Clark Miller, Sheila Jasanoff, Marybeth Long, William Clark,

Nancy Dickson, Alastair Iles, Tom Parris

What are the relationships between assessments, knowledge, and knowledge communities? Assessments are often viewed as documents that communicate information from scientists to policy makers. In this paper, we develop an extended model of assessments as dynamic processes of communication that focuses attention on who is communicating, what is communicated (or not, as the case may be), and how various messages interrelate over time as assessments evolve. In particular, we consider the roles of assessments in framing and communicating information among scientists, policymakers, and citizens in modern democracies, as well as how assessments and their roles in society evolve over time. We then turn to an examination of how this model alters conventional understandings of uncertainty in assessments and the importance of assessments to the growth of new knowledge communities. Finally, we propose a series of questions related to trust and credibility, capacity-building, and assessment evaluation that flow from our proposed model of assessments as a communications process.

CC Climate Change

COP Conference of Parties

DOE Department of Energy

EPA US Environmental Protection Agency

FCC Framework Convention on Climate Change

GEA Harvard University Global Environmental Assessment project

GHG greenhouse gas

GWP global warming potential

INC Intergovernmental Negotiating Committee

ICSU International Council of Scientific Unions

IPCC Intergovernmental Panel on Climate Change

MIT Massachusetts Institute of Technology

NGO non-governmental organization

NIGEC National Institute for Global Environmental Change

NRC National Research Council

OECD Organization for Economic Cooperation and Development

RIVM Dutch Ministry of Health

UN United Nations

UNEP UN Environment Programme

WCC World Climate Conference

WMO World Meteorological Organization

WRI World Resources Institute 

Internationally, and in the United States, sound responses to climate change demand the production of reliable scientific knowledge and its effective integration into policy decisions. Traditionally, environmental assessments have been viewed primarily as bridges to connect these two activities, communicating knowledge from the scientists who produce it to the policymakers who use it. Integrated assessment, for example, has been defined (Rotmans and Dowlatabadi) as "an interdisciplinary process of combining, interpreting, and communicating knowledge from diverse scientific disciplines in such a way that the whole cause-effect chain of a problem can be evaluated from a synoptic perspective...," so as to provide useful information to decisionmakers.

 Assessments function, according to this view, primarily as inputs into policy. From this perspective, the role of assessment is three-fold: (1) to integrate disparate knowledge from many different disciplines and research programs into consensus answers to a set of well-defined questions; (2) to disseminate this consensus to the policy community; and (3) to identify gaps in scientific understanding to be filled by future research. This conceptualization gives rise to a relatively narrow set of questions about the quality of assessments. For example, what did a given assessment say about climate change? Were its conclusions correct? Did it make reasonable policy recommendations? Did it adequately identify gaps in our current knowledge?

 However, research in the Global Environmental Assessment (GEA) Project suggests that assessments may also affect the production of knowledge, in ways that have not always been clearly understood and documented. Not all knowledge, for example, is created equal from the viewpoints of producers or users, and assessments often function over time as devices through which the relevance of particular knowledge claims to particular decisions can be reevaluated. Assessments can identify areas of uncertainty and indicate the need for new areas of substantive research. Assessments, too, can affect the nature of the communities which produce scientific knowledge and use it to inform public decisionmaking. And, assessments may simply provide the resources and opportunity for new groups of individuals to interact and develop common ground with respect to policy choices.

 In the first part of this paper, we present an expanded model of assessments that draws on the research of the GEA project. We focus on the importance of the social and evolutionary dimensions of assessments and their relationship to broad communicative exercises in modern democratic societies. Subsequently, we elaborate some of the implications of this model for understanding the relationship between the production and assessment of knowledge and the communities involved in these processes. We conclude with a set of questions about the implications of this model for (1) creating new knowledge of climate change and (2) building the capacity of societies to respond credibly and effectively to such knowledge. We hope that this paper will encourage future research in this area and help shape the organization and design of future climate assessments.

 Assessments are widely recognized as tools for communicating between the worlds of research and policymaking. Research in the GEA project suggests, however, that the conventional view of assessments as documents that communicate information from scientists to policymakers captures only a limited picture of the roles that assessments play in modern democratic societies. Greater emphasis needs to placed on who is communicating, what is being communicated, and how various messages interrelate over time as assessments evolve within particular communities. Indeed, an important conclusion of several GEA studies to date is that assessments, however broadly conceived, engage only a small subset of the plausible "producers" and "users" in substantive communication. In this section, we articulate three perspectives on assessments that our research suggests may yield important insights: first, assessments may be viewed as communications processes within and among particular communities; second, assessments help frame or structure what knowledge is communicated, what is not, and how communication takes place; and third, assessment processes evolve over time, thus bringing in new participants and adapting to changes in science and politics.

 2.1 Communications Processes

 Assessments are part of broad communicative processes that support democratic governance. Both during the production and use of assessments, information is communicated amongst a wide array of experts, policymakers, and lay persons. A key finding of GEA research, for example, is that assessments both rely on and help promote communication processes within the scientific community. Individuals engaged in assessments often work in different areas of scientific research, are employed by different organizations, and are located in different geographic and sociopolitical regions (see the discussion of knowledge communities below). Integrated assessments can therefore help bridge various scientific disciplines. Assessment processes may help researchers to identify gaps in existing knowledge or new problem areas and communicate them to research managers and funders. Not least, assessment reports may help summarize and communicate basic information broadly within the scientific community, serving as a kind of reference manual; witness, for example, the widespread of citation of IPCC reports in the scientific literature.

