).
Wynn e (1992) emphasizes that the modelling of environmental risk systems requ ires
examination of not only the scientiŽ c evidence and competing interpretations, bu t also
investigation of the nature, assu mptions, and inherent limitations of the scientiŽ c knowl-
edge behind the data and the model. H e identiŽ es fou r types of u ncertainty – risk,
u ncertainty, ignorance, and indeterminacy – each overlaying dimensions of u ncertainty.
R isk refers to a situation when the system behaviou r is well known and the chances of
different ou tcomes can be qu antiŽ ed by probability distribu tions. If, however, the impor-
tant system parameters are known bu t not the associated probabilities, u ncertainty
exists. Ignora nce is that which is not known (or even awareness that we do not know
it) and, for Wynne, is endemic because scientiŽ c knowledge mu st set the bou nds of
u ncertainty in order to fu nction. Indeterminacy captu res the u nbou nded complexity of
causal chains and open networks. U ncertainty, in part, stems not only from an incom-
plete u nderstanding of determinate relationships, bu t from the interaction of these
relationships with contingent and u npredictable actors and processes.
R owe (1994) deŽ nes u ncertainty as ‘the absence of information, information that
may or may not be obtainable’ (p. 743), and identiŽ es fou r dimensions of u ncertainty
predicated on their sou rce inherent in any decision making process: temporal (in both
past and fu tu re states); structural (du e to complexity); metrical (in measu rement); and
translational (derived in explaining u ncertain resu lts). The fou rth dimension, transla-
tional u ncertainty, comes into play only after the Ž rst three have been considered. These
dimensions can exist simu ltaneou sly in any situation, althou gh one or more may tend
to dominate. R owe pays special attention to the issu e of variability and identiŽ es it as
a contribu tor to u ncertainty in all dimensions. R owe (1994) does not offer a distinct
deŽ nition of variability. H owever, he does identify three primary sou rces of variability:
(1) U nderlying Variants – variants inherent in natural systems contribu ting to the spread
of parameter valu es, including randomn ess, inconsistent hu man behaviou r, and chaotic
or nonlinear dynamic systems behaviou r; (2) Collective/Individu al Membership
A ssignment – the distinction between collective behaviou r and a single instance of
behaviou r for a parameter; and, (3) Valu e D iversity – varying perspectives and valu e
systems among people.
A lthou gh different in their characterizations of u ncertainty, all three typologies
attempt to expand the conceptu alization to recognize that u ncertainty is not pu rely of
a technical or physical or biological character, bu t also social, cu ltural, and institutional
in nature. By drawing ou t dimensions of u ncertainty, su ch as epistemological u ncer-
tainty, ignorance, indeterminacy and translational u ncertainty, these conceptu alizations
offer insights into the role of u ncertainty and science in global environmental change
(Wynne 1980, 1987, 1992; Morgan et al., 1990; O ’R iordan and R ayner, 1991; D overs
G lobal change science 171

and H andmer, 1995) . In the areas of environmental policy and resou rce management,
policy makers stru ggle with the need to make decisions u tilizing vagu e and ambigu ou s
concepts (su ch as su stainability), with sparse and imprecise information, in decisions
that have far-reaching, and often irreversible, impacts on both environm ent and society.
Not su rprisingly, effo rts to incorporate u ncertainty into the decision-making process
qu ickly move to the forefront with the advent of decision-making paradigms, su ch as
the precautionary principle, adaptive environ mental management, the preventative
paradigm or stewardship (e.g., Brown, 1997). A s noted by Wyn ne, the shift towards
prevention in environmental policy ‘implies an acceptance of the inherent limitations
of the anticipatory knowledge on which decisions abou t environmental discharges [and
other environmental problems] are based’ (1992, p. 111).
R avetz (1986) takes the concept of ‘u sable knowledge in the context of incomplete
science’ one step fu rther by introdu cing the idea of u sable ignorance. To R avetz,
acknowledging the ‘ignorance factor’ means becoming aware of the limits of ou r knowl-
edge – an idea we pu rsu e below. R avetz argu es that ignorance cannot be overcome
with any amou nt of sophisticated calcu lations. R ather, coping with ignorance demands
a better articulation of the policy process and a greater awareness of how that process
operates. H e recognizes that one can only replace ignorance by gaining more knowl-
edge, bu t stresses that by ‘being aware of ou r ignorance we do not encou nter disastrou s
pitfalls in ou r su pposedly secu re knowledge or su pposedly effective techniqu e’ (p. 429).
Following this lead, D overs and H andmer (1995) offer a step-b y-step framework for
what they term ‘ignorance au diting.’ In this framework, the policy maker is asked to
explicitly deŽ ne the types, causes, and sou rces of u ncertainty – and more speciŽ cally
ignorance – affecting the decision at hand. The policy maker then identiŽ es methods
to address these u ncertainties where possible and consciou sly implements them within
the policy process or management task.
The emphasis on managing u ncertainty rather than mastering it can be traced to
work on resilience in ecology (most notably by H olling, 1973, 1986) . Whereas resis-
tance implies an ability to withstand change or impact within some measu re of
performance, resilience captu res the ability to give with the forcing fu nction, withou t
disru pting the overall health of the system. In this framework, adaptation is an ecolog-
ical mechanism whose aim is not to overcome or control environmental u ncertainty,
bu t to live with, and in some case, thrive u pon, it. It is interesting to note that H olling’s
classical work did not focu s on u ncertainty per se bu t on su rprise.

4. D eŽ ning surprise
The A G CI session on ‘A nticipating G lobal Change Su rprise’ drew u pon many experts
(A ppendix 1) in global environm ental change and from the natural hazards commu nity,
most of whom were largely u nfamiliar with the extensive literature noted above.
Nevertheless, drawing u pon a portion of it, the resu lting deŽ nitions adopted in the session
strongly correspond to principal ideas embedded within that literature. The following is
a description of the workshop’s distinctions among the terms as they relate to su rprise:
(i) risk – the condition in which the event, process or ou tcome, and the proba-
bility that each will occu r, is known.
In reality, of cou rse (coins and u nloaded dice notwithstanding) complete or perfect
knowledge of complex systems rarely exists, which wou ld permit the credible calcu lation of
172 Schneider et al.

