by Judith Curry
The authority of a scientific body is not undermined by questioning, but rather depends upon it – Beatty & Moore
At the Conference Circling the Square that I attended in Nottingham earlier this week, Mike Hulme gave a keynote talk on “Scientists speaking with one voice: panacea or pathology?”, discussing issues around the consensus debate on climate change. Video of the talk can be found [here]. Excerpts (verbatim from slides, plus some from my written notes):
Making and selling climate consensus: “… the peer review has helped ensure a high degree of consensus amongst authors and reviewers regarding the results presented” – Sir John Houghton, July 1990
Defending the climate consensus for political purposes: is done by by politicians and also by social scientists. Hulme introduced the concept of ‘consensus entrepreneurs’, e.g. Nuccitelli and Cook.
Challenging the climate consensus are scientists (e.g. Bob Carter), journalists (e.g. Donna LaFramboise), advocacy groups
Possible motivations for consensus:
- consolidating an epistemic community
- shoring-up the authority of science
- offering a ‘firm foundation’ for policy
- closing-down dissenting voices
Consensus-making processes …
- expert elicitation (e.g. Granger-Morgan)
- group deliberation (e.g. IPCC)
- enumeration of publications (e.g. Cook et al.)
- voting (e.g. Beatty & Moore)
The ethics of consensus-making …
- beware sub-optimal deliberation
- neither reward conformity nor punish dissent
- dissensus in science is a standard of value
- “The requirement of unanimity is pernicious”
Consensus is deemed important because:
- pre-condition for political action
- the linear model of science-policy
- ‘motivating the will to change’
- ‘gateway belief’
How do publics reason? Apparently, not according to consensus-messaging …
Consensus messaging, for the past 25 years, hasn’t really been succeeding – the consensus gap continues to persist. “… increasing knowledge alone is unlikely to overcome the political divide around climate change” [Hart et al., 2015]
How does science acquire authority? Merely offering a consensus is not enough Rather than ‘trust in numbers’ (e.g. 97.1% ), what matters is trust in the process of consensus-building (e.g. IPCC; enumeration methods – Jose Duarte’s critique of Cook et al.). Scientists’ character and integrity matter more (Climategate). “The authority of a scientific body is not undermined by questioning, but rather depends upon it” [Beatty & Moore, 2010]
The IPCC claiming authority:
… “Although … there is a minority of opinions which we have not been able to accommodate, the peer review has helped ensure a high degree of consensus amongst authors and reviewers regarding the results presented. Thus the Assessment is an authoritative statement of the views of the international scientific community at this time.” [IPCC, John Houghton, 1990]
Finding policy agreement despite the ‘consensus gap’ [Howe et al., 2015]
- ‘believe that most scientists think global warming is happening’ (~35%)
- ‘somewhat or strongly support the regulation of CO2 as a pollutant’ (~65%)
Consensus as hidden framing: a way of exerting power over the public discourse on climate change:
- Political goals 2C is a political consensus … But as a substitute for conflicting goals it can close down debate about diverse goals.
- Epistemic claims that AGW is a scientific consensus … But with 97% as a substitute for uncertain knowledge of risk, it can skew public debate
Conclusions about consensus:
- There are different ways of making a consensus
- The quality of a consensus matters more than its numerical strength
- Don’t extend the reach of consensus
- Consensus has limited public leverage and policy efficacy
- A consensus is not forever
Mike Hulme provided several provocative quotes from Beatty & Moore (2010), a paper I am unfamiliar with. I took a look, this is a really good paper.
Should we aim for consensus?
John Beatty & Alfred Moore
Abstract. There can be good reasons to doubt the authority of a group of scientists. But those reasons do not include lack of unanimity among them. Indeed, holding science to a unanimity or near-unanimity standard has a pernicious effect on scientific deliberation, and on the transparency that is so crucial to the authority of science in a democracy. What authorizes a conclusion is the quality of the deliberation that produced it, which is enhanced by the presence of a non-dismissible minority. Scientists can speak as one in more ways than one. We recommend a different sort of consensus that is partly substantive and partly procedural. It is a version of what Margaret Gilbert calls “joint acceptance” –we call it “deliberative acceptance.” It capitalizes on there being a persistent minority, and thereby encourages accurate reporting of the state of agreement and disagreement among deliberators.
