Stockfish 120324 a Disaster

[AndrewGrant]

Hello Andrew!

I would like to take this opportunity to congratulate you on your engine Ethereal! On PlayChess the engine cannot be defeated when playing with normal books. Some other engines (if it wasn't fake, which is not certain on PlayChess) also play surprisingly well, Berserk for example. Stockfish's lead is getting smaller and smaller. I know that you are currently working on Torch. I wish you good luck! I really hope that this engine will be available to everyone at some point.

Stockfish's lead is actually growing by the day! You might be out of touch with the times! I would suggest peeping some of the data out there. Thank you for your kind words on Ethereal.

[smatovic]

Oh boy, we removed the Cloudflare check, people can access TC again, and here we go with beef again must be Friday, elections are gonna be fun...

--
Srdja

[Uri Blass]

Hello Andrew!

I would like to take this opportunity to congratulate you on your engine Ethereal! On PlayChess the engine cannot be defeated when playing with normal books. Some other engines (if it wasn't fake, which is not certain on PlayChess) also play surprisingly well, Berserk for example. Stockfish's lead is getting smaller and smaller. I know that you are currently working on Torch. I wish you good luck! I really hope that this engine will be available to everyone at some point.

Stockfish's lead is actually growing by the day! You might be out of touch with the times! I would suggest peeping some of the data out there. Thank you for your kind words on Ethereal.

I think it is dependent on the conditions.
Eduard post about playing with normal book and not about tcec conditions.

Do you have an evidence that stockfish's lead is actually growing with normal book?
I remember that even CCRL started to use slightly biased book.

If I look at the ssdf rating list I see that lc0 is leading
https://ssdf.bosjo.net/list.htm

[Eduard]

Great, Lc0 is leading at SSDF!

I created a version with changed parameters for the small network (Big-SF 110324) and this version clearly beats SF dev with my test book with variants that we play on servers in practice. I prefer to use this version because it is clearly better for analysis. Download on my homepage. By the way, it work network with Dimension 3072.

A new 3072 network is currently being tested on Fishtest today. And a next version to test is coming soon.

https://tests.stockfishchess.org/nns

I compiled two SF test versions with this new 3072 network today. One is the one that is currently being tested on Fishtest against SF dev 2560 and one is with changed parameters for the small network. The rest identical. Download also on my homepage for anyone who would like to test it.

I am not sure which 3072 network is better. I will test it.

[lucametehau]

Hello Andrew!

I would like to take this opportunity to congratulate you on your engine Ethereal! On PlayChess the engine cannot be defeated when playing with normal books. Some other engines (if it wasn't fake, which is not certain on PlayChess) also play surprisingly well, Berserk for example. Stockfish's lead is getting smaller and smaller. I know that you are currently working on Torch. I wish you good luck! I really hope that this engine will be available to everyone at some point.

Stockfish's lead is actually growing by the day! You might be out of touch with the times! I would suggest peeping some of the data out there. Thank you for your kind words on Ethereal.

I think it is dependent on the conditions.
Eduard post about playing with normal book and not about tcec conditions.

Do you have an evidence that stockfish's lead is actually growing with normal book?
I remember that even CCRL started to use slightly biased book.

If I look at the ssdf rating list I see that lc0 is leading
https://ssdf.bosjo.net/list.htm

Pretty irelevant rating list. Also do you know about error bars?

[Eduard]

And why is it wrong to say "Lc0 leads"?

Are you still allowed to write negatives about Stockfish? You're not allowed! I have no idea what else I can write if it doesn't match the SF formulas and the tests with thousands UHO openings. Then everything is irrelevant!

[Uri Blass]

Hello Andrew!

I would like to take this opportunity to congratulate you on your engine Ethereal! On PlayChess the engine cannot be defeated when playing with normal books. Some other engines (if it wasn't fake, which is not certain on PlayChess) also play surprisingly well, Berserk for example. Stockfish's lead is getting smaller and smaller. I know that you are currently working on Torch. I wish you good luck! I really hope that this engine will be available to everyone at some point.

Stockfish's lead is actually growing by the day! You might be out of touch with the times! I would suggest peeping some of the data out there. Thank you for your kind words on Ethereal.

I think it is dependent on the conditions.
Eduard post about playing with normal book and not about tcec conditions.

Do you have an evidence that stockfish's lead is actually growing with normal book?
I remember that even CCRL started to use slightly biased book.

If I look at the ssdf rating list I see that lc0 is leading
https://ssdf.bosjo.net/list.htm

Pretty irelevant rating list. Also do you know about error bars?

I know about error bars.
I did not claim that lc0 is the best but only that lc0 is leading the ssdf list.

The ssdf rating list is clearly relevant for people who are interested in results in chess and not in biased chess.
In other lists people push moves to the engine that the engine is not going to play in normal games.

I would like to know what is the highest level that it is possible to beat with no book and I would like to see some correspondence tournament when every participants need to beat some constant version of stockfish with no book when the winner is the person who beat stockfish with the biggest number of nodes per move. opponent is deterministic stockfish with 1 core and no tablebases so it is going to be easy to reproduce the games.

[Graham Banks]

I remember that even CCRL started to use slightly biased book.

No.

[Uri Blass]

I remember that even CCRL started to use slightly biased book.

No.

I remembered something like that.

Maybe my mistake is because of reading the following thread when larry kaufman claimed that the few decisive games are from opening that today are not considered balanced.

viewtopic.php?p=950339&sid=6250f0b1dba5 ... 91#p950339

Maybe I am wrong but it seemed to me that I read somewhere that you decided to allow book exit of more than 0.5 but not book exit of more than 0.7 or something like that.

[Dann Corbit]

I guess that engine writers will train for imbalanced openings more and more.
That way, they can do well in the current contests.