objective probabilities. Likewise, the fu ll range of potential ou tcomes is u su ally not known.
Thu s, risk almost always is accompanied by varying degrees of uncertainty.
(ii) u ncertainty – The condition in which the event, process, or ou tcome is known
(factually or hypothetically), bu t the probabilities that it will occu r are not
known, or are highly su bjective estimates.
Typically, when prob abilities are assigned they are su bjective (or depend u pon su bjec-
tive assu mptions – Schneider, 1994), and the ways to establish the reliability of different
su bjective probability estimates are debatable (e.g., Morgan et al., 1990; Morgan and
Keith, 1995; another A G CI worksh op, this time on methods to deal explicitly and
formally with u ncertainty in international assessments, contains fu rther discu ssion and
references – H assol and Katzenberger, 1997).
(iii) su rprise – The condition in which the event, process or ou tcome is not known
or exp ected.
In this ‘strict’ meaning, the attribu tion of su rprise shifts toward the event, process,
or ou tcome itself – is it a new or wholly u nexpected experience or not? We may expect
su rprises to occu r, bu t we are su rprised by the speciŽ c event, process, or ou tcome
involved. This meaning, as noted, begs the issu e of anticipation because the very act
of anticipation implies some level of knowledge or foresight. It is, therefore, not partic-
u larly interesting or u sefu l for policy pu rposes. O ne exception is those cases where the
conditions that might indu ce su rprises – for instance, rapid forcing of non- linear systems
as qu oted earlier from the IPCC 1996 Su mmary for Policymakers – are known, even
thou gh the actual su rprise events are not. This exception, which we cou ld term ‘imag-
inable conditions for su rprise’, cou ld have policy meaning, since actions cou ld be
propose d to mitigate the conditions in which su rprise might be indu ced (e.g., slowing
down the rate of global change forcing, as in Chapter 6 of Schneider, 1997) .
Because of the impracticality of the strict deŽ nition of su rprise for policy making,
variou s studies advocate the u se of another meaning for su rprise, one in which the attri-
bu tion of ‘su rprise’ shifts more towards the expectations of the observer. H olling (1986:
294) recognized this meaning of su rprise as a condition in which perceived reality
departs qu alitatively from expectations. It is this more interpretive or relational meaning
of su rprise – one we label ‘imaginable su rprise’ – that portends to be most u sefu l for
global change studies.
(iv) imaginable su rprise – The event, process, or ou tcome departs from the expec-
tations of the observing commu nity or those affected by the event or process.
‘Seen from this point of view, su rprise abou t one or another aspect of climate
change is an after-the-fact reaction to an observation or new scientiŽ c Ž nding
that, in some sense, lies ou tside ou r range of expectations’ (D armstadter and
Toman, 1993b: 3).
A lmost every event may constitute an imaginable su rprise to someone. But since
global change phenomena and their environmental and societal impacts are a commu-
nity-scale set of issu es, little can be gained for ou r pu rposes by focu sing on whether
someone, somewhere, may or may not have once predicted or hinted at some su rprise
event. More fru itfu l is the recognition that grou ps, commu nities, and cu ltures may share
expectations su ch that a particular event is likely to qu alify as a su rprise for most within
them. In these cases, what gets labelled as a su rprise depends u pon the extent to which
G lobal change science 173

reality departs from commu nity expectations, and on the salience of the problems
imposed.
Imaginable su rprise applies to commu nities of experts, policy makers, managers, and
edu cators who share common ranges of expectation that are generated by grou p
dynamics, leaders and signal processors, including the dominant edu cational and
research paradigms (Kasperson et al., 1988). For these commu nities, shared expecta-
tions follow from dominant interpretations among the expert commu nity (e.g., global
warming is likely, NR C, 1992 or IPCC, 1996) , from their Ž t with broader policy agendas
(e.g., environmentally benign economic development is possible), and from vested
interest, consciou s or u nconsciou s, of an agency or grou p to maintain a particular view
(e.g., global popu lation growth is environmentally damaging, or, alternatively, good for
the economy – e.g., Myers and Simon, 1995). Since policy making often re ects a blend
of pu blic and interest grou p perceptions of reality, the imaginable su rprise formu lation
is mu ch more relevant to global change policy issu es than a strict deŽ nition of su rprise
as an u nimaginable ou tcome.

5. A typology of surprise
The distinction between ‘strict’ su rprise and ‘imaginable’ su rprise is important bu t not
su fŽ ciently developed to be very u sefu l for dealing with su rprise phenomena. More
exhau stive treatments of the su b-categories of su rprise or the sou rces of su rprise reveal
that there are signiŽ cant nu ances with important implications for global change stu dies
or policies (Kates and Clark, 1996). Ju st as with u ncertainty, different deŽ nitions and
typologies of su rprise have been propose d. Brooks (1986), for example, offers a simple
tripartite typology: (i) u nexpected discrete events (su ch as the oil shocks of 1973 and
1979 or the Three Mile Island reactor incident); (ii) discontinu ities in long-term trends
(su ch as the acceleration of U SA oil imports between 1966 and 1973); and (iii) the
su dden emergence into political consciou sness of new information (su ch as the relation
between  u orocarbon produ ction and stratospheric ozone). This typology focu ses on
the qu alitative nature of the disju ncture between the event as it occu rs and the event
as it was anticipated. Timmerman (1986), in contrast, focu ses on the increasing inten-
sity of the effects of su rprises, distingu ishing fou r types of su rprises: anomalies, shocks,
epiphanies, and catastrophes. In this typ ology, su rprises are rated on a scale from anom-
alies whose effects are barely noticed to catastrophes that cause u nrecoverable damage.
A typology by Faber and colleagu es (1992) is particularly u sefu l for the ‘imaginable’
meaning because it highlights the relationships between the sou rce of su rprise and the
observer, in particular identifying different sou rces of ignorance that in u ence what is
registered as a su rprise by the observer. Since ignorance, or its opposite, u nderstanding,
in large part shapes expectations, the A G CI session, post-session review, and ou r own
su bsequ ent work has led u s to develop a variant of the Faber-Manstetten-Proops typo-
logical map (Figu re 1) that emphasizes the sou rce of the expectations themselves and
the impediments to changing them.
The Faber-Manstetten-Proops version recognizes the distinctions made above
between risk and u ncertainty, and imaginable su rprise. Because the two former involve
events, processes, and ou tcomes that are known (expected) by most in the commu nity,
even if the probabilities are not fu lly u nderstood, we may label imaginable su rprise
associated with risk and u ncertainty as a ‘qu asi-su rprise.’ We do not address these cate-
gories, althou gh we su ggest that the impediments to u nderstanding imaginable su rprises
174 Schneider et al.

