Who grades this, and how

This site tells you how strong the evidence behind a claim is. That is only worth anything if you know who is doing the judging and what rule they are applying.

Who

Useful Mind is written and maintained by Devendra K Jha. I am not a psychologist. I am someone who kept meeting the same handful of findings in books and training decks, noticed that several of them had quietly fallen apart, and wanted one place that said which were which. Everything here is a reading of published work, with the source named so that you never have to take my word for it. If that is not good enough for a particular decision, it should not be — go to the paper.

Write to me at hi@digi.ai.in.

How a grade is decided

provenReplicated by people with no stake in the original result — usually a meta-analysis across many studies, a multi-lab replication, or both. If it failed somewhere, the failures are part of the record and it still held.
solidGood evidence, and nothing has knocked it down, but the independent checking is thinner than you would like. Most findings live here, because most findings have not been tested the way the famous ones have.
shakyReal, but smaller, narrower or less certain than the popular telling. Worth knowing, not worth betting on.
untestedNot a claim anybody has tested. A decision rule from a shareholder letter, a technique from a negotiator’s book, a phrase from a 1955 magazine column. Several of these are genuinely useful, which is why they are still here — but calling them evidence would be a lie, so they get their own mark instead.

That fourth mark exists because of a mistake. Everything here was originally graded on one scale, which meant a decision rule from a business book sat at the same grade as a meta-analysis — and 51 of 75 findings had piled up in the middle. Seven entries were never empirical claims at all and now say so, and ten were carrying more confidence than their evidence supports and have been moved down. Every one of those entries explains the change on its own page.

There is no formula. A grade is a judgement about the weight of the evidence, made by one person, and you should treat it as such. What the grade is not is decoration: a finding that fails independent replication does not get a gentler grade for being useful or well known — it moves to Did not hold up and stays there.

How it is funded

independentNo ads. No sponsors. No affiliate links.

Nothing on this site is recommended because somebody paid for it. There is no advertising, no sponsored entry, no commission on any book or product named here, and no paid placement of any kind. There is nothing to subscribe to and nothing to buy. It costs about the price of a coffee a month to run and is paid for out of my own pocket.

If that ever changes — if there is ever a sponsor, a paid tier or an affiliate link — it will be written on this page before it appears anywhere else on the site.

When it is wrong

It will be. Grades are judgements, papers get retracted, and fields move. Two commitments:

Findings that fail are not deleted. They move to Did not hold up with what replaced them, because knowing what to stop believing is worth as much as knowing what to believe.
Corrections to an entry are shown on that entry, not quietly applied. Bind Your Future Self is the current example: its original source was retracted in September 2026, the finding survived on other evidence, and the page says so.

Found a grade you disagree with, or a citation that is wrong? Tell me and it gets looked at. Disagreement with a source attached is the most useful mail I get.

Check the work

43 of the 90 entries link straight to the paper by DOI; the rest link to a search that finds it. Every DOI on this site was checked against Crossref rather than written from memory, because a plausible-looking wrong DOI is worse than no link at all. The whole thing is downloadable as JSON and CSV under CC BY 4.0 if you want to audit it in bulk.

Reading this once will change nothing. Useful Mind shows you one finding at a time, asks what you remember before it tells you, then brings it back a few days later. Three a day, about two minutes.

Try it — takes two minutes