Why Trust Science? Page 19
By contrast, Galileo’s own proposal is dimensionally homogeneous. If we take the distances it dictates (1s, 3s, 5s, 7s, 9s, 11s …) and switch to a unit of time that is twice as long, we find that, by Galileo’s proposal, the distances covered in these longer intervals are 4s (= 1s + 3s), 12s (= 5s + 7s), 20s (= 9s + 11s).… The ratio of 4 to 12 to 20 is the ratio of 1 to 3 to 5—the odd-number sequence that Galileo’s rule demands.12
That there is a dimensionally homogeneous relation among the free-fall distances was tacit common ground in this particular crisis, but it need not be common ground in every crisis. Perhaps only a scientist with Galileo’s ingenuity would have found a way to turn such sparse common ground into a strong argument for one of the rival theories against the others. Finding powerful reasons in a crisis is inevitably going to be difficult. But it is not impossible. (I think that we could take this lesson to heart in many disputes, even outside of science.)
I will conclude with one further point. According to a 2012 Gallup poll, 46% of Americans deny the evolutionary origins of human beings.13 In that same year, a congressman on the House Science Committee referred to evolution and the Big Bang as “lies from the pit of Hell.”14 Whether an American believes that climate change is taking place is highly correlated with that American’s party political affiliation.15 In this kind of political climate, I think we need to do a better job of communicating the rational basis of science. In our classes, we philosophers love to break things; we love to present the failures of various venerable proposals regarding the logic of scientific reasoning. But we need to go beyond incommensurability and underdetermination and the Duhem-Quine thesis and the new riddle of induction16 and the demise of the demarcation problem17 and the pessimistic historical meta-induction.18 Where we can, we need to give a positive account of the logic underlying scientific reasoning. Our students are able to grasp a positive account and they are hungry to see one. We owe it to them—and to ourselves—to supply one.
Chapter 5
PASCAL’S WAGER REFRAMED
Toward Trustworthy Climate Policy Assessments for Risk Societies
Ottmar Edenhofer and Martin Kowarsch
The Trump administration seems to accept climate science, though it argues against ambitious climate change mitigation efforts. In October 2017, the US Environmental Protection Agency (EPA) made a remarkable new proposal as to how climate change impacts can be evaluated in economic terms, i.e., how to calculate the “social costs of carbon” (SCC).1 They suggested social costs of only $1–6 per additional ton of carbon dioxide (CO2) emitted to the atmosphere in the near future. These numbers are extremely optimistic compared to the $45 per ton of CO2 estimated under Obama’s presidency. In contrast to the Obama administration, which included the global damages of climate change in its estimate, the new EPA calculations only take into account US domestic damages. If policy makers and investors base their decisions on the newly suggested numbers, ambitious US climate policy can hardly be justified. Calculating such costs does presuppose basic trust in the underlying climate science. This example illustrates a thorny point: scientific consensus does not imply policy consensus. Instead, calculating SSC implies controversial value judgments. Hence, in Trump’s universe, climate science does not commit the United States to ambitious climate policy, the provision of public goods, or far-reaching international cooperation. Rather, by virtue of heavily discounting future generations and dismissing public goods as globalist twaddle, the United States does no longer seem bound by its Obama-era commitments. Since scientific consensus on human-induced climate change still allows for a multiplicity of “value vectors” and policy pathways, a serious analysis of the entanglement of facts and values in expert studies, the role of scientific expertise, and the design of policies is direly needed in risk societies—i.e., societies that have to deal “with hazards and insecurities induced and introduced by modernisation itself .”2
Naomi Oreskes makes a compelling case for the trustworthiness of science, its social, value-laden, and fallible character, and the conditions for objectivity. She also reveals the rather effective diversion created in public debates by many climate change skeptics. In most cases, skeptics no longer question the overwhelming, credible scientific evidence on anthropogenic, risky climate change. Instead, they argue against particular economic or social implications of potential political responses to climate change. While they balk at the idea of engaging in open dialogue about their concerns as to the various possible (political) responses to climate change, they simply deny that the latter is a scientific problem. They urge the academic community to continuously clarify the science. And even when the science of climate change is accepted they deny the severity of damages caused by climate change. The recent SCC estimate of the Trump administration is one such example. Furthermore, according to their view, evidence-based research can never reduce the uncertainties to such an extent as to allow for the implementation of climate policies.
The often-intractable conflicts about climate policy are thus not necessarily rooted in a lack of trust in climate science—but rather in disagreement on the design of climate policies.3 Any reasonable answer to this crucial question is dependent on disputed ethical values, concepts of intergenerational and intragenerational justice, preferences, and interests (as Oreskes rightly points out). But the estimates of social science concerning the costs of action and inaction, the co-benefits and unintended societal side effects of climate policy must figure prominently too in such an answer. Natural science and technology alone cannot determine appropriate climate policies. Stubbornly insisting on the facts of climate science to counter skepticism just makes environmental controversies worse.4 Due to her emphasis on criteria for the trustworthiness of natural science Oreskes does not answer sufficiently the question as to the conditions—if at all5—under which the policy assessments provided by scientific experts can both be trustworthy and legitimate, particularly given the highly disputed value judgments involved. Therefore, we need to reflect about these integrated, transdisciplinary assessments of complex socioeconomic and political aspects. How can we specify, apply, and perhaps amend Oreskes’s convincing criteria for trustworthiness to provide guidance also for the social sciences in these cases?