SOURCES OF EXPECTATION

OUTCOMES ALL KNOWN OUTCOMES NOT ALL KNOWN

Risk Uncertanty imaginable surprise


(probabilities (probabilities
known) not known)
(Quasi-Surprise)

Education OPEN EXPECTATIONS CLOSED EXPECTATIONS


(willingness and ability to recognize (unwillingness to recognize
that some outcomes are not known) that some outcomes are not known)

"EASY"-TO-ENLARGE EXPECTATIONS "HARD"-TO-ENLARGE EXPECTATIONS

Personal Communal Epistemological Phenomenological


Impediments Impediments Impediments Impediments
(due to level of (due to level of (due to the ways (due to inadequacy or
individual learning) "normal" science communities organize fundamental limits in
directed toward or view the world) existing technology and
the issue) the full range of known
analytical perspectives)
Research

Fig. 1. Sou rces of differences in imaginable su rprise: typological map.

that we articulate below also have some applicability for assessing qu asi-su rprise – i.e.,
risk and u ncertainty.
Imaginable su rprises occu r among events, processes, and ou tcomes that are not all
known: the other path from sou rces of expectation on ou r typological map. This path
divides fu rther on the ability of the observer to change expectations, either ‘closed’ or
‘open’ in kind. Closed expectations are simplest, as they involve denial of the possi-
bility of u nexpected ou tcomes and thu s an u nwillingness (entrenched ignorance) even
to consider the possibility of a larger range of events or u nderstanding. This sou rce of
su rprise cannot be changed u nless cognitive alternatives are entertained by the observer
and, by deŽ nition, to entertain su ch alternatives engages avenu es of open expectations,
the more complex and interesting path in ou r typological map.
O pen expectations – or the recognition or acknowledgment that all events, processes,
and ou tcomes are not known – is fu rther su bdivided into ‘easy-’ and ‘hard-to-enlarge’
expectations. The ‘easy’ su b-categor ies involve two impediments: personal (those of the
individu al), which can be enlarged throu gh learning; and commu nal (those of grou ps,
commu nities and cu ltures), which are enlarged throu gh the application of more ‘normal’
science, i.e., research. The research process is that of open experimentation and critiqu e,
often leading to improved u nderstanding, and hence expectations. Personal learning or
G lobal change science 175

‘edu cation’ can lead an observer (or research can edu cate a commu nity) back to the
‘qu asi-su rprise’ category of risk and u ncertainty, shown by the dashed line in the Ž gu re.
A s we note below, however, nu merou s barriers impede this process of enlarging expec-
tations by learning and research, and these barriers need to be u nderstood explicitly
and taken into account in any attempt to redu ce su rprise.
The ‘hard’ su b-categories include two major classes of impediments to improved
expectations. Phenomenological impediments involve the su ite of factors that act as
barriers to knowledge-bu ilding, at least in the short ru n, su ch as inadequ ate technology
or modes of analysis. This su b-categor y shou ld not be confu sed with closed expecta-
tions or open expectations of the commu nal kind. Closed expectations rest within the
rigid, u nchanging belief system of the observing commu nity, whereas commu nal imped-
iments can be overcome in principle by providing more attention to the problem set
in qu estion by applications of existing ways of knowing. Impediments of the phenom-
enological kind, in contrast, rest in the sometimes momentary inadequ acies of the tools
and skills of the commu nity. Non- momentary, fu ndamental limits may also exist,
su ch as the inability to forecast accurately weather details beyond a few weeks.
U npredictability, owing to the chaotic nature of large-scale atmospheric dynamics, is
an example (e.g., Lorenz, 1969) . A lthou gh so-called normal science may well lead to
breakthrou ghs that will one day permit credible forecasts past cu rrently believed
predictability limits, it remains possible (even likely) that no amou nt of commu nal
research effort can lead to a breakthrou gh that wou ld breach this apparent phenome-
nological impediment.
Finally, epistemological impediments follow from the ways in which the variou s
observing commu nities view the world and bu ild their u nderstanding within it (e.g., the
compu ter progra mming that missed the ozone hole). If alternative epistemologies will
not be explored in principle or because of social conventions, closed exp ectations will
exist, even if u nconsciou sly. For this set of ‘open’ su b-categories, however, the episte-
mological impediment follows largely from an u ncritical belief in a favou red view,
perspective, or paradigm: the kind of view inherent in the progra mming of data analysis
in ou r ozone story. Science now well knows that it operates by way of prevailing
paradigms (ou r epistemologies) that gu ide u nderstanding, and hence expectations, for
particular commu nities. A paradigm may dominate for a time, providing conditions
temporarily not dissimilar to closed expectations. Science commu nities are u su ally su fŽ -
ciently diverse, however, to maintain competing paradigms that constantly challenge
the prevailing perspective. When these challenges are su ccessfu l (i.e., provide better
u nderstanding), the commu nity eventu ally adopts the alternatives. It is important to
recognize that contemporary behaviou ral and social sciences – an important compo-
nent of the study of imaginable su rprise in climate change – has been in a prolonged
state of competing paradigms that constantly challenge one another (Gu ba, 1990).
Ou r ru dimentary typology (Figu re 1) is u sefu l for at least two reasons. First, it places
su rprise in relation to observing commu nities, providing a logic for the ‘imaginable’ kind.
A nd second, the categor ies in the typolo gy provide clues abou t how to deal with su rprises
associated with the sou rce of expectation, a su bject to which we will retu rn below.

6. Some glo bal climate change surprises derived from an expert community
The 1994 A spen G lobal Change Institu te session on ‘A nticipating G lobal Change
Su rprise’ involved experts from the variou s ‘dimensions’ of global change research:
176 Schneider et al.
Table 1. Candidate global-change ‘su rprises’ for the climate change example [Most of the entries
re ect discu ssions at the 1994 A G CI Su mmer Session on A nticipating G lobal Change Su rprises
(Schneider and Tu rner, 1995). The entries, edited by the au thors, are examples and not intended
to be either comprehensive or independent.].

A . Su rprises in anthropogen ic cau ses of climate change


D evelopment
l Sou th remains proportionately behind the North in economic development.
l Transfer of wealth from Sou th to North accelerates, widening the economic disparities
between the two.
l A n u nderclass of nations is maintained owing to the diminished process of globalization.

