Seven axioms, and what follows from them.
One may assume, as every publisher before us has assumed, that a result becomes valuable only once it is enclosed. We do not assume it. Let the following be taken as given; the journal is merely the corollary.
Let K denote all knowledge. K is not partitionable by ability to pay.
A paywall is not a business model applied to knowledge; it is a claim about who knowledge is for. We reject the claim. Every article is free to read on the day it is published and every day after.
A proof is true or it is not.
The institution of its author is not a variable in the proof. Affiliation is collected for correspondence only and is withheld from referees. We publish schoolteachers, engineers and emeritus professors on identical terms.
Let the referee be named. Whence accountability.
Anonymity protects the reviewer from the author. Nothing protects the author from the reviewer. We invert this: reports publish with the paper, signed, including the ones that recommended rejection.
Machines may assist. They may not be concealed.
Authors declare machine assistance on a five-step meter; referees file their own independent assessment. Both figures publish. Where they diverge, the divergence publishes too — it is data about the instrument, not an accusation.
One may assume the reader is capable.
We therefore publish the full apparatus: data, code, the lemmas that did not survive, and the attempts that failed. Nothing is compressed for the convenience of a press release.
Negative results are results.
A closed door is a datum about the corridor. We hold a standing section for null results, disproofs, and counterexamples to conjectures the field would rather keep.
Whence: no author pays to be read, and no reader pays to read.
No submission charge, no article processing charge, no subscription, no embargo. The journal is funded by the labour of its editors and the servers of its signatories, and its accounts are published annually.