Stockfish 16 release around 2024?

[Jouni]

Current progress from SF15 is +8 Elo. SF 14 and 15 have at least +30 Elo in release! Takes looong to get same progress.

[Modern Times]

Yes, obviously it gets harder and harder to make gains the stronger it becomes.

[Eelco de Groot]

Yes, obviously it gets harder and harder to make gains the stronger it becomes.

I have not followed any of the discussions but I would expect it to become more difficult to tune if with a new NNUE, it basically means you have a whole new chessprogram In the philosophy of Rebel and Tal programmers, the eval is the chessprogram. And it does not help if you then also have hybrid evals, sometimes not using the NNUE. On top of that the magnitude of the eval was until recently not standardized. Chessprogram tuning is basically mathematical optimization to the game of chess. But with NNUE like this it is not just optimizing a moving target, it is a legion of targets so to speak. Well, maybe it is more the artillerie that is moving with a new motor every few weeks, not the target. Maar "de beste stuurlui staan aan wal" as they say here in Holland.

[Paloma]

Yes, as we can see here:
https://nextchessmove.com/dev-builds

[Graham Banks]

Meanwhile, ShashChess 26 could well top Stockfish 15 in rating lists.

[Sopel]

Meanwhile, ShashChess 26 could well top Stockfish 15 in rating lists.

That's the minimum we should expect from an engine based on Stockfish 15. Though with +-25 error bounds everything can happen.

[Jouni]

I hope version 15.1 when they fix X/Y time control. Impossible? No. I have easy solution: When time control is 60 moves in 15 minutes use 14 sec for each move . This is better than losing with time or playing last moves in 0,03 sec.

[Magnum]

Current progress from SF15 is +8 Elo. SF 14 and 15 have at least +30 Elo in release! Takes looong to get same progress.

Donate 4x to 8x more hardware that we have to develop now and Stockfish 16 will be out soon.