Political organization / structu re


l The nation state weakens, leading to con ict and collapse of economic growth.
l Political economy of Ru ssia leads to large- scale resou rce degradation / depletion, especially
deforestation.
H ealth
l World mortality patterns are transformed by the emergence of a new, highly contagiou s
viru s.
l Medical breakthrou gh increases life expectancy substantially.
l Chemical pollu tion causes signiŽ cant health effects in hu mans and other species, possibly
redu cing fertility or creating demands for redu ction in su lfate emissions in A sia and Russia
and in the u se of agricu ltural biocides worldwide.
Popu lation
l Hu man popu lation growth rate does not signiŽ cantly decrease; the demographic transition
does not stay on track globally.
l Smoot h popu lation trajectories foreseen in all standard projections of world popu lation
become woefu lly inaccurate in the face of sharp departu res from monotonic trends.
Technology / policy
l Fu nding stops for technology development that wou ld facilitate a cost-effective low-carbon
fu ture.
l Montreal Protocol to redu ce emissions of ozone depleting su bstances weakened by refu sal
of some developed countries to pay their agreed-u pon shares to help developing cou ntries
acqu ire su bstitutes.
Personal valu es
l Change takes place in the political consciou sness of the valu e of natu re.

E conom y
l The global market does not control local allocation of natu ral resou rces, especially for
land and water u se; rather non- market institu tions (e.g., command-and- control econom ies,
qu asi-market economies, local institutions) remain important.
l InsufŽ cient economic growth and alternative migration patterns work to stabilize or
reverse the rates of deforestation in Sou th A merica and sou th-east A sia.
E nergy / resou rces
l India matches China in CO 2 and su lfate emissions.
l Several catastrophic nu clear plant accidents lead to ban on nu clear power before inexpen-
sive non- carbon backstop technology is available.
l A very inexpensive, low-carbon backstop technology is developed (e.g., fu el cells).
l China bu rns mu ch of its coal withou t signiŽ cant improvement in produ ction or end-u se
efŽ ciency.
G lobal change science 177
Table 1. Continu ed

l China shifts to low-carbon alternative energy sou rce (e.g., Ž nds ample su pply of natu ral
gas or develops viable biomass indu stry).
l Energy u se reverts to a parallel track with economic growth because (i) the cost of energy
conservation proves too expensive or politically intractable or (ii) a switch from an
indu strial to a service econom y proceeds slowly, or both.
l CO 2 emissions from developing cou ntries do not increase.

B . Su rprises in non- anthropogen ic cau ses of climate change


O ceanic
l A redu ction in ‘conveyor belt’ oceanic overtu rning leading to cooling at high latitudes
occu rs, despite general (bu t slower) global warming.
l H eat stored in the ocean at intermediate depths is released to the atmosphere, leading to
rapid warming.
l D imethyl sulŽ de emissions decline with redu ced sea ice, causing clou d brightness to
decrease and warming to accelerate.
l D imethyl sulŽ de emissions change with sea-su rface temperature change.

G eophysical
l A ntarctic volcanoes lu bricate ice-stream  ow causing glacial su rge and rapid sea level rise.
l Changes in volcanism is indu ced by change in climate (e.g., via sea level change).

A tmospheric
l Stratospheric cooling causes increased stratospheric clou ds and greater loss of ozone at
high latitu des.
l O zone depletion accelerates du e to lax compliance with Montreal Protocol or refusal of
certain states to live u p to their commitments.
O ther
l The G reenland ice sheet su rges.
l H igh latitude forests are not a su stained CO 2 sink.
l Positive or negative biogeochemical feedbacks become signiŽ cant to climate forcing.
l Solar radiation increases (decreases) by 0.5% , dramatically enhancing (redu cing) anthro-
pogenic warming.

C. Su rprises in environm ental consequ ences


For natu re
l D ifferential movement of species ranges in response to global environmental change
causes irreversible or very long-term ecological damage (extinction or cascading effects).
l Warmer climate becomes more stable.
l Warmer climate becomes less variable.
l Enhanced hydrological cycle leads to unanticipated extreme  oods or drou ghts.
l Clou d liquid water content increases causing increased clou d re ectivity which restrains
warming.
l Increased snow accu mu lation compensates faster ou t ow in West A ntarctica when the
R oss Ice Shelf disintegrates
l Land-cover stabilizes in Sou th A merica.
l Hurricane intensity changes signiŽ cantly with warming.
l Synergism of habitat fragmentation, artiŽ cial chemicals, introdu ction of exotic species
and anthropogenic climate change affect ecosystems in u nforeseen ways that redu ce
biodiversity.
178 Schneider et al.
Table 1. Continu ed

l Spatially varying (regional scale) competing forces create u nforeseen regional climate
anomalies (e.g., land-u se changes, aerosols or troposphe ric ozone).
For society
l R egional climate anomalies lead to econom ic and political dislocations.
l R egional environmental degradation has global impacts on economic and political systems,
which feed back on climate policy responses.

D . Su rprises in hu m an response to the advent or prospect of global change


Technology
l G eo- engineering is adopted, works as intended and mitigates most anthropoge nic climatic
changes at low costs.
l G eo- engineering is practised intermittently by only a few nations causing international
political con icts and greater environmental instability.
Policy / International accords
l The climate convention increases funding for low-cost, low-carbon backstop technologies.
l The creation of expanded wildlife reserves and migration corridors lowers impact on
biodiversity.
l More credible climate-change scenarios and better understanding of regional climate
impacts identiŽ es speciŽ c winners and losers and thereby destroys consensu s in the
international commu nity for emissions redu ctions.
l CO 2 bu ild-up in the atmosphere stalls for Ž ve years, derailing the climate convention
process.
l Intense drou ght, forest Ž res, devastating  oods and powerfu l tropical storms create an
‘emissions control’ stampede.
Valu e/Norms
l Society of the twenty-Ž rst centu ry chooses to be relatively carbon- free and resilient to
climate change.
l Society of the twenty-Ž rst centu ry chooses to maximize growth in G D P and ignores
potential long-term climatic consequ ences.

