TCEC 23 Superfinal started

[pohl4711]

TCEC 23 Superfinal is running!

https://tcec-chess.com

[CornfedForever]

TCEC 23 Superfinal is running!

https://tcec-chess.com

I moved the 10/27 version of Stockfish off my desired deep analysis engine. I can't remember exactly why, but do recall comparing it to another fairly recent version and it just did not seem quite as good. Of course, I could well be wrong.

[Jouni]

I don't expect much suspense. SF score 62.5% after 4 games.

[Jouni]

Now only 53.57% for SF. SF draws from 4.26 evaluation! Downscaling of evaluation really needed. There is already pull request.

[CornfedForever]

Now only 53.57% for SF. SF draws from 4.26 evaluation! Downscaling of evaluation really needed. There is already pull request.

Like I said...10/27 just smelled 'fishy'. I'm not sure other

I think you are referring specifically to the fortress of Rook vs Queen with many pawns game.

Perhaps the 'issue' may be more a product of over optimistic NNUE than anything...a non-NNUE engine would likely not have steered for some of the positions and kept play 'alive'.

[Jouni]

Stockfish wins game 46 after 18 moves .
[fen]r3k2r/pp1nbp2/1qp1pn1p/7P/3P1N1p/5P1B/PPPBQ2P/2KR3R b kq - 3 17[/fen]
Lc0 takes pawn d4 Stockfish 18. Ng6 +3,40 over. Why are they playing 3 hours unnecessarily?