Openings according to Alpha Zero

[CID]

Opinions ?

[MikeB]

Opinions ?

Let's put it this way - it appears Alpha Zero would be very good at creating an excellent opening book.

[CID]

What I understand from Deep Mind's papers:
1. 1.c4 it's the best
2. The best answer to 1.e4 is e5 (Ruy Lopez) and for 1.d4 is 1....Nf6
Good moves are too 1.d4 and 1...d5 as answer to 1.d4.
Am I wrong ?

[jdart]

This is not really what GM (including correspondence GM) practice is these days.

Ruy Lopez has declined in popularity, I think because the Marshall Gambit is drawish, if you are booked up on it. And the anti-Marshall systems are not very promising for White, either. But some of this may be changeable fashion and not bedrock truth.

--Jon

[Michael Sherwin]

This is not really what GM (including correspondence GM) practice is these days.

Ruy Lopez has declined in popularity, I think because the Marshall Gambit is drawish, if you are booked up on it. And the anti-Marshall systems are not very promising for White, either. But some of this may be changeable fashion and not bedrock truth.

--Jon

What about the Berlin Wall? That is enough to make me switch from 1.e4 if I wanted to win.

[jdart]

Yes, that is true: the Berlin is quite popular but I am not clear that is what AlphaZero would aim for. It is a relatively safe option for Black and has a good chance to be a draw but it is not as explored, and there are a bunch of complex resulting endgames that either side might have a chance to win.

--Jon

[Michael Sherwin]

I watch tournaments on YouTube like the Sinquefield Cup at the St. Louis Chess Club. Yasser and company talked about the Berlin defense almost every round. They complained heavily when there were so many of them because the games were dull and lifeless and resulted in all draws. Then when there were no berlins in a round they were elated because there was a greater chance for decisive games. If ever I play in a tournament again there are a couple of prepared lines to try. They'd never fool a top grandmaster but they might fool a lesser player. One is this.

1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.0-0 Nxe4 5.h3

This is about equal but if black tries to secure his extra pawn by 5. ... d6 then black will have to play very well to avoid a disadvantage.

[Ovyron]

This is not really what GM (including correspondence GM) practice is these days.

That's because they don't have access to A0. Since it showed on games positions that current engines claim are 0.00 and one side is actually winning, A0 could change the Correspondence game landscape because who knows if current games that were agreed to a draw were actually winning for one of the sides.

[Ponti]

Ruy Lopez has declined in popularity, I think because the Marshall Gambit is drawish, if you are booked up on it.

--Jon

Perhaps A0 doesn´t agree that Marshall Gambit is drawish...