Cerebellum says: play the French and semi-Slav

[Lyudmil Tsvetkov]

Cerebellum is arguably the strongest opening book ever. Its strength lies in the memorized best value score of a given move. If it has not seen that move before then it behaves like a normal Stockfish chess engine/book.

With absolute certainty, the unchanging first move of Cerebellum as white is e4 and its favorite variation is the Giuoco Piano. With black against e4, it plays the constant e5.

So, it is possible to find ways to neutralize by thorough preparation and accumulation of the best move sequence against these. This is both a strength by having a large collection of store best move and at the same time a weakness as other book makers could assimilate that large best move collection and then use it against Cerebellum with the injection of more best moves that has not been memorized.

In theory, chess games should end a draw with correct play. Good opening books makes it easier to play correctly. The best opening book actively maintained thus far is Cerebellum but may not sustain its perceived strength when somebody else will have the focus to create an alternative book that will neutralize it.

Unfortunately, the best reply on 1.e4 is not 1...e5, but 1...c5.
Also, the giuoco (goddam wops, how can they insert 3 vowels one after another ) piano is much inferior for white than the Ruy Lopez and a range of other sequences.

Biggest downside of this project though is that this book, no matter the hours spent analysing and the depth reached, will be easily beaten by any book created by some 200 elo stronger than current SF engine in ..... 5 minutes.

[Henk]

Engines are only ELO crunchers. They are not made to help you play better chess. Even if some chess openings would be clearly better they may not match your playing style.

[drj4759]

Yes, for me the best reply for e4 is Sicilian defense. That's what I always use when I play. But that is human judgement. According to Stockfish, the best defense for e4 is e5 with a probable score of 45 while sicilian is 32. That's a big marign. You can not argue with Stockfish when you try to challenge it and you play the sicilian. You will be smashed.

You are talking about the distant future where a chess engine will have 200 ELO advantage over Stockfish and beat Brain Fish/Cerebellum. That's daydreaming. The opening book that can neutralize Cerebellum is here. I invite you to visit the ChessOwl site later where I post the result of the ongoing tournament between Brain Fish/Cerebellum vs. Fauzi 2.2 and Fishbook Dynamic 161016. Partial results showed the overall superiority of Cerebellum but lags behind Fishbook.

See it for yourself. The opening book links will also be published.

[jdart]

Scores are subject to the horizon effect. And the farther they are from the tips of the tree (end of book lines), the more subject to error they are, because there are branches along the way that have not been visited.

--Jon

[clumma]

You have forgotten that the values of Cerebellum based on Stockfish only

I haven't forgotten. It's the strongest engine, especially at short time controls, so it seems an obvious choice.

It may be possible to do better by blending evaluations of different engines, say, by running them in multi-PV mode and summing scores (after calibrating) per-move. It would be an interesting experiment to try...

Moreover the idea of "perfect play" has not a well established definition.

Of course it does: perfect play is play that guarantees the game outcome won't be worse than the minimax game outcome of the position it started from.

-Carl

[clumma]

- move the project headquarters from Germany to the US, where electricity costs will be much lower;

Not necessary! Just use AWS, from anywhere in the world. Stockfish license lets us run it on as many machines as we like for free.

- start a donation subscription on this forum, with a minimal contribution requirement of 1000 bucks;

A Kickstarter could probably succeed. I would set the goal at $50,000. But it would require some computer chess people to cooperate. Fat chance!

-Carl

[clumma]

Cerebellum is arguably the strongest opening book ever. Its strength lies in the memorized best value score of a given move. If it has not seen that move before then it behaves like a normal Stockfish chess engine/book.
With absolute certainty, the unchanging first move of Cerebellum as white is e4 and its favorite variation is the Giuoco Piano. With black against e4, it plays the constant e5.

You may be thinking of Brainfish. Cerebellum demo is a database only, so it doesn't play anything, but it says the French is superior to the Italian for black.

The demo may be using a different version of the database than Brainfish.

-Carl

[clumma]

Even if some chess openings would be clearly better they may not match your playing style.

True! I've been thinking about this. An eval difference of 0.02 may not be a good reason to choose French over Italian -- even if the evals are accurate to 2 centipawns.

For humans, and maybe also computers, it may be more important that we can find the best moves later (or that our opponent cannot).

I'm not sure how to quantify this. We want to put our opponent in positions where several moves look equivalent but only one is actually good. We want to maximize the horizon effect for our opponent.

With something like Cerebellum, it would be interesting to know the revision in eval due to the backsolving. I wonder if Thomas has stored this in the database. Thomas?

-Carl

[corres]

[quote="clumma"]
Of course it does: perfect play is play that guarantees the game outcome won't be worse than the minimax game outcome of the position it started from.
l[/quote]

The main issue is that there is no man who knows the real value of starting position. Values displayed by engines reflect the opinion of developers only.
Their opinion is based on practical games without theoretical evidence.

[clumma]

You are talking about the distant future where a chess engine will have 200 ELO advantage over Stockfish and beat Brain Fish/Cerebellum. That's daydreaming.

Not at all. We've gained about 50 Elo/year from software improvement since Rybka 4. If that continues we will have perfect chess in < 8 years.

(Hardware improvement gave us 15 Elo/year from 2007-2013 but I don't think we've gotten much from it in the last three years.)

What's exciting is that cloud computing & backsolving may put us in striking range today. Computer chess people are wasting a lot of resources buying expensive computers and throwing out their calculations after every game! OK, CCRL and some other testers do make evals available (and somebody should throw them into a hash table) but much more could be done.

The opening book that can neutralize Cerebellum is here. I invite you to visit the ChessOwl site later where I post the result of the ongoing tournament between Brain Fish/Cerebellum vs. Fauzi 2.2 and Fishbook Dynamic 161016. Partial results showed the overall superiority of Cerebellum but lags behind Fishbook. See it for yourself. The opening book links will also be published.

Cool, I will check it out! But I should say I don't think beating Brainfish is very interesting. Of course it's possible to tune some book to beat some other book. I'm more interested in solving the game. Brainfish is a proof of concept that backsolving works, and it got the attention of computer chess people. I don't think it has any use beyond that.

-Carl