natural, social, and policy sciences. These participants were less involved with theoretical
issu es like typologies of su rprises and more attu ned to actual global change problems.
Therefore, they were asked to imagine what kinds of events, processes or ou tcomes within
their domains of expertise wou ld qu alify, in their view, as (imaginable) su rprise. O r, to
reverse the qu estion, imagine what qu aliŽ ed as su fŽ ciently deviant from conventional
views that many within each expert commu nity wou ld be su rprised shou ld it occu r (or not
occu r). Not su rprisingly, some of the responses su ggested su rprises that were opposites,
re ecting instances in which members of variou s expert commu nities held polarized or
u nclear positions, both within and across su ch commu nities.
Table 1 lists the imaginable global change (bu t linked to climate change issu es)
su rprises generated, grou ped into fou r broad categories: (A ) anthropoge nic and (B)
non- anthropogenic cau ses or ‘forcing-fu nctions’ of change, (C) environmental conse-
qu ences of these forcing fu nctions, and (D ) hu man response to the advent or prospect
of climate changes. Some of the su rprises are pegged to speciŽ c ou tcomes, others have
more general implications. Some are immediately associated with the su rprise to which
G lobal change science 179

they are linked (proxim al relations), while others constitute u nderlying factors (distal
relations) that operate throu gh variou s media before connecting to global change. For
example, u nder the su bheading of health (anthropoge nic causes) is this su rprise: world
mortality patterns are transform ed by the emergence of a new, highly contagiou s viru s.
The su rprise here is that su ch patterns might be affected su fŽ ciently as to lower the
rate of global popu lation growth su bstantially, thu s redu cing variou s demands on land
and fossil fu el consu mption that contribu te to greenhou se gases. E ach of the more
‘distal’ entries shou ld be read analogou sly.
Tables 2A and 2B take some of these candidate global climate-change su rprises and
rework them according to their Ž t by ‘su rprise arenas,’ thu s linking each directly to a
speciŽ c ou tcome. The Ž rst arena (Table 2A ) involves the scenario that by 2050 green-
hou se gas concentrations are far higher than wou ld be typically anticipated; carbon
sinks become saturated or emission rates escalate, or both, thereby causing concentra-
tions to be mu ch greater than a radiative equ ivalent of dou bling CO 2. The A G CI expert
commu nity wou ld be su rprised bu t can imagine, for example: if research and develop-
ment on low carbon sou rces of energy were effectively to cease; or if the general trend
to slower popu lation growth (the demogr aphic transition) faded, thu s driving u p
demands for energy; or if deforestation of Siberia were a major sou rce of greenhou se
gases. The second arena (Table 2B) explores su rprises that provide the opposite scenario
for 2050; greenhou se gas concentrations are lower than expected and the rates of their
deliveries continu e to fall. For instance, most of ou r exp ert commu nity wou ld be
su rprised bu t can imagine if this scenario followed from strongly implemented inter-
national accords, su ch as those to lower carbon dioxide emissions; rapid deployment of
low carbon ‘backstop’ technologies, su ch as hydroge n fu el-cell-powered cars; or because
deforestation stabilizes in the tropical world.

7. Improving the anticipation of surprise in global change


A lthough we cannot explore the sou rces of every su rprise that appears in ou r tables,
as this wou ld requ ire an extensive research endeavou r, most of them wou ld Ž t well
within the categories of the ‘open expectations’ path in ou r typological map. Those
su rprises involving the environm ental forcing fu nctions and consequ ences of climate
change often tend to follow from inadequ ate attention given to the problem (commu nal
impediments) or ou r inability at this time to measu re or model adequ ately the processes
involved (pheno menological impediments). E xamples of commu nal impediments
include:
l D imethyl su lŽ de emissions decline with redu ced sea ice.
l Climate variability changes with global temperature rise.
l Hu rricane intensity changes with global warming.
E xpanded conventional research efforts cou ld, in principle, redu ce ignorance and thu s
alter expectations in these catego ries. E xamples where fu ndamental levels of knowl-
edge of baseline data or fu nctional relationships create signiŽ cant phenomenological
impediments include:
l Cascading ecological effects (e.g., where loss of one species causes a dramatic
pest explosion).
l Land- cover changes in Sou th A merica.
180 Schneider et al.
Table 2. Candidate global climate change su rprises arranged so that they (A ) create a condition
mu ch exceeding a dou bling of CO 2 by 2050, and (B) create a condition of mu ch less than dou bling
of CO 2 by 2050.

A . G reenhou se gas concentrations are far m ore than equ ivalent to a dou bling of CO 2 by 2050
– More than 50% of incremental CO 2 remains in the atmosphere, sinks become satu rated,
and world emission rates grow sharply –
(i) No signiŽ cant policies are adopt ed because
l improved u nderstanding of climate impacts identiŽ es speciŽ c winners and losers,
thereby compromising consensu s in the international commu nity for emission
redu ctions;
l people remain unaware of, or place low valu e on, environmental impacts like
biodiversity loss;
l of the weakening of the nation state leading to con ict and collapse of treaty
obligation enforcement
(ii) D ecarbonization of the energy system stops because
l R &D on low-carbon sou rces halts;
l nu clear accidents cause a shu tdown of all nu clear plants;
l China continu es its commitment to coal u se;
l India increases coal-based energy signiŽ cantly.

(iii) E nergy/G NP ratio stops declining because


l the D emographic Transition (birth rates decline following increased af u ence) does
not take place in the developing world;
l fossil energy prices remain low;
l cost of energy conservation proves too expensive to implement;
l a transition to a service-dominated economy in the non- Western world proceeds
slowly.
(iv) Increased deforestation takes place because
l Siberia incu rs major deforestation and degradation;
l the developing world remains propor tionately behind the econom ies of the devel-
oped world, leading to su stained land-cover changes.

B . G reenhou se gas concentrations are far less than equ ivalent to a dou bling of CO 2 by 2050
– World emission rates peak and decline in the near fu ture –
(i) Strong international agreements to constrain emissions are implemented.
l D ramatic weather extremes motivate political action.
l Biodiversity preservation becomes of major societal value.

(ii) R apid decarbonization of energy systems takes place because


l low-cost biomass alternatives are developed;
l artiŽ cial photosynthesis is mastered;
l inherently safe, inexpensive nu clear power is developed;
l large natu ral gas discoveries are made in India and China;
l China and Brazil develop a large biomass-energy indu stry;
l fu el-cell-powered cars trigger an efŽ cient, hydrogen- based economy.

(iii) E nergy/G NP ratio declines sharply because


l low-energy u sing technology is improved and adopt ed globally;
l development increases per capita G NP sharply in developing world.