 Even if the principal purpose of assessments is to communicate between scientists and policymakers, GEA and other research alerts us to the bi-directional character of this communication. As Working Group 1 has elaborated in its theme paper, political processes and policy elites can shape the conduct and content of assessments in myriad direct and indirect ways (Agrawala et al., 1997). This is particularly important if we take an extended view of assessments over time. As we summarize later in example 1, the U.S. has over thirty years of assessment experience in climate change. Individual assessments often occurred over many years, while assessment processes within particular institutions and communities have lasted over decades. During such extended periods of time, policymakers have numerous opportunities to communicate their needs for specific information to assessors, either directly, through, for example, negotiation of the assessment mandate, or indirectly, as, for example, assessors tailor their choices to fit perceived needs of the policy process. To quote Bert Bolin, "As scientists, we must first of all ask ourselves: What is important to people and politicians?" (Bolin, 1994) Equally important, communication from assessors to policymakers over the long term is not limited to scientific facts. Assessors can also communicate to policymakers their needs (and societies’ needs) for clearer guidance on political issues and value choices (Jasanoff, 1990). Numerous examples of the need for policy guidance have occurred in the development and comparison of national inventories of greenhouse gas emissions, where policymakers have been asked to choose between different methods of attributing accountability for particular sources and sinks.

 Assessments also provide one means for elite scientists and policymakers to communicate risks and response strategies to the lay public. The credibility of this communication is central to maintaining the public authority of governing institutions in modern democracies (Ezrahi, 1990). Thus, societies often subject important substantive and procedural aspects of assessment to oversight and regulation by legislative or judicial bodies (Porter, 1995; Jasanoff, 1995). For example, the federal court system has recently held that the National Academy of Sciences is sufficiently important to the functioning of the U.S. government for its panels to be subject to the Federal Advisory Committee Act. However, the degree of oversight and other procedural features of assessment, such as public participation or access to data and reports and the weight accorded to formal assessment in policymaking, may vary from culture to culture (Brickman et al., 1985). Indeed, Kandlikar and Sagar (1997) argue in their contribution to the GEA project that assessments as we understand them in the U.S. are essentially unknown in Indian politics.

 Besides broadening our understanding of who is communicating, GEA research also suggests that greater emphasis be put on what is being communicated. In particular, we should look at how assessments frame and reframe climate change issues for research, policy, or further assessment. Framing, as we use the term here, refers to the way in which people define a particular problem, which is shaped in turn by their knowledge and underlying views of the world (Schoen and Rein, 1994). Assessments are linked to framing in two important ways: first, frames provide the interpretive context through which assessors (and later users) judge the relevance, reliability, and weight of competing knowledge claims; and second, assessments can help stabilize (or destabilize) the framing of issues in public discourse.

 In providing the interpretive context for evaluating knowledge, frames help shape a variety of factors that are important to assessments. These can include the definitions of risk, the terms of participation, the range of policy responses, and the nature of political debate. In his contribution to the GEA project, for example, Patt (1997) notes that different assessors often frame such features as extreme outcomes in quite different ways depending on their policy goals. Below, in example 1, we describe several other aspects of framing that have been important in the evolution of climate change assessments. We also show that assessments can help stabilize frames in public discourse. For example, assessments can present alternative issue framings to the public, helping to clarify differences in interpretation and approach. Assessments commissioned to provide particular information for particular decisions, such as mitigation assessments or national inventories, can also help embed issues in emerging legal or political frameworks, often establishing which kinds of policy responses will be considered and which will not. Assessments can also link existing problems to new scientific practices or new policy initiatives by reframing problems in new ways. By reframing the issue, however, assessments may also provide new opportunities and new incentives for controversy and backlash, such as the criticism of general circulation models that has occurred since emissions reductions began to be seriously considered in the mid-1980s.

 2.3 Dynamic Processes

 Finally, GEA research points to the importance of considering assessments as dynamic processes that (1) evolve over time, (2) occur sequentially and often iteratively, and (3) typically involve long-term interactions between scientists, assessors, and decisionmakers. Changes often occur within particular assessment processes. Assessors choose and recruit participants, define and redefine problems, draft and review texts, and publish and publicize results -- all of which can substantially affect the final outcome and effectiveness of communication. The 1983 National Academy report Changing Climate began as an assessment of the U.S. government’s policy on synfuels, but ultimately took a much broader form. Changes also occur between assessments as assessors develop new research programs and new research communities, refine their ideas of the needs of policymakers, respond to new policy initiatives or exogenous crises, and reorganize assessment practices and institutions.

 The IPCC assessments, which iteratively engaged many different types of scientists and decisionmakers, are a good example of both kinds of dynamism. Over time, as shown by Agrawala (1997) in his contribution to the GEA project, interactions among participants, continued reflection on lessons learned, and responses to ongoing policy efforts and political controversies contributed to the evolution of the IPCC on several fronts: alterations in the overall assessment structure; new rules governing peer review; increased emphasis on regionally-focused assessment activities; and technical papers produced in response to the express needs of the Framework Convention on Climate Change.

 2.4 Example 1. Historical changes in the climate problem

 A brief overview of the definition of the climate problem over the past thirty years offers a glimpse into many of the features of assessments described above. Table 1 presents a summary of much of the material presented below.

 Beginning in the early 1970s, numerous U.S. institutions participated in the assessment of climate change. Their contributions illustrate the wide diversity of available definitions of risk and possible policy responses in U.S. climate politics. During the 1970s, assessors offered at least four different interpretations of how society should conceptualize and respond to climate change. The 1970 MIT Study of Critical Environmental Problems, for example, viewed carbon dioxide emissions to the atmosphere as a form of global pollution that posed a relatively unspecified threat to the global environment. The 1977 National Academy report Energy and Climate argued that the real problem was less environmental protection than long-term energy planning, of which climate change was only a small component. The 1979 World Climate Conference, in turn, suggested that the threat was less from global pollution or energy consumption than from myriad human activities at all scales (including the emission of carbon dioxide and other greenhouse gases into the atmosphere) that disrupted regional weather patterns and caused abnormal droughts and flooding. Finally, a 1980 Department of Energy report suggested that far from being a threat, climate change, if controlled, could substantially benefit human society.

