Claims

../

2026_04_30_kovac: A working note generated by ChatGPT 5.5 Pro and posted by Kovač claims that for random sign coefficients the ratio of real roots to log n has lower limit at most one over pi with positive probability; posted raw, no review.

2026_05_07_kwon_zou: Kwon and Zou claim that almost surely infinitely many degrees have no real root outside the unit interval, so the ratio of real roots to log n has lower limit one over pi almost surely; a note in a public repository, no review.

2026_07_21_sneiderman: Sneiderman, working with GPT-5.6 models, claims that the ratio of real roots to log n has lower limit one over pi almost surely, by a finite-prefix restart of the April cone-record note; a full claim on the site's tab, no comments.

2026_07_23_snyder: A Lean 4 theorem credited to Colin Snyder of Star Fleet Math states that the ratio of distinct real roots to log n does not converge almost surely to two over pi; linked by the formal-conjectures catalog, not audited by this corpus.

2026_08_01_an_lin: An and Lin, working with GPT 5.5 Pro, claim that almost surely infinitely many even degrees have every real zero in the unit interval, so the ratio of real roots to log n has lower limit one over pi; a claim on the site's tab.

2026_08_26_alexeev: A Lean 4 development in Boris Alexeev's repository, with Codex as formal author, proves that the ratio of distinct real roots to log n has lower limit one over pi almost surely for random signs; accepted on Lean built here.

---