(iv) World economic growth rates decline sharply because


l of the demise of the nation state leading to con ict and collapse;
l of the emergence of a new, qu ick-acting and highly contagiou s viru s redu cing
popu lations globally.
(v) Minimal deforestation takes place because
l land-u se cover stabilizes in Sou th A merica and elsewhere in the tropics.
G lobal change science 181

Some of the su rprises involving anthropogenic causes of and responses to climate change
involve these kinds of impediments, bu t many more also follow from variou s com-
mu nities preferred (consciou s or not) perspectives and interpretations and their inability
to reconcile the differences across commu nity perspectives (epistemological impedi-
ments). For example:
l A n u nderclass of nations is maintained owing to the diminished process of glob-
alization.
l Ideological preconceptions and nationalism lock ou t political consciou sness of
‘global commons’ problems at scales larger than nation states.
In either of these cases, competing epistemologies have fu ndamentally different (and
often irreconcilable) interpretations of the intent, fu nction, and ou tcomes of the preva-
lent capitalist system as well as abou t the priority to be given to economic versu s other
social stru ctures in these interpretations. R ecognizing that commu nal, phenomeno-
logical, and epistemological in u ences on the range of exp ectations pu t restrictions on
the range of imaginable su rprise, the global change commu nity can tap variou s means
of dealing with them. For example, global change studies are forced to deal with the
(i) connectivity and (ii) complexity of natu ral and hu man systems which, in turn, requ ire
that problem domains, disciplines, and perspectives be crossed (also D armstadter and
Toman, 1993b, p. 3). Connectivity and complexity may ru n cou nter to the predominant
approach to prob lem solving in modern science – to increase u nderstanding by ever
narrowing and reŽ ning problems and analyses – often within well-deŽ ned disciplines.
Su ch approaches, of cou rse, carry with them implications for ou r three impediments.
We su ggest that some of the problems inherent in complexity and connectivity can be
ameliorated: (a) by su pporting work at and across the edges of the research cores that
dominate the problem domains, disciplines, and perspectives (e.g., Schneider, 1988;
Tu rner, 1991); (b) by complementing small- and large-scale approaches; and (c) by
encou raging the role of synthesis and synthesizers in both (a) and (b). None of these
su ggestions denies the fu ndamental importance of research within cores and by the core
specialist – the dominant cu rrent disciplinary science approach – or the recognition that
the root of global change studies involves systemic processes (e.g., embedded su b-cycles
within the global nitrogen cycle), each of which mu st be u nderstood as completely as
possible at smaller scales. It does, however, argu e for more integration and better
balance among activities that cou ld be labelled synthesis-edges versu s specialization-
cores. It is also important to recognize that systemic processes at large scales may not
be captu red adequ ately by a focu s on the patterns or trends of su bsystems at small
scales and that improved u nderstanding can be gained only by cycling back and forth
between small- and large-scale levels of analysis, what R oot and Schneider (1995) call
strategic cyclical scaling.
G lobal climate change research, for example, is, as it shou ld be, dominated by the
leading scientists and science centres (natu ral and social). These individu als and insti-
tutions typically develop and share common views and opinions abou t variou s facets
of the problem su ch as the likelihood that climate warming is taking place and the
central role of models in the analysis. The dominance of these commu nities in the acad-
emies and, hence, their role in agenda-setting for research and research fu nding (e.g.,
see the debate on this issu e between Bru nner, 1996, and E dwards, 1996), may obscu re
alternative views and theses, making it difŽ cu lt for the alternatives to be evalu ated
properly (i.e., leading to epistemological or commu nal impediments). Since global
182 Schneider et al.

change science, both natural and social, remains in a range of developmental stages,
the u nknowns are su fŽ ciently large to warrant attention to divergent themes abou t
similar processes and ou tcomes. To facilitate this range of research, measu res shou ld
be taken to ensu re a more open discou rse and evalu ation of alternatives, su ch as (a)
by a more open airing and professional evalu ation (as oppo sed to u ncritical, ‘equ al
time’ and equ al credibility often afforde d to polarized viewpoints in the popu lar media
– e.g. Schneider, 1990, Chapter 7) of less dominant or u nconventional views, including
those by advocacy science and scientists; and (b) by redu cing the redu ndancy of research
focu sed on the dominant views and theses while still preserving a diversity of approaches
within dominant paradigms, that is, create research ‘overlap-withou t-cloning.’
Finally, global change portends alterations to the basic processes that gove rn the state
of the biosphere. G lobal-change research, therefore, might do well to anticipate these
alterations, an effort that will requ ire u s to do more than study extant processes and
conditions alone. Variou s modes of analysis and approaches appropriate for su ch explo-
rations, bu t typically u nderu tilized in the research commu nity, shou ld be encou raged.
A mong these are: (a) backcasting scenarios from posited fu ture states and/or recon-
stru cting past scenarios in alternative ways to identify events or processes that might
happen (recognizing, of cou rse, that diffu sion processes u su ally are not reversible and
diffu sion-dominated systems cannot be meaningfu lly backcast); (b) increasing attention
to and su pport for the study of ‘ou tlier’ ou tcomes, searching for the reasons they appear
deviant and the lessons that might be drawn from them (e.g., H assol and Katzenberger,
1997); and, (c) exploring the ‘resilience’ paradigm (e.g., precautionary principle) along-
side the ‘optimization’ paradigm (e.g., cost-beneŽ t analyses) to inform policy making
and diagnose alternative ou tcomes and risk management strategies. O ther means of
improving the anticipation of su rprise in global change science wou ld emerge from
convening additional expert grou ps and asking them for more exhau stive assessments
of the issu es than we have attempted here. We su spect, however, that balanced assess-
ments will consistently lead to recommendations that ‘research as u su al’ be tempered
with more ‘less-than-u su al’ or even u nu su al research alternatives. Improvements in
dealing with scientiŽ c su rprise in climate change in particular and global change in
general, therefore, requ ire the research and fu nding commu nities to seek a better
balance among traditional and experimental research alternatives (also Kates and Clark,
1996: 31). This aim, in turn, requ ires strategies that will facilitate this balance, including
the difŽ cu lt problem of assessing ‘qu ality’ in an interdisciplinary context (e.g., Schneider,
1988).
O f cou rse, strategies that shift resou rces to more problem-oriented, high risk, or inte-
grated studies are hard enou gh to fashion with exp anding research su pport bu dgets, let
alone the cu rrent situation where traditionally federally fu nded global change research
(and its standard measu res of excellence) are increasingly u nder scru tiny by ideolog-
ical ‘anti-big-gove rnment’ politicians or bu dget-cu tting-for-its-own- sake legislators.
Nevertheless, we contend that the biggest societal and environmental returns on
research investments may, at least at the margins, lie in the su pport of u nconventional
studies focu sed on identifying and u nderstanding ou tlier ou tcomes of the variou s inter-
connected global change ‘experiments’ now u nder way on Planet E arth.