 The differences among these assessments cannot be dismissed simply as disagreements over the science of climate change; all four relied on essentially similar information and came to similar conclusions about its reliability. Instead, the four assessments illustrate four different framings of climate change that emerged in different expert communities and with different policy implications for society. In turn, these differences illustrate a broader feature of American risk assessment, the pluralism of both the views presented in public debate and the access to the resources necessary to express those views through the conduct of scientific assessments. Not all societies share both of those features, and many share neither.

 Combined, these assessments and numerous others offered American policymakers and the American public a clearer picture of how choices about both broad values and specific policies could influence their ability to respond to climate change. In the early 1980s, however, changes in American public values began to emphasize the global dimensions of climate change. In their GEA paper, Long and Iles (1997) show that before the early 1980s climate impact assessment focused mainly on changes in local weather patterns; after the turn of the decade, interest increasingly turned to the threat posed by worldwide shifts in climates and global sea level rise to human society and global ecosystems. This change also reflected changes in the practices of meteorologists. Through the late 1970s, statistical techniques for aggregating regional climatic information dominated the study of climate within the meteorological community. By the early 1980s, however, assessments of climate change, such as the 1979 and 1982 National Academy panels headed by Charney and Smagorinsky, increasingly established general circulation models of the climate system as the preeminent tool for predicting future climatology. Similarly, in her GEA contribution, Fisher-Vanden (1997) shows that policy response options were also beginning to shift away (1) from an exclusive focus on energy to broader considerations of curbing deforestation, carbon sequestration, and reducing emissions of other greenhouse gases such as methane and nitrous oxide; and (2) from national responses to international treaties and global policy instruments, such as emissions trading and joint implementation. These changes in assessment focus helped stabilize the global framing of the climate problem by providing links to new policy initiatives that were less politically sensitive, particularly in the U.S., than energy planning (e.g., reducing methane and nitrous oxide emissions), or that enjoyed broad public support (e.g., eliminating CFCs and deforestation).

In turn, stabilizing a global framing of climate change fed back into the design of later assessments. Perhaps most important, the late 1980s brought new emphasis to international assessment processes. As Franz (1997) indicates in her contribution to the GEA project, early international assessments, such as the 1985 UNEP/WMO/ICSU assessment and its successors, played important roles in catalyzing the actions of particular communities of scientists and policymakers. The lack of direct oversight by authoritative government representatives in these processes ultimately proved unsatisfactory, however, and prompted a number of changes in IPCC assessment and review procedures. For example, governments nominate participants to the IPCC, coordinate national reviews of chapters, and negotiate summaries for policymakers on a line by line basis (Agrawala, GEA 1997). Similarly, during the 1990s, the production of national inventories of greenhouse gas emissions (another form of assessment) has been structured by the legal framework of the climate convention (e.g., the inventories are compiled by states, are subject to weak international review, and involve only anthropogenic as opposed to natural sources and sinks). In turn, emissions inventories have helped structure the evolution of the climate regime (e.g., by providing an authoritative source of information on the conduct of other countries that has enabled negotiations of emissions targets to go forward).

However, the framing of climate change in the early to mid-1980s as a problem of worldwide emissions reductions also opened up significant new opportunities for backlash and controversy in climate politics. For example, in the late-1980s, the possibility of international negotiations created new incentives for American industry to question the necessity of reducing emissions, and the new emphasis on climate models as the scientific basis for policy decisions opened up a significant avenue for criticism. During the 1970s, no one doubted scientific arguments that changes in regional climates had led to droughts in the Sahel and elsewhere, with significant implications for human societies. Since the mid-1980s, however, the ability of climate models to accurately predict future climates has come under continuous attack. Similarly, the global framing has been questioned by developing countries such as India and China, who view climate change less in terms of the ecological limits of the planet than the overconsumption of resources by the industrialized nations of Europe and North America (Jasanoff, 1993). Finally, the relatively unspecified threats to the global environment that form the basis of the climate convention have proven unsatisfactory to many countries that either (1) would like more specific understandings of how changes will affect their regional climate; or (2) are grappling with how to respond to short-term climate changes that are already underway. These countries continue to press both other parties to the climate convention and the IPCC for greater attention to regional climate change and adaptation.

Treating assessments as communicative processes focuses attention, in part, on questions of what is being communicated and the form taken by communication. One area that stands out in this connection is the uncertainty associated with particular scientific claims and the way such uncertainty is analyzed and expressed in assessments. In this section, we suggest additional ways in which the proposed model of assessments as a dynamic process of communication influences our understanding of knowledge and uncertainty.

Identifying, expressing, and communicating the certainty with which scientists know particular facts are important goals of climate assessments. Typically, assessments have assumed that uncertainty arises from technical issues such as inadequate data, experimental error, computational approximations, and differences in model assumptions or approaches. Consequently, assessors have assumed that uncertainty can best be reduced either through more scientific research, such as improved models and new observations, or through the standardization of different approaches. The latter, for example, has recently become important in energy forecasting and other areas of modeling with the development of "modeling forums" that test different models using identical assumptions about baseline social and economic factors.

 Methods for reducing uncertainty such as these assume that the certainty of knowledge is entirely a product of debates within the scientific community. However, GEA research indicates that the credibility of knowledge claims is also often called into question during the formulation of public policies and the effort to secure public trust in them. As new participants from non-scientific communities enter the debate, they may question previously unchallenged assumptions and methods, sometimes with considerable authority. On the strength of ethical arguments, for example, Indian representatives to the IPCC succeeded in altering the calculation of climate damages that had been based on what they believed were prejudicial assumptions about the value of statistical life. Such challenges may arise from a number of factors, many of which relate to how risks and responses are framed:

• differently motivated stakeholders and interest groups interpret the certainty of scientific claims according to different criteria;  
• public understandings of risks and policies do not necessarily mesh with expert analyses; 
• policy development takes place in transnational fora where differences occur in cross-cultural understandings of environmental risk; 
• even when public understandings of climate risks are shared, social goals that might be affected by climate policy may differ across countries, leading to different definitions of risks, costs, and benefits.