Oreskes does not highlight the indeterminateness of climate policy. Rather she presents a simplified version of Pascal’s Wager6 as applied to climate policy (see table 1). Oreskes suggests that ambitious climate policy would be beneficial even if the science supporting anthropogenic climate change turned out to be wrong. A risk-neutral decision-maker would choose ambitious climate policy, if p > C / (V-E). In other words, ambitious climate policy is a no-regret option irrespective of the probability of climate change if the mitigation costs are negative due to the co-benefits of climate action. Such benefits include reduced local air pollution after phasing out coal or less dependency on fossil fuel imports. In contrast to Oreskes, the Trump administration has discounted the damages (V) to such an extent that investments in ambitious climate policy cannot be evaluated as beneficial for US society, even if the probability of dangerous climate change is very high.
TABLE 1. Pascal’s Wager Applied to Climate Policy
Dangerous Climate Change Probability p
Harmless Climate Change Probability 1−p
Ambitious Climate Policy
Low Damages (E) + Mitigation Costs (C)
Mitigation Costs (C)
No Climate Policy
Irreversible High Damages (V)
Zero Net Costs (0)
The payoffs in the climate policy wager are not merely givens that result from a state of nature—they should be understood as an emergent outcome of a social learning process. Most experts would evaluate Oreskes’s no-regret argument for ambitious climate policy as overly optimistic.7 It is, however, equally pessimistic for Trump to ignore the damages of climate change for humankind as a whole by discounting future climate damages to an excessive degree—particularly when their partly irreversibl
e character is taken into account. Rational decision-makers would choose to reduce emissions both immediately and considerably. They would launch a societal learning process designed to adapt climate policies according to new insights about future damages and mitigation costs as well as other effects and risks associated with these policies. Such a social learning process would draw on evidence from ex-post policy analysis to determine which policy instruments worked and which did not. Recent analyses of the EU ETS8 are examples of such studies. The iterative nature of the learning process is conducive to making incremental, yet successful steps in climate policy on various governance levels while avoiding irreversible lock-in effects. If democratic societies want to tackle climate change as a serious risk for current and future societies they need to engage in rational discussions and learning processes about alternative, available solutions and their (often uncertain) implications. Such discussions must include an examination of the specific risks and the pros and cons of different policy pathways as well as interdependencies with different policy fields, governance levels, and time scales. A scenario in which human-induced climate change transpires to be a mistaken hypothesis would then be only one of many different possible scenarios—though a rather extreme and unlikely one, as Oreskes herself suggests.
Salient examples of difficult climate policy issues include: the evaluation and comparison of climate damages in different regions; appropriate prices on CO2 emissions for different countries; the sector(s) that should be affected by a carbon pricing scheme; the specific distributional effects of carbon pricing on different social groups; policies that could soften the blow of the side effects of bioenergy, such as food insecurity, deforestation, and endangered biodiversity; the timing and size of subsidies for renewables; the socially acceptable location of wind-power turbines; the benefits and the risks of nuclear energy; the appropriate volume of international technology transfer; the potential and risks of negative emission technologies (carbon dioxide removal) or solar radiation management; and the employment effects of a rapid diffusion of electric cars.
Several ineffective and inefficient climate policy decisions have already been made precisely because such questions were not answered well. The complexity of these questions requires serious, integrated assessments to facilitate a learning process about the available policy pathways. This presupposes collaborative, scientific explorations of the various societally relevant ramifications of different pathways from different disciplinary perspectives and viewpoints.9 A more complex version of Pascal’s Wager on climate policy is needed particularly when assuming the irreversibility of some climate damages. The burden of proof in risk societies thus mainly lies on assessments of policies to show that a particular climate policy choice is better than its viable alternatives. Such an assessment has to consider a policy’s overall effects, side effects, and co-benefits that must be gauged against the backdrop of Sustainable Development Goals and other policy goals or values. Jointly with the stakeholders and decision-makers involved, scientific experts would act as cartographers of policy alternatives while decision-makers remain the navigators.10 We highly appreciate Oreskes’s thoughts concerning the trustworthiness of (climate) science. Yet we stress the need for more focus on the policy assessments to facilitate a better public debate about different solutions and their implications.
The frequently contested value judgments implied in ex-post and ex-ante evaluation of policies are certainly one of the key challenges in determining their trustworthiness and legitimacy. As Oreskes mentions, facts and values—including cognitive, epistemic, and ethical values—are always intertwined in scientific research.11 It is somewhat surprising that Oreskes does not discuss at greater length how objectivity can emerge concerning disputable ethical values implied in scientific knowledge. Oreskes only points to her hope that there is perhaps much more overlap between our deepest values than we often think—a rather remarkable belief in ethical common ground. Talking about our values and identifying overlap is indeed desirable. She admits, however, that substantial dissent remains regarding some fundamental values and, particularly, their more specific meaning with respect to current policy choices that usually exhibit complexity and uncertainty. Divergent sets of values also play a central role concerning the increasing political divide in several Western countries.