James L. Sweeney
D ept. of E ngineering E conomic Systems and O perations R esearch, Stanford
G lobal change science 183

References
A yyu b, B.M. Gu pta, M.M. and Kanal, L.N. (eds) (1992) A nalysis and M anagem ent of Uncertainty,
A msterdam: North-H olland Pu blishers.
Bogen, K.T. and Spear, R .C. (1987) Integrating u ncertainty and interindividu al variability in envi-
ronmental risk assessment, R isk A nalysis 7, 427–36.
Bonissone, P.P. and Tong, R .M. (1985) R easoning with u ncertainty in expert systems, International
Jou rnal of M an- M achine Stu dies, 22, 241–50.
Broecker, W.S. (1994) Massive iceberg discharges as triggers for global climate change, N atu re,
372, 421–24.
Brooks, H . (1986) The typology of su rprises in technology, institutions, and development, in W.C.
Clark and R .E . Mu nn, (eds) (1986) Su stainable D evelopm ent of the B iosphere,
pp. 325–48, Cambridge, Cambridge U niversity Press for the International Institute for A pplied
Systems A nalysis.
Brown, P. (1997) Stewardship of climate: an editorial for Clim atic Change, 37, 329–34.
Brunner, R .D . (1996) Policy and global change research: a modest proposal, Clim atic Change 32,
121–47.
Bu ttenŽ eld, B. and Beard, K. (1991) Visu alizing the data qu ality of spatial infor-
mation, Proceedings, A UT O-CA R T O 10. 25–28 March 1991, Baltimore, MD , vol. 6,
pp. 423–427.
Casti, J. L. (1994) Com plexiŽ cation: E xplaining a Paradox ical W orld through the Science of
Su rprise, New York: H arper Collins.
Clark, W.C. (1988) Visions of the 21st Centu ry: conventional wisdom and other su rprises in the
global interactions of popu lation, technology, and environment, in K.T. Newton Schweitzer
and Voyer, J.-P. (eds) Perspective 2000: Proceedings of a Conference Sponsored by the
E conom ic Cou ncil of Canada, Economic Cou ncil of Canada, D ecember 1988.
Clark, W.C. and Mu nn, R .E . (eds) (1986) Su stainable D evelopm ent of the B iosphere, Cambridge:
Cambridge U niversity Press for the International Institute for A pplied Systems A nalysis.
Clemen, R .T. (1991) M ak ing H ard D ecisions: A n introduction to D ecision A nalysis, Boston: PWS-
Kent.
Cohen, J. and Stewart, I. (1994) T he Collapse of Chaos: D iscovering Sim plicity in a Com plex
W orld, New York: Viking.
D armstadter, J. and Toman, M.A . (eds) (1993a) A ssessing su rprises and nonlinearities in green-
hou se warming, Proceedings of an Interdisciplinary W ork shop, Washington, D C, R esources
for the Fu ture.
D armstadter, J. and Toman, M.A . (eds) (1993b) Nonlinearities and su rprises in climate change:
an introdu ction and overview, in A ssessing Su rprise and Nonlinearities in G reenhou se
Warming. Proceedings of an Interdisciplinary W orkshop, March 1992, Washington, D C,
R esou rces for the Fu ture, pp. 1–10.
D ou glas, M. and Wildavsky, A . (1982) R isk and Cu ltu re, Berkeley: U C Press.
D overs, S.R . and H andmer, J. (1995) Ignorance, the precautionary principle, and sustainability,
A m bio, 24, 92–7.
E dwards, P.N. (1996) G lobal comprehensive models in politics and policymaking. E ditorial essay,
Clim atic Change, 32, 149–61.
Faber, M., Manstetten, R . and Proops, J.L.R . (1992) Hu mankind and the environment: an
anatomy of su rprise and ignorance, E nvironm ental V alu es, 1, 217–41.
Farman, J.C., G ardiner, B.G . and Shanklin, J.D . (1985) Larger losses of total ozone in A ntarctica
reveal seasonal ClO x /NO x interaction, N atu re, 315, 207–10.
Fischhoff, B., Slovic, P. and Lichtenstein, S. (1982) Lay foibles and expert fables in ju dgments
abou t risk, T he A m erican Statistician, 36, 240–55.
Fu nctowicz, S.O . and R avetz, J.R . (1990) Uncertainty and Qu ality in K nowledge for Policy,
D ordrecht: Klu wer.
Gu ba, E .G . (ed.) (1990) T he Paradigm D ialog, Newbu ry Park, CA : Sage.
184 Schneider et al.