Consider, for example, the most commonly recognized area of uncertainty: inadequate data. Assessments may help identify and communicate "gaps" in available knowledge in several different ways depending on what communities are engaged in the assessments.

 Most familiar are those cases in which gaps in scientific information are identified by individual researchers or disciplines and are communicated to and evaluated by the broader scientific and research management communities during assessments. For example, climate modelers first argued in the 1960s that aerosols could matter to the radiative forcing of the atmosphere. Only in the mid- to late-1980s, however—as political controversy emerged over differences between actual and predicted temperatures and a new emphasis was placed on regional climatic variation—did research on aerosols begin to receive significant funding and attention within the global change research and assessment communities.

 Note, however, that not all assessments may identify and communicate knowledge gaps equally well. The dynamics of different assessment processes may encourage or discourage different forms of the critical appraisal of gaps in knowledge. Some assessments, for example, may permit disciplinary specialists to engage in detailed reviews of specific results, while others encourage interaction across disciplines and feature more general criticisms.

 A more interesting result of GEA and other recent research is that, by facilitating communication between scientists and policymakers, assessments can also identify additional areas of uncertainty. These may be gaps in information and or a lack of adequate methods (e.g., extrapolation techniques and model building for assessing risks) that are specifically relevant to the needs of public policymaking, but may not have been previously recognized or given importance by the scientific community. For example, as assessments of climate change began in the early 1980s to incorporate other greenhouse gases besides carbon dioxide, the perceived policy need to compare gases pointed to the desirability of a common, scientifically credible measure of global warming potentials (GWPs). Not all of the missing information that assessments may identify in this way will necessarily be technical; some may also involve political or value choices. Efforts in the late 1980s and early 1990s to use GWPs revealed unresolved political and moral issues, such as whether responsibility for climate change should be defined in historical or contemporary terms (see our discussion of GWPs below under example 2).

 In communicating risks between elite scientists and policymakers and the lay public, assessments can bring to light still other forms of uncertainty. For example, public perception of environmental risks often differs from expert analyses (Patt, GEA 1997); yet information about public perceptions of environmental risk was entirely missing from early climate assessments. The European Union’s ULYSSES project, RIVM’s efforts to include cultural theory into its models, and some of Carnegie Mellon’s recent exercises offer novel, if still preliminary attempts to represent and incorporate public perceptions of risk in integrated assessments. These efforts indicate that a new area of uncertainty has arisen over how perceptions should influence the evaluation of risk in integrated assessments. Most assessment activity still pays little attention to how the public perceives threats to the global environment.

 Assessments can affect the definition and evaluation of technical uncertainty, such as experimental error or computational approximations, when they help change the prevailing framing of environmental risks. During the 1970s, for example, climate impacts were typically viewed in terms of variations in local weather patterns. As a consequence, assessments from the 1970 MIT Study of Critical Environmental Problems to the 1979 World Climate Conference expended considerable effort determining how well scientists understood the effects of urbanization on climate and the evolution of heat from energy production facilities. As the view of climate change altered to that of a threat to the global environment, however, assessments stopped considering the reliability of knowledge in these areas and focused increasingly on uncertainties in climate models.

 GEA research also suggests that uncertainty can arise which is related not to the knowledge communicated but rather to the design of the communications process. Uncertainty can arise, for example, in the delineation of boundaries between different spheres of expertise and authority. The particular character of such uncertainty will depend on the specific communities involved. For example, controversies often arise over (1) the nature of the relevant expertise (e.g., should the problem be dealt with by social or natural scientists, biologists or oceanographers); (2) how to classify a problem (e.g., does it belong in the domain of science, transscience, policy, or politics); and (3) the basis for trust in research results or their interpretations (e.g., when should the public trust policy decisions made on the basis of expert knowledge and when should it rely on other forms of understanding).

Uncertainty may also arise in decisions about participation in particular assessments. Should participants be limited to those whose research is being specifically assessed or should they include scientists with broader and less specialized expertise (e.g., members of the National Academy of Sciences)? Should they include governmental experts (IPCC), stakeholders (Montreal Protocol Technical Assessment Panels), policymakers (Enquete Kommission), interest groups (NGO or industry assessments), or members of the lay public (ULYSSES). Equally important, once participants are settled, uncertainty may still arise over the precise forms of participation necessary to achieve particular goals (e.g., public legitimation of assessment processes or the reliability of expert contributions). For example, should experts be allowed to participate as independent individuals or only as national representatives? Should NGO and industry activists enter as active participants or only as observers?

Table 2 summarizes the different forms of uncertainty that may emerge during assessments. The rows represent major types of communication commonly involved in assessment processes. We have included as illustration scientific communication, communication between scientists and policymakers, and communication between the public and elite scientists and policymakers. The columns represent the different forms in which uncertainty may be manifested. Again for illustration we have included the three specifically identified above: inadequate data, boundaries, and participation.

It is important to note, however, that this table is not meant to represent a set of static categories into which all uncertainties can mechanically be sorted. Rather, the table should be viewed as a space in which assessments dynamically operate. Controversy, and the uncertainty which results from it, can arise in numerous forms in any assessment process, and the dynamics of assessment can move uncertainty from one box to another. In the two examples below, we offer illustrations of how, over time, some areas of uncertainty can be resolved and others opened up as assessment processes evolve, new knowledge is produced, new policy responses are proposed, and new participants enter the debate.

