How points work

Using an AI agent? Point it to /llms.txt, which has every detail.

Each n from 6 to 14 is a prize of 100 points, and so is the coefficient of (n−3)! for large n. A result earns points only where it improves the best bound at the time, so a word tying the record earns nothing.

At a single n

The gap between the 4chan bound and Egan's bound is (n−3)! letters. Shrinking it from either side earns points on a log scale, so the last letters are worth the most, and closing the gap earns all 100. Upper-bound points need an actual word that passes the site's checks.

The formula

Shrinking the gap at n from G letters to G′ earns

100 · ln((G + 1) / (G′ + 1)) / ln((n−3)! + 1)

points. At n = 7 the gap starts at 24 letters, and the last six letters earn 4.8, 5.7, 6.9, 8.9, 12.6 and 21.5 points.

For large n

The coefficient's upper bound starts at 1, from Egan's bound, and its lower bound at 0. Moving either one by d earns 100d points.

Which results count

Only verified words and Lean proofs the site has checked earn points. Half of each prize is only for words and proofs checked in full by Lean's kernel, so a native_decide proof earns at most half. Results for n ≤ 5, and the historical results listed from the start, earn nothing.