What observation would prove this wrong?
Reasons from 1934–1980
Click & holdScroll to dive in
Independent research intelligence.
The reviewer that doesn’t think like the others.
Every AI learned from the same pages, so they miss the same things. Feynman reads your argument three different ways and shows you where it gives.
Feynman · in four rows
001Six claims went in. Eight objections came back.
A real review of the deadlines and procrastination study, as it stood. Nothing on this page was written to look good.
ReviewDeadlines and procrastination
Claim 0102 objections
Claim 0202 objections
Claim 0301 objection
Claim 0401 objection
Claim 0501 objection
Claim 0601 objection
Sources retrieved14
Sources cited12
Tests proposed01
Open this review002Ten lenses. Each asks one question.
A lens is one way of reasoning, named for what it tests. Each is built from work published no later than 1980, so none learned to think from today’s consensus. These are the lenses a review chooses from. It uses three.
Does the conclusion follow?
Do the model’s assumptions match the thing modelled?
Reasons from 1928–1957
Is the mechanism real?
What experiment separates the explanation from a good story?
Reasons from 1948–1980
What does the outlier show that the average hides?
Reasons from 1931–1980
Does measuring or acting on the system change it?
Reasons from 1943–1964
Is the measurement sound?
Which distinctions survive the measurement, and which are lost?
Reasons from 1937–1961
What would a passed test actually establish?
Reasons from 1936–1954
Does it hold elsewhere?
Who or what was left out of the sample?
Reasons from 1897–1963
What did the real decision process have time to consider?
Reasons from 1947–1980
Which unknowns can the stated odds not cover?
Reasons from 1919–1946
003You already know how to use it.
001 / Bring it
Paste a draft, a DOI or a PDF.
002 / Agree on it
Feynman says your argument back as numbered claims. You fix what it got wrong.
003 / See where it gives
One sentence first. Under it, rows you can open: who objected to what, the evidence, and the test that would settle it. Then ask it anything else.
004What it costs, and what we haven’t shown.
For doctoral and postgraduate researchers, early-career scientists, mathematicians and independent researchers. Sign-ups aren’t open yet.
Membership3 months
Lenses per review03
Open indexes searched04
Citations checkedevery one
Shown to beat a plain AInot yet
The comparison is fixed under fingerprint 8cc6e2ec4fda and has not been run. It will be published whichever way it comes out. Read the design.