In 1990, the Washington-based World Resources Institute (WRI) published national inventories of greenhouse gas emissions. WRI used the observed rate at which various gases were observed to accumulate in the atmosphere, along with the differences in radiative forcing they imposed on the atmosphere, to calculate a global warming potential (GWP); they called this a "greenhouse index" to distinguish it from alternative definitions of GWPs such as that offered by the IPCC (WRI, 1990; Hammond et al., 1991). In WRI’s formulation, the problem with using GWPs was one of scientific uncertainty (upper left box of Table 2): a gap in their knowledge of relative national contributions to climate change that would impede negotiators’ ability to adopt global policies for reducing GHGs. Many of WRI’s critics within the U.S., and there were many, also framed the problem in this way and sought to "improve" WRI’s methodology. Others, however, saw the problem differently. For example, David Victor, commenting on WRI’s methods in the March 1991 issue of Environment, began with the observation that public policymaking depends on stable institutional arrangements. According to Victor, the WRI calculations relied on parameters (notably the fraction of each gas that was observed to remain in the atmosphere) that could fluctuate significantly from year to year. As a result, whatever the accuracy of WRI’s estimates, Victor viewed the uncertainty they would create for policy planning beyond one year as detrimental to their utility for policymakers. The disagreement between WRI and Victor illuminated a different kind of data gap, namely, uncertainty about what kind of data policymakers need to enable policies to work effectively, and shifted the debate to the middle box of the left-hand column.

In her comment in the March 1991 issue of Environment, Susan Subak shifted the debate to still a third location within Table 2. Subak noted that, while WRI had defended their view of "the problem as it really is," the problem could in fact be framed in other ways. If the issue was to determine how many greenhouse gases were emitted into the atmosphere from a particular geographic region in one year, then WRI had the right answer. If, however, the problem was how countries were to be held accountable for all of their emissions into the atmosphere (and Subak argued that this was the problem), then the issues became primarily political and ethical. In particular, Subak argued that WRI had ignored in their definition of national responsibility the greenhouse gases that industrialized countries had pumped into the atmosphere over the past 200 years. For Subak, the problem was about boundaries; she had located the debate in the center box of Table 2. Was the issue to be resolved by science or by politics, and how would one draw the line between them?

Arguably WRI’s most effective critics, Anil Agarwal and Sunita Narain, took a similar line in their Global Warming in an Unequal World, published by the Center for Science and the Environment, an Indian NGO (Agarwal and Narain, 1991). Asserting that WRI had based its analysis "less on science and more on politically motivated mathematical jugglery," Agarwal and Narain proposed rather that the blame for climate change lay in the "consumption of the developed countries, particularly the United States." Articulating a different set of assumptions about how responsibility should be attributed by a climate treaty, they went on to show that these assumptions would result in different totals for national inventories. Agarwal and Narain’s conclusions thus combined many Western analysts’ view that WRI used inaccurate data and methodologies with Subak’s critique that WRI had stepped outside the bounds of science. In so doing, however, they raised new uncertainties about who should participate in the compilation of national inventories of greenhouse gas emissions; Agarwal and Narain had identified column 3 as the important area of concern.

The arguments of Agarwal and Narain resonate strongly with those articulated by negotiators of the Framework Convention on Climate Change during the same period in 1991-92. The legal framework established by the treaty and subsequent decisions of the Convention’s Conference of Parties commit signatories to compiling national inventories of greenhouse gas emissions using "comparable" methodologies developed by the IPCC. Questions of the sort raised by Subak and Agarwal and Narain have, for the most part, been temporarily resolved by incorporating government experts as participants in the development and review of inventory methodologies and in the compilation of national inventories. The authority of these experts to make decisions on behalf of states and on their presumed ability to recognize and flag issues that need to be addressed by policymakers, have shifted most of the debate back from columns 2 and 3 to the upper-left hand box; for the most part, current discussions focus on the identification of data needs for compiling accurate inventories. Three cautionary notes should be sounded, however. New areas of uncertainty may arise (1) as countries develop national response strategies that implicate new interest-group participants in the climate debate; (2) as countries introduce new political goals which use information from national inventories for purposes other than national accountability (e.g., to monitor joint implementation projects, plan national policy responses, or account for emissions trading); and (3) as developing countries enter into the negotiation of protocols to the Framework Convention for emissions reductions.

Soon after the formation of the IPCC in 1988, as Agrawala (1997) has pointed out in his GEA paper, observers and assessors began to question the inadequate participation of developing countries in the workings of the IPCC. In response, the IPCC formed a special task force to expand the participation of developing countries and to establish a special trust fund to support travel by developing country representatives to IPCC meetings.

Differences quickly arose, however, as to why developing countries should participate in the IPCC and what benefits the IPCC could expect to gain from developing country participation. In part, these differences reflected uncertainty about whether the activities of the IPCC were entirely scientific, clearly political, or somewhere in between (or, in other words, uncertainty about where the boundaries of science and policy lay in international discussions of climate change; center box in Table 2). Some participants argued, for example, that the IPCC was a scientific organization. From this perspective, developing country participation was only necessary insofar as developing country scientists brought knowledge which was unavailable to other participants (e.g., of southern hemisphere climates and ecosystems). Developing country participants, by this reasoning, should be disciplinary specialists (upper right box).

Other participants argued, however, that the IPCC played important policy roles. For some, the IPCC helped to educate leaders about the dangers of climate change and possible policy responses. From this perspective, developing country participants should be generalists (again, upper right box) with the ability to digest the information presented in IPCC meetings and assessments, as well as the political connections necessary to bring that information to developing country decisionmakers. For others, the IPCC acted more as an authoritative stamp of approval to confirm for states that they needed to respond to climate change. From this perspective, developing country participants should be well-recognized experts (column 3, middle box) who might or might not be connected to specific policymakers, but whose participation would lend authority and credibility to the IPCC process in their own countries.

The debates among IPCC leaders over why developing countries should participate in the IPCC also reflected uncertainties about how the IPCC should set criteria and develop policies for improving the effectiveness of developing country participation. For those who viewed the IPCC entirely in scientific terms, improving the participation of developing countries in the IPCC required the development of new research and training programs in those countries. If the goal was to encourage societal learning about climate change, however, others suggested that the IPCC should develop more broadly based information workshops. Finally, for those who viewed credibility as the most important reason for increasing developing country participation, the best short-term policy appeared to be to raise the status of those developing country representatives who already participated and to add participants from other developing countries as quickly as possible.

