8Chess 1.0.10 — AGPL engine derived from Reckless, own 1024-wide NNUE (source, binaries, PGNs)

[hemeshk]

8Chess is not small but however Graham is always like made from scratch ask the devs from sf 1 that why it is not from scratch and glaurung fork sf is powerfull but in starting it had glaurung fork.

[Ozymandias]

Then Stockfish 1 is Glaurung fork

Man, why are you getting so worked up over this little engine-8Chess? You’re talking about Stockfish, Reckless, Owen... that’s history-prehistory, even! AI is picking up speed... Triumviratus, Coda & Co. are the future. Look at that PHP vulnerability-it was discovered by AI! People are obsessed with other things: resources, money, wars, and so on. The science we preach is only an approximation-just a tiny bubble within fundamental understanding. I wonder what they’ll do with their money in the afterlife?!? And here’s a dilemma: which came first-the chicken or the egg? I bet AI will solve that, too!

The egg containing the first creature to be genetically considered a chicken came from a creature that wasn't 100% genetically a chicken.

[Ozymandias]

Self-reported result (not independently verified)
1,000 games vs Stockfish 19, TC 1.0+0.01, no opening book, direct UCI, Rust referee.
+487 =381 -132, 67.75%, +129 ±17 Elo. PGNs: link to results/.
I'm not asking anyone to take that at face value — it's why I'm posting here. Conditions, command lines, hardware and PGNs are public; I'll send binaries to anyone who wants to re-run a shorter match.

About -20 Elo vs SF 20261017-1125. 6000 games test at about 10 seconds per side and game. -20 Elo compared to Reckless 0.90 result under the same conditions. Reckless was +25 Elo over SF, 8chess is only +5 Elo. Openings based on SALCv5. SF had a time disadvantage in both cases to bring them closer.

[hemeshk]

did you test it vs sf 19?

[Ozymandias]

did you test it vs sf 19?

No