What Scientific Consensus Means, and When It Was Wrong

Scientific consensus is one of the most quoted and least explained phrases in public life. It is used to end arguments, and attacked as if it meant that scientists vote on the truth.

Both uses miss what it is. A consensus is the collective judgement of the people who know a field best, built through a specific process. It deserves weight, and it can be wrong. The history of science shows both, and how the two fit together.

What a consensus is

A scientific consensus is the collective judgement, position and opinion of the vast majority of active, qualified experts in a field. It is not a poll of all scientists, and it is not unanimity; some experts will always disagree.

Some bodies use an even higher bar. Agencies such as the US Food and Drug Administration and the Intergovernmental Panel on Climate Change speak of significant scientific agreement, meaning a claim backed by the totality of the evidence, not just by most opinions.

How one forms

Consensus is not declared; it accumulates. Researchers publish results that other experts review before publication. Others try to replicate them. Meta-analyses combine many studies, and review articles and guidelines summarise where a field stands. Conferences, and sometimes formal consensus conferences, bring the arguments into one room.

The result is less like a vote and more like a long trial, in which evidence is presented, challenged and tested until most of the people who understand it are persuaded. That is why consensus forms slowly, and why it is hard to shift without new evidence.

Never proven, only not yet disproved

Philosophers of science have long warned against treating consensus as certainty. Karl Popper argued in 1934 that scientific theories can never be proven, only tested and potentially falsified. Thomas Kuhn argued in 1962 that science changes not by smooth progress but through paradigm shifts, in which an accepted framework gives way to a new one.

Both ideas explain why consensus is provisional. It is the best judgement on the evidence so far, open to revision when better evidence arrives.

When the consensus was wrong: drifting continents

On 6 January 1912 the German scientist Alfred Wegener presented a startling idea: the continents had once formed a single landmass, which he called Pangaea, and had since drifted apart. He set out his evidence in a 1915 book, including coastlines that fit together, matching fossils on separate continents and matching rock formations.

Most geologists rejected the idea for decades. Wegener could not explain what force moved the continents, his estimate of their speed was far too high, and he was not a geologist. Acceptance came only after new evidence: mantle convection was proposed in 1931, and in the late 1950s and early 1960s magnetic stripes on the ocean floor showed new crust forming at mid-ocean ridges. By 1967, plate tectonics was widely accepted.

When the consensus was wrong: ulcers

For decades doctors agreed that stomach ulcers were caused by stress, spicy food and too much acid. In 1982 the Australian physician Barry Marshall and the pathologist Robin Warren cultured a bacterium, Helicobacter pylori, and argued that it was a major cause. In 1983 their findings were rejected by reviewers for a national gastroenterology meeting.

In 1984, after failing to infect piglets, Marshall drank a broth of the bacteria himself, developed gastritis, and recovered the bacterium from his own stomach. The evidence mounted, the consensus changed, and in 2005 Marshall and Warren received the Nobel Prize in Physiology or Medicine.

How to read a consensus claim

When someone says the science is settled, three questions help. First, consensus among whom? A statement backed by the main scientific bodies of a field carries far more weight than a list of signatures from people outside it. Second, how was agreement measured? Surveys of experts, reviews of published research and formal statements are different kinds of evidence, and the strongest cases use several.

Third, what exactly is agreed? A consensus usually covers a core finding, such as that humans are warming the climate, not every detail of how much, how fast or what to do about it. Many public arguments mix up the settled core with the open questions around it, and separating the two clears up a great deal of confusion.

What those stories really show

These cases are often used to argue that any consensus might be wrong, so it can be ignored. They show the opposite. In both, the consensus changed because of new evidence, gathered and tested by scientists, and the system worked as intended, if slowly. Dissenters who are right win by producing better evidence, not by declaring the experts biased.

There is also a modern pattern in the other direction: interests that portray expert disagreement as larger than it is. Tobacco is the classic case, where action came long after the science was largely settled. Our article on where the 97 percent climate figure comes from shows how agreement can be measured, and our guide to evaluating a claim explains how to weigh it.