Over time, it has been the last indicated model of participation which has received the greatest attention within the IPCC. The IPCC has increasingly funded the travel of developing country participants but has used its limited resources primarily to fund individual representatives from each country to attend IPCC plenary meetings (Agrawala, GEA 1997). Donor countries have, for the most part, proven unwilling to extend large sums of money to developing countries to enable them to send multiple participants to the many meetings of the IPCC working groups or to begin to build credible climate science research programs. Developing countries, in turn, have opted to use their limited resources in other ways. Thus, while the number of non-OECD countries participating in IPCC plenaries had risen to nearly a hundred by 1995 (Agrawala, GEA 1997), the number of individuals from developing countries listed as authors and contributors to the 1995 IPCC assessment have remained much smaller (Kandlikar and Sagar, GEA 1997).

 Many developing country participants, however, brought an entirely different interpretation of their need to participate from those described above. In the final report of the IPCC task force, for example, developing countries noted that many issues relevant to climate change are political (center box of Table 2). Poverty, development, equity, and access to technological and financial resources (including intellectual property rights) are fundamental, they argued, to any effort to respond effectively to climate change. Jean Ripert, the chair of the task force, noted at the Second World Climate Conference, "the struggle to master a very important aspect of the future of our planet cannot be dissociated from other efforts which the international community must make to favour a general process of development, to ensure an equitable rise in standards of living, and to equalize opportunities between peoples."

 Consequently, developing countries argued that they needed to participate in the IPCC in order to help frame how global society deals with these issues. By this standard, developing country participation in the IPCC was woefully inadequate. This raised considerable uncertainty in the minds of developing country leaders about whether the IPCC could act as an appropriate forum for formulating global responses to climate change. This uncertainty in turn had important consequences during UN debates over how to organize international climate negotiations. Between June and December, 1990, the UN Environment Programme Governing Council and the UN General Assembly debated whether the IPCC should become the official institution for negotiating the Framework Convention. Although many industrialized countries and the UN Environment Programme supported this choice, developing countries, largely as a consequence of the connection they saw between the IPCC’s technical focus and their inadequate ability to participate effectively in its operations, voted overwhelmingly to authorize the UN General Assembly to form the Intergovernmental Negotiating Committee under its own auspices.

 Treating assessments as communicative processes also encourages attention to who is communicating. Traditionally, little attention has been paid to the important roles assessments can play in the growth of research, assessment, and policy communities around issues like climate change. GEA research indicates, however, that assessments can influence the formation and evolution of these "knowledge communities" in numerous ways. In this section, we suggest several.

 Knowledge communities, as we use the term here, refers to the communities of scientists, assessors, policymakers, interest groups, and citizens who interact around particular issues and, importantly, often use assessments as part of their communicative processes. These communities may be narrow or broad, depending on the particular assessments or issue in question. They may share common policy goals, methodologies, disciplinary identities, or framings, or they may share only an interest in climate change and social responses to it.

 These communities are essential to the successful communication of risk in democratic societies. Such communities can help, for example, to (1) improve knowledge of the facts and values relevant to climate change; (2) judge the relevance, reliability, and importance of particular knowledge or value claims; (3) develop standard methods and practices that can be used in assessments; (4) link knowledge to particular decisionmaking processes; and (5) conduct reviews of both assessments and decisions. Recall, however, that various communities may bring differing perspectives to any of these tasks. During the 1970s, for example, biologists and oceanographers disagreed over how to judge the relative reliability of ocean uptake models and estimates of global deforestation. Similarly, in the mid-1990s, policymakers and scientists in India and the U.S. disagreed over the appropriateness of calculating damages from climate impacts using values of statistical life based on national incomes.

 Knowledge communities can also serve important functions in disseminating information. Such social networks, however, may be much broader than has typically been considered in climate assessments. For example, Cash (1997) argues in his contribution to the GEA project that, in gathering information for making agricultural decisions, farmers in the U.S. draw on information from a wide array of sources, including agricultural extension services, regulatory agencies (local, state, and national), the media, corporations, agricultural consulting firms, local community interactions, and land-grant universities. These networks are also often structured for particular kinds of information relevant to particular kinds of decisions. Long-term interactions may be necessary between assessors and users in order for effective communication to occur. Thus, although EPA has been developing policies for several years to mitigate methane emissions from cattle agriculture, they are only now in the process of establishing demonstration farms to show the potential benefits of their programs to the agricultural community. One important area for future research is to understand much more clearly what "niche" formal assessments such as the IPCC do, could, or should play in this rich array of communication activities.

 A key finding of GEA research is that ongoing assessment activities often help extend, enhance, and connect knowledge communities, either by identifying the relevance of communities that had previously not widely participated or by enabling new communities to form. Assessment processes, for example, have been instrumental in bringing together diverse experts and enabling them to transcend geographic, political, and disciplinary boundaries. In some cases, such as climate impacts research in the insurance and public health sectors, the interaction of experts during assessments has stimulated the formation of entirely new communities of researchers and policymakers with common interests. As a result, the entire landscape of social networks around climate impacts has changed dramatically (Long and Iles, GEA 1997). Assessments can also help establish networks of scientists and policymakers through the elaboration of common policy agendas. Franz (1997) indicates, for example, in her historical account of the Villach 1985 assessment for the GEA project, that one of the assessment’s most important consequences was to energize the activities of a small group of scientific and policy entrepreneurs in the international scientific community and U.S. domestic governing agencies.

 However, the formation and extension of knowledge communities is not easy. Limited resources, lack of institutional continuity, and methodological differences among experts can pose significant difficulties to the growth of new knowledge communities. Interaction between experts often requires considerable time, communication, and resources to defray the costs of travel, and the development of standard research protocols can require extensive training and the purchase of new equipment. Those who do not possess adequate resources for these activities are often prevented from fully participating. To the extent that assessments act to alleviate some of these barriers, they can stimulate community development. If structured differently, however, they may serve simply to strengthen already existing social networks.