There is good reason to believe that rational discussions about value-laden policy issues are possible, even if they have become partly ideological and almost “religious” disputes. Centuries ago, Pascal made the revolutionary attempt to initiate a rational discussion about nothing less than the most fundamental religious question, namely the very existence of God. If it transpired that God did not exist, the personal costs and benefits of having lived a religious life are comparatively lower than not believing in God, Pascal reasoned. Despite several shortcomings of Pascal’s pragmatic wager, including the narrow set of alternatives presented, the latter was one of the first conceptual frameworks for decision-making under uncertainty. John Dewey’s (1859–1952) pragmatist philosophy helped to develop this framework further.12 Similar to Pascal, Dewey emphasizes the philosophical necessity of exploring and evaluating the diverse practical consequences of particular hypotheses, including normative, methodical, and empirical assumptions. All scientific claims and other hypotheses are conceptualized as means of achieving practical ends that are somehow relevant to humankind, assuming an ends-means continuum.13
Dewey argues for the possibility of trustworthiness and objectivity of value-laden but always fallible scientific claims—including those of assessments of policies or ethical debates. On the basis of Dewey’s viewpoint, hypotheses—as potential means for resolving problematic situations—can be regarded as trustworthy if they turn out to be reliable in terms of their practical consequences, i.e., if they repeatedly help to transform an indeterminate problematic situation into a determined one in a reliable way. According to this “natural realism,” successful results of a pragmatist inquiry could be applied to similar situations; indeed they can serve as the premises for further inquiries as cumulative experience, potentially qualifying as objective, “warranted assertability” under good-enough conditions for such an inquiry.14 Ends cannot justify the means and both have to be critically assessed via their practical consequences. If, for instance, even the best available climate policy has severe side effects, then the initial policy goals or even the underlying values might have to be revised.
This Deweyan perspective on trustworthiness, albeit largely in alignment with Oreskes’s view, offers a nuanced amendment. While the idea of experimentation is inspired by the success of natural science it is applied to all sorts of inquiries—including highly value-laden ones. Moreover, the decisive “practical implications” go well beyond instrumental ones and rather include everything that matters for human existence, for instance in spiritual terms.
Instead of presenting allegedly value-free facts or lobbying for a particular policy option scientists can help facilitate more constructive discussions among stakeholders about highly value-laden policy issues like climate policy. Drawing on Dewey’s thoughts we propose to go further than normative transparency in scientific assessments and consistently embed divergent values and principles like equality, liberty, purity, nationalism, etc. in different future scenarios and policy pathways. In an inter- and transdisciplinary manner jointly with stakeholders, the various practical implications of these alternative policy pathways can then be critically compared and evaluated.15 These implications include costs and benefits in the narrow economic sense. Beyond that they also include everything that matters for a society and threatens the legitimacy of policies. “Costs” can, for instance, be perceived as prohibitively high when fundamental rights or procedures are violated.
Such an evaluation of policy pathways may lead to revisions or, at least, reinterpretations of initially assumed, perhaps one-sided sets of values, principles, policy goals, etc. This could happen if it turned out that there were ad
verse side effects and considerable limitations, e.g., by neglecting other societally relevant values. This exercise can thus enable diverse stakeholders to clarify their policy positions. It can also facilitate the identification of previously unidentified overlap between different viewpoints at least on particular policy instruments and pathways. Left-wing liberals and right-wing conservatives, for example, may still agree about effective carbon pricing.
The essence of the Deweyan approach to value-laden policy assessment consequently lies in its ability to transform heated, entrenched policy conflicts into much more constructive discussions and learning processes about policy alternatives and their complex practical implications. Epistemic trustworthiness can be achieved through a careful exploration and evaluation of the diverse direct and indirect implications of future policy pathways—i.e., of hypotheses conceived of as means within an ends-means continuum—via the feedback loop between ends, means, and consequences. Moreover, the legitimacy of these assessments can mainly be fostered by exploring alternative future policy pathways, which represent different prominent sets of values and policy beliefs, and by actively engaging with a diversity of stakeholders during the assessment process.
The proposed model is largely different from both current practices and the literature regarding the science-policy interface. For instance, our model emphasizes that there is no “value-neutral way” to assess policy pathways and that the feedback loop between policy goals, means, and their practical implications is crucial. Usually, only small sets of alternatives are explored based on narrow evaluation criteria without serious transdisciplinary collaboration or explicit assessing of the underlying policy goals and ethical values via their practical implications. Instead of mere discussion or brokerage of alternative options, the cartography of the implications of alternative futures puts emphasis on learning among all actors involved. This includes learning about alternative problem framings and worldviews.