Published in Episteme, [link]. Excerpts:
Suppose a body of scientists has been convened to consider an issue on which a political decision hangs, and has submitted its report. Would you have more confidence in the report, and be more likely to defer to the position recommended if it were unanimously endorsed, or if there were a minority opposed? And in the latter case, would you be even more confident if the minority were sizeable, reasonable, and well-informed on the issue in question? Not that your own outlook is different in the case of science than in the case of politics, but you can surely imagine reasons for judging the two cases differently. After all, no one expects unanimity in politics; not in the light of culture, class, gender, and other differences. If a diverse voting body were to report unanimous agreement on an issue or a candidate, one might well wonder if all parties had freely spoken their minds. Science, on the other hand, is supposedly conducted by rules of reasoning that lead from the same evidence to the same conclusions, no matter who does the reckoning. In the case of science, lack of consensus might suggest that conclusions had been reached prematurely, or that the “personal or social attributes” of the protagonists had interfered with their deliberations.
Unanimity is not required to authorize the result of a scientific deliberation, i.e., to entitle deference to the outcome. Worse, the requirement of unanimity is pernicious. It has a detrimental effect on scientific deliberation, encouraging agreement where there is none in order to protect the authority of the group. It encourages misleading reports of the state of scientific agreement to the public. It undermines the epistemic equality of the deliberators. And it unfairly privileges the status quo with regard to any decisions that hinge on the outcome of the contest.
Furthermore, the persistence of a minority is not only “okay,” but is rather a benefit: minorities are not confidence busters, but confidence boosters, for the reasons suggested by Elster in the case of politics, reasons that apply in the case of science as well. What matters is the quality of the deliberation, which is enhanced by the existence of a non-dismissible minority. The kind of agreement that authorizes the result of a scientific deliberation is partly substantive and partly procedural. It is a version of what Margaret Gilbert calls “joint acceptance” (1987, 1996); we call it “deliberative acceptance.”
Our point will be that authority is not undermined by questioning, but rather strengthened by it. Authority also requires transparency, which is easier to provide when hard questions are seen as a strength rather than a weakness.
Deemphasizing consensus on substantive issues (though not entirely), and stressing consensus on deliberative quality would not only take away the temptation to hide a persistent minority position, but would instead provide a good use for it. What better way to inspire confidence in a deliberative outcome than to show that 1) the position in question had been tested against a worthy alternative; 2) the minority felt that they had been heard, that they had been treated as deliberative equals; and 3) having been heard, even the minority agreed to let the position in question stand as the group’s.
JC reflections
I found Mike Hulme’s talk to be very interesting, particularly as a complement to my own paper on the topic No consensus on consensus – Hulme takes more of a social science perspective than I did.
My main concern re the IPCC consensus seeking and the consensus entrepreneurs is that this is extremely ill-suited to a complex, highly uncertain area of science, and that it acts to bias the science. Scientists defending the consensus end up conducting acts that undermine the consensus through loss of trust in the scientists.
Beatty & Moore make really important points about the importance of minority opinions in terms of strengthening the consensus and building trust in the process.
I can’t find the Howe et al (2015) paper referenced by Hulme (ordvic spotted it here), but I think some very important points are emerging about the failure of consensus messaging to motivate the will to act and as a gateway for belief. This, taken in combination with the failure of the linear model of policy making (speaking consensus to power), suggest that a re-frame is really needed for climate policy. The public and policy makers seem to support many of the policies that are being pushed under the climate banner (in spite of the consensus gap), simply because these policies make sense for other reasons (i.e. no regret policies).
As I’ve written many times before, it is time for the IPCC to abandon their consensus seeking process, and stick to a more straightforward assessment of what we know, what we don’t know, progress since the last assessment, and provide a much broader range of possible future climate scenarios (including natural variability).
And its time to put the ‘consensus entrepreneurs’ back in a box. Imagine a world without consensus entrepreneurs; there would be no need for ‘deniers’!
Filed under: Consensus