 4.1 Example 4. Climate Change Impacts on Human Health

Individuals and institutions working through assessment processes have recently been instrumental in facilitating the growth of a knowledge community around the subject of climate change impacts on human health (Long and Iles, GEA 1997). The increased strength and salience of this community are evidenced, for example, by contrasting how the IPCC addressed the health impacts issue in preparing its 1990 and 1995 reports. Whereas the IPCC involved only a handful of health experts and provided relatively cursory treatment of health/climate change interactions in its work prior to 1990, it later engaged a much broader, international community of health experts and provided a much more comprehensive treatment of the subject in its 1995 report.

Many different assessments contributed to growth of the health impacts community. One of the first community-building events took place when Dr. Paul Epstein of Harvard organized authors to write a set of articles on health impacts for the Lancet. This increased awareness of health impacts issues among medical professionals and was one of the first attempts to initiate interaction among health experts who shared an interest in the effects of climate change. Community membership broadened through assessments spearheaded by the National Research Council of the National Academy of Sciences and several international institutions. In particular, recent National Academy conferences on human health have promoted interaction among health and climate specialists. The international dimensions of the community have also developed through preparation of the report, Climate Change and Human Health (1996) sponsored by the World Health Organization, the World Meteorological Organization and the United National Environment Program; and through the IPCC’s analysis of the implications of climate change for the health of human populations in 1995.

4.2 Example 5. Measuring and Monitoring Emissions of Greenhouse Gases

In recent years, many countries have established institutions for compiling national inventories of greenhouse gas emissions (GHGs). These developments provide another good example of the fruitful interaction of assessment processes with the growth of new social networks and knowledge communities. Before the late 1970s, few sources of GHGs except fossil fuels had been estimated at all. The growing importance of climate change, and the need to assess its human causes, spurred the development of a number of small, interdisciplinary groups of researchers focused on measuring and modeling GHG emissions. These interdisciplinary groups of ecologists, biologists, agricultural scientists, engineers, and atmospheric chemists have organized primarily around individual gases and sources and now provide on ongoing stream of new measurements and rapidly proliferating communities of expertise from which emissions inventories can draw.

One important area in which assessments have influenced the growth of these communities has been through the standardization of methodologies. All of the various communities involved in measuring greenhouse gas emissions have adopted some form of standardization (e.g., instrumentation, laboratory quality control mechanisms, sampling practices, field protocols, etc.). These communities have focused on standardization primarily because of disciplinary differences in experimental practice, and assessments have provided an important framework in which methodological differences can be negotiated and the goals of standardization defined.

Most of these communities, however, face continual resource shortages and a lack of institutional continuity. Because much of this work falls outside traditional disciplinary boundaries, it typically relies on special sources of funding, most of which are ad hoc. Programs such as those developed by EPA to assess U.S. emissions of methane from cattle agriculture and landfills and volatile organic compounds from biogenic sources have been essential to providing support to these communities not only for research activities but also to hold workshops and to standardize experimental and computational practices. Similarly the International Global Atmospheric Chemistry Program’s effort to assess global methane emissions from rice agriculture has helped support the development of a community of researchers in this area.

Barriers to community growth have been particularly high as these scientific networks have sought to extend membership to developing country researchers. Resource constraints often hamper travel opportunities, prevent researchers from acquiring standard equipment, and raise difficulties in even obtaining access to the publications of other researchers. In their work for the GEA project, Kandlikar and Sagar (1997) have shown that researchers in India studying the emission of methane from rice agriculture have formed a relatively strong local network. However, these researchers rarely interact with their counterparts in the U.S. and Europe, they typically publish in the Indian scientific literature (much of which is unavailable in locations outside India), and they use methods which differ from those adopted as "best practice" by U.S. and European researchers. As a result, little of their work is accorded much credibility (if it is even read) in the international scientific community. Even when programs are explicitly designed to build networks between developing and developed country scientists, however, such as the U.S. EPA’s programs to transfer techniques for measuring methane emissions from agriculture, these programs typically face severe difficulties in building trust and credibility among researchers. American researchers participating in these programs have complained that government officials in developing countries interfere in their work and publish preliminary data without permission, that scientists are beholden to producing data that will "look good" politically, and that basic infrastructural inadequacies render sophisticated measuring techniques no longer credible.

Similarly, the development, review, and publication of national inventories has helped forge new networks of communication between scientific researchers, interest groups, policymakers, and the public. The compilation of national inventories by government officials draws on research carried out by scientists and often includes experts, at least minimally, as reviewers. The Framework Convention secretariat has also developed processes for subjecting national inventories to review by officials of other governments. Perhaps most importantly, the publication of national inventories provides policymakers, citizens, and interest groups with information about the relative performance of their own and other countries in protecting the global environment.

Standardization has also been central to these efforts to build larger, more inclusive knowledge communities around the measurement and monitoring of greenhouse gas emissions. Research communities, under the auspices of the IPCC and OECD, have developed standard methodologies for use by states in compiling national inventories. In fact, the IPCC/OECD program would like to establish strong expert communities to oversee future changes in, and ultimately implementation of, its methodologies. They are hampered, however, by a lack of resources to allow expert groups to meet regularly. Instead, groups are convened in an ad hoc fashion that depends largely on participants’ willingness to volunteer time and travel support to the endeavor. As a consequence, the membership of expert groups has occasionally changed dramatically from meeting to meeting and has, at least once (for methane emissions from rice agriculture in the 1997 update), resulted in a complete change in basic methodology.

