What Adam Is Reading · Anti-Anti-Science
The Counterpoise
Eighteen sources on whether science really has stagnated, where peer review actually came from, and what happens when you do your own research.
This week's opening covered an interview with Eric Weinstein. He is a mathematician who argues that American science stopped working around 1970, and that the people who know why are being kept out of sight. Checking his claims turned out to be more interesting than the interview, so here is the longer version.
The pattern in what follows is consistent. The careful research agrees with a lot of what he says is broken and disagrees with what he says is breaking it. Peer review really is a recent invention, and it really did arrive for political reasons. Scientists really are getting less progress per dollar than they used to. Physics really has been stuck for decades. None of that requires anyone to be hiding anything, and every serious account of it is stranger than the hidden version.
Part one is about his claims. Part two is about the method he recommends, which is where the research gets uncomfortable for the rest of us, myself included.
The one to read first. Three weeks ago Nature published a correction to the most cited proof that science has stopped producing breakthroughs. It turns out a large share of the papers scored as most original were simply database records missing their reference lists. The scoring method reads a missing reference list as a paper that broke with everything before it. Take those out and most of the decline goes away. That is reference 3.
Part One: The Claims
Has science stagnated, where did peer review come from, is anything being hidden, and was he really ignored.
| 1Scientific Autonomy, Public Accountability, and the Rise of “Peer Review” in the Cold War United StatesMelinda Baldwin · Isis 109(3) · 2018 |
| Most of us assume peer review is as old as science itself. Baldwin, a historian, shows it actually spread through American science in the 1960s and 1970s. Congress wanted a defensible answer to why taxpayers should pay for research, and peer review supplied one. Nature did not require it for every paper until 1973. So Weinstein is right that peer review is young and that politics helped put it there. He is wrong that this makes it a plot. |
| Free PDF·Journal |
| 2Papers and patents are becoming less disruptive over timePark, Leahey & Funk · Nature 613:138–144 · January 2023 |
| This is the study people point to when they say science has stopped producing breakthroughs. The authors scored 45 million papers and 3.9 million patents on one question. Did each one send its field somewhere new, or just add another brick to the existing wall? That score fell steadily for sixty years. It is the reason the phrase “science is stagnating” entered ordinary conversation. Read it before you read the next one. |
| nature.com |
| 3Dataset artefacts can partially drive the measured decline in disruptionHolst, Algaba, Tori, Wenmackers & Ginis · Nature 656:E7–E13 · 12 August 2026 |
| This is the correction to the study above, and it is the most important thing on this list. Many entries in those databases were recorded as citing nothing at all. The scoring method treats that as the most original work possible. In fact those papers do cite other research, and the records were simply incomplete. Remove those broken records and the sixty-year decline shrinks by 87 percent in one database and 97 percent in the other, falling from −0.31 to roughly zero. Some of the problem traces back to an update in the software library used to draw the charts. |
| nature.com |
| 4The disruption index is biased by citation inflationQuantitative Science Studies 5(4):936 · 2024 |
| A second, separate problem with the same measurement. Papers today cite far more sources than papers did in 1960, and the originality score is sensitive to that, so some of what looks like a decline in bold thinking is really just the growth of modern bibliographies. In other words, the ruler changed length while people were using it to measure the patient. |
| MIT Press |
| 5Is there a secular decline in disruptive patents? Correcting for measurement biasResearch Policy · 2024 |
| The same kind of correction, applied to patents rather than papers. Once these authors account for how patent citation habits have changed over the decades, the apparent collapse in inventive originality largely disappears. That is two independent teams, working on two different datasets, reaching the same conclusion. A good deal of the stagnation everyone worried about was in the measurement. |
| Journal·Free preprint |
| 6Are Ideas Getting Harder to Find?Bloom, Jones, Van Reenen & Webb · American Economic Review 110(4) · 2020 |
| If you read only one argument that the slowdown is real, read this one, because it survives everything above. These economists measured how much research effort it takes to produce a given amount of progress, in computer chips, crop yields, and medicine. It takes far more people and far more money than it used to. The slowdown is real. It just looks like the ordinary difficulty of picking fruit from higher and higher branches, rather than anyone standing at the bottom of the tree stopping you. |
| Free PDF·NBER |
| 7Historical Record Report, Volume 1All-domain Anomaly Resolution Office, US Department of Defense · March 2024 |
| Weinstein suggests the government is sitting on recovered physics it will not share. This is the Pentagon office assigned to investigate unidentified aerial sightings, reviewing the historical record and finding no evidence of any program taking apart non-human technology. The more useful finding is further inside the report. Many people who sincerely believed they had brushed against such a program had in fact brushed against real classified projects that were entirely human-made. Secrecy is very good at manufacturing the impression of a deeper secret. |
| defense.gov PDF |
| 8A Response to Geometric UnityTimothy Nguyen & Theo Polya · February 2021 |
| Weinstein has his own theory of everything, called Geometric Unity, which he has presented in a lecture and on podcasts but never in a physics journal. This is the only paper that sits down and works through its actual mathematics, and it concludes the theory has holes that stop it doing what he says it does. Read it if you want to judge the science rather than the grievance. |
| Free PDF |
| 9Eric Weinstein vs. Sean Carroll: Pomp & FuryDecoding the Gurus, episode 130 · May 2025 |
| A podcast walking through Weinstein's televised argument with Sean Carroll, a working theoretical physicist, on Piers Morgan's show. It is worth the time because it surfaces his claim that he originated an important set of equations and was denied the credit. That grievance sits underneath most of his argument about institutions. Carroll's answer is the plain one, that describing a theory on podcasts is not the same as publishing it where other physicists can check it. |
| Episode |
| 10The Present Phase of Stagnation in the Foundations of Physics Is Not NormalSabine Hossenfelder · Nautilus |
| Hossenfelder is a physicist who accepts Weinstein's diagnosis and rejects his explanation. She agrees the deepest parts of physics have gone decades without a real advance. She blames ordinary academic incentives, and a professional habit of chasing theories because they are beautiful rather than because they can be tested. Same symptoms, no cover-up, written by someone inside the field who has paid a professional price for saying it. |
| nautil.us |
| 11Mansfield Amendment curb on basic research may spreadScience 167(3924) · 1970 |
| Weinstein blames one specific law. Between 1969 and 1971, the Mansfield Amendment barred the military from paying for university research unless it had a direct military purpose. This is Science reporting on it while it was happening, so you can see what researchers at the time actually feared would follow. Worth reading because the law and the alarm were both real. Worth treating carefully because I could not find historians who agree it broke American physics, and he presents that as settled fact. |
| science.org |
Part Two: The Method
On doing your own research, and on what the evidence says about the people who say they do.
