Scientific Method & History
The Gatekeepers Who Let Mendel Sleep for Forty Years
The most important quality-control system in science was invented to catch errors, but its most consequential failures have been acts of silence.
The Idea
Peer review feels ancient and authoritative — the bedrock ritual of scientific credibility. But its modern form is surprisingly recent. Most journals only made it standard practice in the second half of the twentieth century. Before that, editors largely decided what got published based on their own judgment, reputation, and occasionally whim. The system we now treat as the gold standard of scientific legitimacy is barely older than colour television. What peer review actually does is subtler than most people assume. It is not a truth-detection machine. Reviewers — typically two or three unpaid experts, reading anonymously, on their own time — are checking for logical coherence, methodological soundness, and whether conclusions are supported by the data. They are not, in most cases, rerunning experiments. Which means peer review is a filter for plausibility, not a guarantee of replicability. This matters because the system has two distinct failure modes that rarely get equal airtime. The first is the one everyone worries about: bad science slipping through. The second, less discussed, is good science being delayed or blocked — work that challenges prevailing assumptions, comes from unfamiliar institutions, or simply arrives before the field is ready to receive it. Peer review is not just a quality filter; it is also a taste filter, run by humans embedded in the same paradigms they are being asked to evaluate.
In the World
Gregor Mendel published his laws of genetic inheritance in 1866 in the proceedings of a regional natural history society in Brno, what is now the Czech Republic. The paper was methodologically rigorous — arguably more statistically sophisticated than most biology being published at the time. It was ignored for thirty-four years. When it was finally 'rediscovered' in 1900, simultaneously by three separate botanists, it immediately transformed biology. Within a decade it had become the foundation of modern genetics. Nothing about the science had changed. The ideas were sitting there the whole time. Mendel's case is extreme, but the dynamic it illustrates is not. Barry Marshall spent years unable to convince the medical community that peptic ulcers were caused by a bacterium — a claim so contrary to the consensus that reviewers treated it as not worth serious engagement. He eventually infected himself with Helicobacter pylori to prove the point. He won the Nobel Prize in 2005. More recently, the replication crisis that swept psychology, nutrition science, and parts of medicine from around 2011 onward revealed something peer review had quietly certified for decades: a significant proportion of published findings in high-impact journals could not be reproduced. The reviewers had not been negligent, exactly. The system had simply been optimised to publish interesting results, and interesting results are disproportionately the ones that are wrong.
Why It Matters
Understanding peer review's real architecture changes how you read scientific news — and how much epistemic weight you give a headline that says 'scientists find' or 'new study shows.' A peer-reviewed publication is an entry point into scientific conversation, not a verdict. It also reframes how you think about consensus. Scientific consensus is valuable and worth taking seriously, but it is built by humans operating within institutions that have their own dynamics, incentives, and blind spots. Consensus can be right, it can be slow, and occasionally it can be systematically wrong in one direction for a long time. None of this is an argument for scepticism about science — quite the opposite. The replication crisis, for instance, was identified and exposed by scientists using the scientific method. The system is self-correcting, but correction takes time, and in that interval, a lot of confident claims get treated as settled. Knowing this makes you a more careful reader, a better evaluator of risk and certainty, and — usefully — someone who is neither credulous nor reflexively dismissive when the next dramatic finding lands in your feed.
A Question to Ponder
If peer review filters for what experts currently find plausible, what kind of true ideas is it structurally least likely to let through — and who tends to have those ideas?
Get a new one of these every morning.
Start learning with Thinkable