In standardizing methods for national inventories, the IPCC/OECD program has also run into difficulties in meeting the needs of developing countries. Many countries do not have access to the instrumentation and expertise required to implement rigorous experimental techniques such as those currently used in state of the art research on GHG emissions. Nonetheless, most countries have been able (with substantial financial contribution from the U.S. and others) to compile national inventories adequate to the standards of the Framework Convention. This has resulted, for the most part, from IPCC/OECD efforts to ensure that their methodologies cover only the compilation of inventories and, consequently, not the production of the data on which that compilation is based.

Similarly, the extension of knowledge communities around national inventories to developing countries has been much slower than it was to OECD countries. In part, this results from lack of resources for travel. Most national experts from developing countries have attended only one or two meetings during the previous five years of the IPCC project’s existence. In part, however, delay also results from differences in political culture that place less emphasis on the importance of scientific information in the formulation of public policy, as well as from differences in political experience. Many of the OECD countries, for example, already participated in a Europe-wide community of inventory specialists developed during the European acid rain negotiations. They are therefore already socialized to participate in international negotiations concerning policy-relevant science.

In previous sections, we have briefly described GEA research that relates assessments to both the production of knowledge and the growth of knowledge communities in modern democratic societies. Assessments, we have argued, are one part of the broad, evolving processes of communication among scientists, policymakers, and citizens that arguably are helping to constitute a global civil society. Through assessments, societies frame risks, develop the knowledge necessary for sound policy responses, and help secure public confidence that governmental actions will adequately protect the environment without sacrificing other social goals. Most importantly, we have tried to illustrate how we can improve our understanding of these processes through a more thorough probing of who is communicating, what is communicated, and how communications among multiple communities interact over time.

We have not attempted, however, to provide a definitive account of climate assessments. In fact, we hope that the communicative model we have proposed here raises more questions than it answers. Below, we suggest three sets of questions that have emerged from our own thinking about assessments, risk communication, and the production of knowledge about climate change that we believe could lead to productive discussion.

How can the view of assessments as dynamic processes of communication improve our understanding of how assessments enhance the credibility of knowledge and contribute to the building of trust among scientists, policymakers, and citizens?

The credibility of scientific knowledge is often assumed to be identical to the certainty that scientists attach to it. Considerable research for the GEA project and elsewhere indicates, however, that this view is too simple. The historian of science Steven Shapin has argued, for example, that credibility and trust may be secured by very different means in communications that are internal to particular scientific communities and in those between expert communities, between experts and policymakers, or between experts and the lay public (Shapin, 1995).

One important subject for discussion, therefore, is the particular roles that assessments may play in enhancing (or detracting from) the credibility of scientific claims in each of these different forms of communication. How, for example, do assessments enhance credibility within the scientific community? Do the views of scientists about how credibility is generated influence the organization of assessments, even if the assessments are meant to communicate information to policymakers or the public? Similarly, how do assessments influence the credibility of communication between scientists and policymakers or between experts and the public?

Comparative research across countries argues that the credibility of policy-relevant knowledge may be determined differently in different countries (see, e.g., Brickman et al., 1985). Another topic for discussion might therefore focus on differences (or similarities) between the role of assessments in enhancing credibility in the U.S. and other countries. What, for example, are some of the determinants of credibility for assessments within the U.S.? How do these draw on a presumed universalism of expertise and expert knowledge? Is this presumption shared across countries like, for example, India (Kandlikar and Sagar, GEA 1997)? Do stakeholder interests shape the credibility of knowledge claims and assessments differently in different countries? How do stakeholders or the public evaluate the credibility of experts and expert assessments (Wynne, 1996; Cash, GEA 1997), and do public evaluations of the credibility of expert assessments differ from country to country?

Finally, discussions might also productively focus on how the credibility of knowledge helps secure trust within and among scientists, policymakers, and citizens and, in particular, what mediating role assessments play in generating that trust.

How can the insights gained from viewing assessments as dynamic processes of communication improve the capacity of societies to deal appropriately with global environmental change?

Most discussions of capacity-building tend to focus on the scientific and technical infrastructure of developing countries. However, few countries, and certainly not the nations of the world as a whole, have the capacity to undertake the profound transformations of society which may be required to respond effectively to climate change. We have tried to indicate above the crucial importance of building successful networks within and among scientific, policy, and lay communities in order to effectively communicate risks, to develop sound policies for responding to environmental change, and to secure public trust in the proper functioning of government.

One potentially productive area for discussion might be to examine the role of assessments in helping Western societies to communicate risk, develop policy responses, and secure public trust, as well as how those roles might be improved in the future. What challenges, for example, do OECD countries currently face in trying to achieve significant emissions reductions, and how might assessments be designed to help overcome them? How might assessments be targeted to reduce some of the uncertainties raised in the minds of policymakers and the public by significant mitigation efforts? More generally, how can assessors design assessments so as to identify uncertainties that will be both amenable to further research and useful to progress in policy formulation?

Productive discussions might also focus on capacity-building in developing countries. Most discussions of capacity-building in developing countries focus on the deficit of technically trained people and scientific infrastructures. However, other areas of communication may also need attention. How, for example, do assessments fit into policy formulation, risk communication, and the securing of public trust in developing countries? Do social networks exist for linking scientists, policymakers, and citizens in developing countries? If not, what can be done to help developing countries forge these communities? If so, what can be done to link them effectively and productively into international knowledge communities?

What implications does the proposed view of assessments as dynamic processes of communication carry for the evaluation of assessments?

Conventionally, assessments are viewed as successful when, for example, they lead to reductions of technical uncertainties or the elimination of data gaps or when they are seen as successfully influencing policy decisions. The view presented here, however, suggests that these criteria may be too narrow to do justice to the many roles assessments can play. Discussions might fruitfully explore, for example, criteria for evaluating the ability of assessments to (a) adapt to changes or uncertainties in scientific knowledge or political circumstances by dynamically reframing problem definitions; (b) encourage (or perhaps discourage) the growth of new communities of researchers, policymakers, and citizens; (c) enhance credibility and trust within various communications processes; or (d) build the capacity of societies to deal effectively with global environmental change.