| 12Online searches to evaluate misinformation can increase its perceived veracityAslett, Sanderson, Godel, Persily, Nagler & Tucker · Nature 625:548–556 · December 2023 |
| This is the strongest argument against “do your own research,” and it is not the argument you would expect. Across five experiments, people told to search online to check whether a false news story was true came away believing it more, by five to eleven percentage points. The reason is what the search returns. Searching a false claim pulls up unreliable sources 38 percent of the time, against 15 percent for a true story. That climbs to 77 percent when people paste the headline straight into the search box, which is what nearly everybody does. The failure is not laziness. It is that the search box itself is where the manipulation happens. |
| nature.com |
| 13Data Voids: Where Missing Data Can Easily Be ExploitedMichael Golebiewski & danah boyd · Data & Society · 2019 |
| The report that named the mechanism behind the study above. A data void is a search term almost nobody has written anything good about, which means whoever writes the bad material owns the results. People spreading a claim will coin a distinctive phrase for it, fill the void with their own content, and then tell their audience to go and look up that exact phrase. The instruction to check for yourself is not an accident. It is part of the technique. |
| Report·PDF |
| 14Support for “doing your own research” is associated with COVID-19 misperceptions and scientific mistrustSedona Chinn & Ariel Hasell · HKS Misinformation Review · June 2023 |
| Researchers surveyed the same group of Americans twice, a few months apart, and asked how they felt about the phrase “do your own research.” The people who liked it went on to hold more false beliefs about COVID and to trust scientific institutions less, but they were no more worried about the disease than anybody else. That last part is the tell. If the phrase described real caution it should have come with real concern, and instead it behaves like a way of signaling which side you are on. |
| Free PDF |
| 15Do Your Own ResearchBallantyne, Celniker & Dunning · Social Epistemology 38(3) · 2022 |
| Three philosophers, one of them the Dunning of the Dunning-Kruger effect, taking the advice seriously instead of mocking it. They ask when looking into something yourself genuinely improves your understanding and when it only supplies you with reasons for what you already believed. Their answer is that it depends on whether you already know enough to tell a good source from a bad one. That is exactly the thing the person doing the research is in no position to judge. |
| Journal·Free preprint |
| 16Motivated numeracy and enlightened self-governmentKahan, Peters, Dawson & Slovic · Behavioural Public Policy · 2017 |
| The best study on why giving people more data does not settle arguments. Volunteers were handed a table of numbers and asked what it showed. When the table was labeled a trial of skin cream, they mostly read it correctly. When the identical numbers were labeled a study of gun control, they got it wrong, in the direction of their politics. The startling part is that the volunteers who were best at math were the most divided, not the least. Being good with numbers turns out to make people better at arriving where they already wanted to go. |
| Free PDF |
| 17Do Your Own Research (?): A Weak Link between Conspiracism and Preference for First Hand EvidenceBaudouin, Altay & Mercier · October 2025 |
| Included because a reading list that only pushed one way would be doing the thing it complains about. These researchers gave people a real choice. Work out a hard visual puzzle yourself, or accept an expert's answer. People drawn to conspiracy theories mostly took the expert, the same as everyone else. The authors also wrote down their predictions before running the studies, so they could not go hunting for a flattering result afterward. The lesson is that “do your own research” may be a flag people wave more than a thing they actually do. |
| SSRN |
| 18The Skeptics' Guide to the UniverseSteven Novella · 2018 |
| Novella draws the cleanest line anyone has drawn between real curiosity and its imitation, and it is worth carrying around. A scientist asks a question because they want the answer, and changes their mind when they get one. Someone who is not really asking will put the same question again after it has been answered, and again after that. The difference is not the question. It is what happens next. |
| theskepticsguide.org |
So What
The honest version of nearly every claim in this interview is already published, and it is more interesting than the version with a conspiracy in it.
Peer review is younger and more political than most people think. Progress per researcher really has fallen. Physics really is stuck. And the single most cited proof that science stopped producing breakthroughs lost most of its effect this month, in Nature, to incomplete database records and a charting library.
Reference 3 was published on 12 August 2026, three weeks before this list. If you read one thing here, read that, then go back to reference 2 and notice how confidently it was received for three years.
Sources
The interview: “Eric Weinstein: The State of American Science, Breakthrough Coverups, and the Danger of Physics.” All-In, 26 August 2026, 90 minutes, conducted by David Friedberg. Spotify
Getting hold of these: References 2, 3 and 12 are all in Nature and sit behind the same paywall. References 1, 5, 6, 8, 13, 14, 15, 16 and 17 have free full text at the links above, and the rest are readable through most library or institutional access.
One gap: As of 29 August 2026 nobody has published a review or rebuttal of the interview itself. Everything here addresses the underlying claims rather than the episode.