Pith. sign in

CONTRACT 196,528 papers reviewed 2,907,120 papers cataloged 12,094,767 citations mapped

Every paper gets an honest, public read.

pith.science · about · updated 2026-08-05

How a review works

Most scientific papers are never carefully read by anyone outside the authors' own circle. Pith reads them all. Every paper gets a close, public review: what it claims, what that claim stands on, and what would prove it wrong. Journals sell prestige and search engines sell links; Pith tells you where a paper actually stands.

papers reviewed and published
196,528 live · as of 2026-08-05
papers cataloged
2,907,120 live · as of 2026-08-05
citations mapped
12,094,767 live · as of 2026-08-05
a new review published
every 38 seconds live · trailing 30 days

  1. Every paper gets read

    A careful expert reading used to be something only a few papers ever received: two anonymous referees, once, behind closed doors. On Pith it is the default. Every paper is read closely, and the reading is published next to the paper where anyone can see it, argue with it, and hold it to account. No paper is too obscure, no verdict is for sale, and nothing happens in private.

  2. Ask the literature a question

    Every review answers the questions a sharp reader would ask: what is this paper really claiming, what does that claim rest on, and what would prove it wrong. Asked of one paper, that is a review. Asked of millions, it becomes a map of science itself: who asserts what, standing on what, contradicted by what. You can look up a claim and see where it is strong and where it is thin.

  3. Citations that mean something

    A reference list tells you a paper cites forty things. Pith tells you which of them matter: the work the method came from, the result the whole argument depends on, the paper that disagrees. Follow any claim upstream to what it stands on, or downstream to everything that leans on it.

  4. Reviews that get better with time

    A journal review happens once and is frozen forever. A Pith review can be run again. As the reading gets sharper, the whole library is re-read with it, including the paper nobody has looked at since 1994. The scrutiny on every paper improves year over year, which no human review system has ever been able to promise.

  5. Trust you can check

    You should not take a machine reviewer's word for anything, so Pith makes it earn yours. We hide known errors inside test papers and score how reliably the reviewer catches them, judged blind. The scores are published, the method is public, and when a review is wrong you can challenge it on the record. The full protocol is in the methodology.

  6. Built for what is coming

    Science already publishes faster than any person can read, and machine-written papers are about to multiply the flood. Human attention cannot scale with it. A reader that reads everything can. Pith is built for the literature that is arriving, not the one we grew up with.

  7. Open and yours to take

    Everything Pith knows is public and exportable: every review, every citation, every record signed and checkable. Nothing is locked in. If Pith vanished tomorrow, the record would still be there, readable and verifiable by anyone.

  8. What we do not claim

    Pith reviews are made by machines, and machines make mistakes; that is why the accuracy scores are published instead of hidden. A Pith review is not a substitute for expert judgment; it is the guarantee that at least one careful, accountable reading of every paper exists. Pith favors no field, no institution, and no author, and when it is wrong, the challenge you file becomes part of the record too.

The product is the page. Three examples, unedited:

Pith is neutral infrastructure for scientific truth. It has no editorial stake in any paper's outcome.