Consilience, convergence, and consensus

In an editorial in this week's Science, the journal's editor Holden Thorp develops an argument that the notion of “scientific consensus” has confused public discussion of science. In Thorp's view, the public misunderstands “consensus” as something like the result of an opinion poll. He cites the communication researcher Kathleen Hall Jamieson, who observes that arguments invoking “consensus” are easy for opponents to discredit merely by finding some scientists who disagree.

Thorp notes that what scientists mean by “consensus” is much deeper than a popularity contest. He describes it as “a process in which evidence from independent lines of inquiry leads collectively toward the same conclusion.” Leaning into this idea, Thorp argues that policymakers should stop talking about “scientific consensus” and instead use a different term: “convergence of evidence”.

It would be a big move for a magazine representing the entire breadth of American science to reject the idea of scientific consensus.

For the last twenty years the idea of “scientific consensus” has been widely adopted by scientific organizations and policymakers, especially applied to politically contentious topics such as climate change, vaccine hesitancy, and COVID-19 response. Many organizations shifted their policy advocacy by issuing statements reflecting the consensus of their members. Science itself helped launch this era with the publication of Naomi Oreskes' 2004 article, “The Scientific Consensus on Climate Change”. This article recounted the number of organizations representing scientists that had issued statements or policy documents about the evidence for human-induced climate change.

The purpose of such statements was to counter public perceptions that there might be significant scientific disagreement about climate change. Corporate interests—memorably named “merchants of doubt” by Oreskes—often work against the notion of consensus by funding and promoting studies that contradict mainstream results. A range of social science research has shown that many people are more receptive to scientific ideas if they are told that the ideas reflect a scientific consensus. Yet it takes only a few vocal dissenters to raise the prospect that organizational statements may emphasize “consensus” but in reality are a power play to silence debate.

Without question, the idea of “scientific consensus” has had a rough time lately. Public statements from organizations that represent scientists don't carry the weight that they did twenty years ago. Some of that damage was self-inflicted, epitomized by the 2006 “Pluto is not a planet” decision by the International Astronomical Union. Scientists in some fields have adopted the frame of “consensus” when discussing topics where substantial disagreement exists—for example, the risks and benefits of gain-of-function research in viruses. Many people increasingly distrust scientists' motivations and see them as disconnected from their own everyday experience.

Back in the 1980s and 1990s the growing discipline of science and technology studies worked to understand how scientists develop shared ideas—that is, how does “consensus” form, and how do they employ claims about consensus to shape their work? Researchers in this area observed that science is intertwined with political and social currents.

For example, Sheila Jasanoff studied ways that scientific facts are co-created with social order. She described how consensus emerges through a process of setting boundaries for what is legitimate or illegitimate, and noted the strategic element to the negotiations over what will be accepted by the community. These negotiations over what Thorp describes as “vetted and validated science” often seem to have little to do with a steady buildup of evidence. They really can seem like a popularity contest.

From that perspective, changing the conversation to talk about “convergence of evidence” might seem like a good idea. Thorp urges us to “take ever more care in clearly delineating scientific results from scientists’ opinions”. His preferred term, “convergence of evidence”, was most widely used in the 1950s.

But I'm skeptical that such a change in terminology will be any better. Usually scientists who talk about a convergence of evidence are looking within a discipline or small research area. The problem of “scientific consensus” is not only a communication problem; it is a problem in the way scientific work is conceived and assessed. Researchers within the same field tend to formulate their research and choose their tests in the same ways, after being trained by the same advisors, funded by the same agencies, and assessed by the same reviewers. The boundary-setting that Jasanoff wrote about doesn't only happen at the end of the scientific process; it happens from the very beginning.

In my own work I often question ideas that seem to be consensus. That doesn't mean that I reflexively reject every idea that has wide support among my colleagues, but it does mean that I don't take other scientists' opinions at face value. I try to critically examine the assumptions—often unstated—that underlie widely-shared ideas. I think a willingness to question consensus is essential to scientific progress.

Some years ago I was invited to be part of a transdisciplinary symposium to share perspectives on the late Edward O. Wilson's book, Consilience: The Unity of Knowledge. Wilson's main idea was that all the fields of academic study ultimately will be unified; that they all provide insights about the same reality at different levels. That idea of consilience has stuck with me over time.

The idea was formulated by the nineteenth-century philosopher William Whewell, who also coined the word scientist. Whewell wanted to understand how observations give rise to theories. His idea was that the induction of a hypothesis or theory from observations requires another step, a step in which evidence developed by other means of observation must also show consistency with the same theory or hypothesis. He used the term “consilience” for this matching of evidence of different kinds. 

Michael Ruse noted Charles Darwin's work as a hallmark of the consilience approach. Darwin brought together evidence from entirely different fields of inquiry: animal and plant breeding, geology, natural history, biogeography, sociology, and many others. He had a remarkable ability to answer questions in one field by examining data in another field entirely. The ability to bring together observations that seem disconnected from each other, explain all of them with one unifying explanation is a powerful mode of scientific thinking.

Consilience of evidence also helps to answer criticism that scientists are closing off debate by excluding ideas that do not fit within their disciplinary boundaries. Where “convergence of evidence” may seem inward-facing, confined to a single research tradition, consilience is explicitly outward-reaching. It requires translation and integration across disciplinary boundaries and sometimes even across different ways of knowing.

The public asks more of science than scientists often ask of their own fields. That may seem unreasonable. But I think scientists can answer the call to expect more in their work, to communicate more effectively to more people, and to be transparent about what they know and don't know. It's not the process of “vetting and validating” that accomplishes these goals; it's the transparency, replicability, and openness that makes it possible for anyone to evaluate the work.

References

Jasanoff, S. S. (1987). Contested Boundaries in Policy-Relevant Science. Social Studies of Science, 17(2), 195–230. https://doi.org/10.1177/030631287017002001

Jasanoff, S. (2004). States of Knowledge: The Co-Production of Science and the Social Order. Routledge. https://doi.org/10.4324/9780203413845

Oreskes, N. (2004). The Scientific Consensus on Climate Change. Science, 306(5702), 1686–1686. https://doi.org/10.1126/science.1103618

Ruse, M. (2007). Charles Darwin. In M. Matthen & C. Stephens (Eds.), Philosophy of Biology (pp. 1–35). North-Holland. https://doi.org/10.1016/B978-044451543-8/50003-4

Thorp, H. H. (2025). Convergence and consensus. Science, 388(6745), 339–339. https://doi.org/10.1126/science.ady3211

Wilson, E. O. (1999). Consilience: The Unity of Knowledge. Knopf Doubleday Publishing Group.