June 25, 2015, by Andrew Gibson

Consensus in science (Brigitte Nerlich)

At the Circling the Square 2 conference there was a lot of talk about ‘consensus’ and Mike Hulme gave an inspiring key note lecture about the concept from a philosophical and sociological perspective. All this made me think a bit more about the meaning of consensus. It seems to me that the word has (at least) two meanings and that these meanings were sometimes confused during the conference. There is consensus as an unintentional emergent property and a fundamental aspect of science as a process; and there is consensus as an intentional act with a product in mind. In the first case we are dealing with the always provisional results of an invisible hand process or the unintended cumulative effects resulting from many individual intentional actions; that is to say, the emerging consensus is not part of a ‘plan’. In the second case we are dealing with the intentional outcome of concerted action, akin to majority voting. Mike Hulme distingushed between accidental and deliberate consensus, a distinction that partially overlaps with the one I am making here. In science we are, it seems to me, mostly dealing with the former, while in society we are mostly dealing with the latter. Scientific knowledge is, generally speaking, not a product of consensus as intentional or planned action.

IPCC and consensus

At the conference there was also a lot of talk about the Intergovernmental Panel on Climate Change or the IPCC. What does consensus mean in this context? In my view, the IPCC, i.e. the thousands of scientists who work on this panel, monitor the scientific process and the findings that emerge from it with regard to climate change. Over the years these scientists have noted and plotted an emergent consensus about some basic tenets of climate science and made this consensus public. The process of arriving at these scientific overviews and summaries can also be seen as being guided by the invisible hand process. As Mike Hulme noticed at the conference, he, as an individual IPCC panel member made intentional contributions to the reviewing process but the overall outcome was an unintended product of a myriad of social interactions between the scientists. However, the IPCC scientists also engage in a more deliberative consensual process through which they try to pull out some policy relevant conclusions from the literature they monitor, review and summarise. Whereas in the first type of activity the scientists take snapshots of the various end points of the scientific process regarding climate change and which are, it turns out, indicative of an increasing consensus, in the second activity they provide various agreed-upon pointers indicating policy options that can be explored by policy makers or not. (For those interested in how the word ‘consensus’ is used in AR5, see here – the report is searchable; however the word ‘confidence’ is used more frequently)

Consensus in science

But what about consensus in science more generally? What does it mean there? Do scientists use the word and how do they use it? To find out I went to Scopus, put the word ‘consensus’ into the search field and searched all articles, excluding social science ones. This gave me 167,776 document results, spread over a variety of scientific fields: medicine (89,753), biochemistry, genetics and molecular biology (42,499), immunology and microbiology (13,524), engineering (12,413), computer science (10,623), agricultural and biological sciences (9,872), neuroscience (6,048), pharmacology, toxicology and pharmaceutics (6,009), mathematics (5,645), environmental science (4448). Not much climate science I thought, but I’ll come back to that later.

Let’s first provide some examples of how the word ‘consensus’ is used in the first ten abstracts listed by Scopus: “noconsensus exists on proper donor toe length for restoration of hand function and optimal donor flap needed for recipient site healing”; “mean square consensus”; “years of consensus expert opinion as to mean arterial pressure (MAP) target and vasopressor choice in septic shock management”; “the key concepts behind the method are the modeling of the uncertainty associated with the inverse perspective projection of image features and a parameter space voting scheme to find a consensus on the vehicle state among tracked features”; “a two-layer improved average consensus algorithm (IACA) of multi-agent system (MAS) is proposed, and the consensus characteristic of which is analyzed in detail;” “it has limited sensitivity and results in a consensus sequence showing the most prevalent nucleotide per position“; “a synthesis of the state of knowledge and consensus trends regarding the various cis-acting and trans-acting components that define spatio-temporal regulation of rice genes”. Only one article used the word in the way ordinary language users would do: “No scientific consensus on GMO safety”.

When looking at how ‘consensus’ is used in article titles (24,246 document results), I found even more obscure uses, at least obscure to this non-specialist: “improved average consensus algorithm”; “eavesdropping-based gossip algorithms for distributed consensus in wireless sensor networks”; “unified random sample consensus”; “pursuit formation of double-integrator dynamics using consensus control approach” and so on. Again only two articles had titles I could vaguely get my head round: one referred to the “international society of urologic pathology consensus conference” and the other to a “consensus report”.

Three aspects of ‘consensus’

This seems to indicate that we might be dealing with three aspects of consensus in science: (a) consensus as part of the scientific process, which is however not articulated in the articles that are involved in shaping it; (b) consensus as an esoteric jargon term in various sciences, especially medicine and mathematics; and (c) consensus as used in ordinary language, focusing mostly on the intentional product side of consensus making. Overall then, (a) is invisible, but structures how science works; (b) is incomprehensible to non-specialists, but works within the fields in which it is used, and (c) is the meaning that people are used to and might therefore try to map, confusingly, onto (a).

Consensus and climate

But what about consensus in climate science? To test the waters in this context, I put ‘consensus’ plus ‘climate’ into the Scopus search engine, again excluding the social sciences. I got 1494 document results, spread over the following fields: environmental science (550), earth and planetary sciences (429), agricultural and biological sciences (300), medicine (256) and engineering (153). I looked again at the first ten abstracts and got the following results of how ‘consensus’ is used in these fields: “to learn from each other and build consensus on decision support tools developed”; “there is currently no consensus on such a rationale “ [for intentional climate law]; “swarm intelligence and consensus clustering”; “this paper also highlights the potential benefit of a consensus position on the specification of time horizon for the development of future IAMs [integrated assessment models] to aid policy making”, “there is a general consensus that temperature variation can significantly affect the fate and behavior of pesticides and microbial activity in soil; “principal component analysis (PCA) of those variables and k-means clustering on the PCA outputs revealed a reanalysis ensembleconsensus for eight macro-climate zones”; “all proxies show pronounced change after 1.5 ka, with the greatest divergence from average Holocene values occurring during the Little Ice Age (LIA), supporting the growing consensusthat the LIA was the coldest multi-centennial interval of the Holocene”.

So, while jargon uses can still be found, it seems that the word ‘consensus’ is used in the climate science articles I sampled mostly in its ordinary language meaning as ‘agreement’ amongst scientists: “general consensus”, “growing consensus”, “consensus position” – all uses that hint at the meaning of scientific consensus as emerging and potentially settling. One article in the journal Climatic Change uses the compound ‘scientific consensus’ (that we also saw above in the context of GM): “Recent research focusing on social factors affecting risk perceptions has suggested that social networks might help to explain why differences of opinion about climate change persist across segments of the lay public despite the scientific consensus.”

This is the ‘consensus gap‘ which ‘consensus entrepreneurs’, as Mike Hulme called them in his lecture, try to fill. Here we leave the emergent scientific consensus as an invisible hand process behind and enter a new phase where this procedural consensus is turned into a product that is used by some to persuade others of its existence.


This article was originally published on the Making Science Public blog

Posted in Uncategorized