H assol, S.J. and Katzenberger, J. (eds) (1997) E lem ents of Change 1996, pp. 268, A spen, Colorado,
A spen G lobal Change Institute.
H olling, C.S. (1973) R esilience and stability of ecological systems, A nnu al R eview of E cology and
System atics, 4, 1–23.
H olling, C.S. (1986) The resilience of terrestrial ecosystems: local su rprise and global change,
in W.C. Clark and R .E . Mu nn, (eds) Su stainable D evelopm ent of the B iosphere, pp. 292–317,
Cambridge, Cambridge U niversity Press for the International Institute for A pplied Systems
A nalysis.
IPCC (1996) Clim ate Change 1995 – T he Science of Clim ate Change: T he Second A ssessment R eport
of the Intergovernm ental Panel on Clim ate Change. Working G rou p I, J.T. H ou ghton, L.G . Meira
Filho, B.A . Callander, N. H arris, A . Kattenberg and K. Maskell (eds) Cambridge, Cambridge
U niversity Press.
Kahneman, D ., Slovic, P. and Tversky, A . (1982) Ju dgm ent Under Uncertainty, New York:
Cambridge U niversity Press.
Kasperson, R .E ., R enn, O ., Slovic, P., Brown, H ., E mel, J., G oble, R ., Kasperson, J.X. and
R atick, S.J. (1988) The social ampliŽ cation of risk: a conceptu al framework, R isk A nalysis, 8,
177–87.
Kates, R .W. (1985) Su ccess, strain and su rprise, Issues in Science and T echnology, 2, 46–58.
Kates, R .W. and Clark, W.C. (1996) E nvironmental su rprise: expecting the unexpected?
E nvironm ent, 38, 28–34.
Lorenz, E .N. (1969) The predictability of a  ow which possesses many scales of motion, T ellus, 21,
289–307.
MacE achren, A . (1992) Visu alizing u ncertain information, Cartographic Perspectives, (13), 10–19.
Morgan, M. G ., H enrion, M. and Small, M. (1990) Uncertainty: A Gu ide to D ealing with Uncertainty
in Qu antitative Risk and Policy A nalysis, New York: Cambridge U niversity Press.
Morgan, M. G . and Keith, D .W. (1995) Su bjective ju dgments by climate experts, E nvironm ental
Science and T echnology, 29, 468A –76A .
Moser, S. (1997) Mapping the territory of u ncertainty and ignorance: broadening cu rrent assessment
and policy approaches to sea-level rise. Ph.D . D issertation, G radu ate School of G eography,
Clark U niversity, Worcester, MA .
Moss, R . and Schneider, S.H . (1997) Characterizing and commu nicating scientiŽ c u ncertainty: mov-
ing ahead from the IPCC second assessment, in S.J. H assol and J. Katzenberger, (eds) E lements
of Change, 1996, pp. 90–135, A spen, CO , A spen G lobal Change Institute.
Myers, N. and Simon, J. (1995) Scarcity or A bu ndance: A D ebate on the E nvironm ent, New York:
Norton.
National R esearch Cou ncil (1992) Policy implications of greenhou se warming: mitigation, adapta-
tion, and the science base. Panel on Policy Implications of G reenhou se Warming, Committee on
Science, E ngineering, and Pu blic Policy. National A cademy Press, Washington, D C.
Nordhau s, W.D . (1993) M anaging the G lobal Com m ons: T he E conom ics of Clim ate Change
Cambridge, MA : MIT Press.
Nordhau s, W.D . (1991) To slow or not to slow: the economics of the greenhou se effect, T he
E conom ic Jou rnal 101, 920–37.
O ’R iordan, T. and Cameron, J. (1994) The history and contemporary signiŽ cance of the precau-
tionary principle, in T. O ’R iordan and R ayner, S. (eds) Interpreting the Precau tionary Principle,
pp. 12–30, London, E arthscan Pu blications.
O ’R iordan, T. and R ayner, S. (1991) R isk management for global environmental change, G lobal
E nvironm ental Change 1, 91–108.
R aiffa, H . (1968) D ecision A nalysis: Introdu ctory L ectures on Choices Under Uncertainty, R eading,
MA : A ddison-Wesley.
R avetz, J.R . (1986) U sable knowledge, u sable ignorance: incomplete science with policy implica-
tions, in W.C. Clark and R .E . Mu nn (eds) Su stainable D evelopm ent of the B iosphere, pp. 415–432,
New York, Cambridge U niversity Press.
G lobal change science 185

R obinson, A .L. (1982) Physicists try to Ž nd order in chaos, Science, 218, 554–57.
R owe, W. (1994) U nderstanding u ncertainty, R isk A nalysis, 14, 743–50.
R oot, T.L. and Schneider, S.H . (1995) E cology and climate: research strategies and implications,
Science 269, 334–41.
Schneider, S.H ., (1979) Comparative risk assessment of energy systems, E nergy 4, 919–31.
Schneider, S.H . (1988) The whole earth dialogu e, Issues in Science and T echnology 4, 93–9.
Schneider, S.H . (1990) G lobal W arm ing: A re we E ntering the G reenhou se Centu ry? New York:
Vintage.
Schneider, S.H . (1994) D etecting climatic change signals: A re there any ‘Ž ngerprints’? Science, 263,
341–47.
Schneider, S.H . (1997) L aboratory E arth. T he Planetary G am ble W e Can’t A fford to L ose, New
York: Basic Books.
Schneider, S.H . and Tu rner, B.L. (1995) A nticipating global change surprise, in S.J. H assol and
J. Katzenberger, (eds) E lements of Change 1994, pp. 130–45, A spen, CO , A spen G lobal Change
Institute.
Smithson, M. (1988) Ignorance and Uncertainty: E m erging Paradigm s, New York: Springer.
Svedin, U . and A niansson, B. (eds) (1987) Su rprising Fu tu res: Notes from an International
Workshop on Long-Term D evelopment, Friiberg Manor, Sweden, Janu ary 1986, Stockholm,
Swedish Cou ncil for Planning and Coordination of R esearch.
Timmerman, P. (1986) Mythology and su rprise in the su stainable development of the biosphere, in
W. C. Clark and R . E . Mu nn (eds) Su stainable D evelopm ent of the B iosphere, pp. 436–53,
Cambridge, Cambridge U niversity Press for the International Institu te for A pplied Systems
A nalysis.
Tonn, B. (1991) R esearch policy and review 34: the development of ideas of uncertainty represen-
tation, E nvironm ent and Planning A , 23, 783–812.
Toth, F.L., E . H izsnyik and Clark, W.C. (eds) (1989) Scenarios of a Critical R eview, R R 89–4.
Laxenbu rg, Au stria: International Institute for A pplied Systems A nalysis.
Tu rner, B.L., II. (1991) Thou ghts on linking the physical and hu man sciences in the stu dy of global
environm ental change, R esearch and E xploration 7, 133–35.
Weinberg, A . (1972) Science and trans-science, M inerva 10, 209–22.
Wynne, B. (1980) Technology, risk, and participation: on the social treatment of u ncertainty.
in J. Conrad (ed.), Society, T echnology, and R isk , pp. 167–202, New York, A cademic
Press.
Wynne, B. (1987) U ncertainty – technical and social, in H . Brooks and C.L. Coope r (eds) Science
for Pu blic Policy, pp. 95–113, New York, Pergamon Press.
Wynne, B. (1992) U ncertainty and environmental learning: reconceiving science and policy in the
preventive paradigm, G lobal E nvironm ental Change, 111–27.
Yager, R .R . (1992) D ecision making u nder D empster-Shafer u ncertainty, International Journal of
G eneral Systems 20, 233–45.
Yohe, G .W. (1991) U ncertainty, climate change, and the economic valu e of information: an eco-
nomic methodology for evalu ating the timing and relative efŽ cacy of alternative response to cli-
mate change with application to protecting developed property from greenhou se indu ced sea
level rise, Policy Science 24, 245–69.
Yohe, G .W. (1993) Sorting ou r facts and u ncertainties in economic response to the physical effects
of global climate change, assessing su rprises and nonlinearities in greenhou se warming.
Proceedings of an Interdisciplinary W ork shop March 1992 at R esou rces for the Fu tu re,
Washington, D C. J. D armstadter and M.A . Toman (eds). pp. 109–32. Washington, D C:
R esou rces for the Fu ture.
Z adeh, L.A . (1965) Fu zzy sets, information and control 8, 338–53.
Z adeh, L.A . (1990) The birth and evolu tion of fu zzy logic, International Jou rnal of G eneral
Systems 17, 95–105.

Das könnte Ihnen auch gefallen