Mystery engine at CCC

[Eduard]

It will therefore not be easy for the new engine to get past Stockfish.

Torch has the advantage of being close source, chess.com is donating hardware to its development (something they never did for Stockfish or any other engine as far as I know) and top-notch developers that now won't contribute to Stockfish.
Wouldn't surprise me if influencers like Gotham and Hikaru start using Torch (when its weaker) just because chess.com "asked them" to advertise it.

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

Will Torch only be tested on Chess.com, or is there a chance that the engine will play in the TCEC tournament as well?

[Chessqueen]

It will therefore not be easy for the new engine to get past Stockfish.

Torch has the advantage of being close source, chess.com is donating hardware to its development (something they never did for Stockfish or any other engine as far as I know) and top-notch developers that now won't contribute to Stockfish.
Wouldn't surprise me if influencers like Gotham and Hikaru start using Torch (when its weaker) just because chess.com "asked them" to advertise it.

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

Will Torch only be tested on Chess.com, or is there a chance that the engine will play in the TCEC tournament as well?

I have a feeling that if Stockfish play 100 games against Mystery, it will beat Mystery by at least 55 games to 45 ==>https://www.chess.com/de/computer-chess-championship

[lkaufman]

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

What will become of komodo- Dragon, Larry?
Komodochess.com site doesn't allow log in anymore, does it?

We are still encouraged to improve Dragon for a possible final release of a Dragon 3.2 upgrade, but with minimal resources available and most attention on Torch it isn't looking very likely, unless some Torch idea is easily transferable to Dragon. A final upgrade would remove any obligation for chess.com to offer partial refunds for subscriptions, so they are in favor of it, but so far we don't have a meaningful upgrade to offer.

[lkaufman]

It will therefore not be easy for the new engine to get past Stockfish.

Torch has the advantage of being close source, chess.com is donating hardware to its development (something they never did for Stockfish or any other engine as far as I know) and top-notch developers that now won't contribute to Stockfish.
Wouldn't surprise me if influencers like Gotham and Hikaru start using Torch (when its weaker) just because chess.com "asked them" to advertise it.

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

Will Torch only be tested on Chess.com, or is there a chance that the engine will play in the TCEC tournament as well?

I have a feeling that if Stockfish play 100 games against Mystery, it will beat Mystery by at least 55 games to 45 ==>https://www.chess.com/de/computer-chess-championship

This statement has no meaning without specifying the opening book, and perhaps also time control and threads. With UHO openings I am sure you are correct, in fact it is surely an underestimate. With normal openings (meaning roughly par advantage for White as in normal serious GM play) and good hardware and Rapid time controls it would be 50 to 50 because all 100 games would be drawn. Maybe with normal openings but typical home pc hardware and blitz (2'1") tc your estimate might be reasonable (probably 10 to 0 with 90 draws or so). These details are crucial now.

[Chessqueen]

It will therefore not be easy for the new engine to get past Stockfish.

Torch has the advantage of being close source, chess.com is donating hardware to its development (something they never did for Stockfish or any other engine as far as I know) and top-notch developers that now won't contribute to Stockfish.
Wouldn't surprise me if influencers like Gotham and Hikaru start using Torch (when its weaker) just because chess.com "asked them" to advertise it.

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

Will Torch only be tested on Chess.com, or is there a chance that the engine will play in the TCEC tournament as well?

I have a feeling that if Stockfish play 100 games against Mystery, it will beat Mystery by at least 55 games to 45 ==>https://www.chess.com/de/computer-chess-championship

This statement has no meaning without specifying the opening book, and perhaps also time control and threads. With UHO openings I am sure you are correct, in fact it is surely an underestimate. With normal openings (meaning roughly par advantage for White as in normal serious GM play) and good hardware and Rapid time controls it would be 50 to 50 because all 100 games would be drawn. Maybe with normal openings but typical home pc hardware and blitz (2'1") tc your estimate might be reasonable (probably 10 to 0 with 90 draws or so). These details are crucial now.

A good hardware is considered to be in this range of Ryzen 7 to Ryzen 9 , but NOT too many gamers would be able to afford this type of PC

[ImNotStockfish]

It will therefore not be easy for the new engine to get past Stockfish.

Torch has the advantage of being close source, chess.com is donating hardware to its development (something they never did for Stockfish or any other engine as far as I know) and top-notch developers that now won't contribute to Stockfish.
Wouldn't surprise me if influencers like Gotham and Hikaru start using Torch (when its weaker) just because chess.com "asked them" to advertise it.

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

Maybe I should've been more specific and say "any other engine not totally or partially owned by them".

[tmokonen]

So what happens now with Ethereal, Berserk and Koivisto? Are all these projects done? High level computer chess competitions will be less interesting with less worthy opponents if they are done.

[Chessqueen]

So what happens now with Ethereal, Berserk and Koivisto? Are all these projects done? High level computer chess competitions will be less interesting with less worthy opponents if they are done.

Each of these programmers acquired more knowledge from the other programmers and probably continue developing their engines even if they stop making it available for us.

[Chessqueen]

Chess.com did donate hardware for Komodo Dragon development as part of our partnership from 2018 thru 2022. They also made some of their employees available to help us out as needed on a part-time basis, notably Dmitry Pervov who wrote the NNUE code we used.

What will become of komodo- Dragon, Larry?
Komodochess.com site doesn't allow log in anymore, does it?

We are still encouraged to improve Dragon for a possible final release of a Dragon 3.2 upgrade, but with minimal resources available and most attention on Torch it isn't looking very likely, unless some Torch idea is easily transferable to Dragon. A final upgrade would remove any obligation for chess.com to offer partial refunds for subscriptions, so they are in favor of it, but so far we don't have a meaningful upgrade to offer.

Now that the Torch project is under development, I finally have hope that if I live another 40 years I might be able to see Chess being solved in my lifetime
NOTE: What we know for sure is that So far, chess is solved when there are 7 or fewer pieces (including kings) on the board.
Each additional piece makes things much more difficult.
https://www.chess.com/de/computer-chess-championship

[AndrewGrant]

I continue to work very passively on Ethereal. I have boxes still generating training data for months now. Very hard to training billions of positions without reusing Stockfish + Leela data, which is very lame imo!

As for search improvements and such, I'll try some things now and again. But mostly things that are already present in other engines I'm afraid. Any original ideas I have would go into Torch. Which means that future Ethereal is a bit bland in that regard. But at least I'm one of few using my own data + trainer, so I have that going for me.