Computer vs Computer

Agree with both comments. I think it's funny that the analysis started by looking at just one game. The most recent competition between the two top engines is interesting because Black actually won more games, which now places a bigger onus on being able to prove that White has an advantage. And if it can't be proven for chess engines, probably even harder to prove for human play.

I've proven it before, at least to my satisfaction. I used two identical engines, identical settings, same opening book, etc. But I had to play a lot of games. I don't know if I kept the tournament results, though.

How did you get two engines to play each other, and what engine did you use?

I'm sure I used Arena, but many GUIs can be used. Just create two identical engines, play them against each other, then count up the White wins, Black wins, and draws.

You can use any fairly strong engine. I probably used some version of Stockfish.

Anyone interested in the very best of computer versus computer, there is a new king in town, and it is an AI called AlphaZero.

[I am puzzled about claims above about black doing well in a computer match. White has done better in every big enough sample I have seen, and the latest is no exception].

Thanks for info about Arena.

As for Black scoring better - it was just a small number of games (the recent Super-final of the chess.com engine tournament). I doubt Black actually has an advantage - but it does hi-lite that winning from Black is possible (even with best play from the latest engines).

The "Draw death" that some have claimed hasn't happened yet (chess is still a fair game, and winning from either side is very much possible).

i like playing h4 sometimes on move 1

oops wrong thread lmfao

I think that the computer plays better as Black because it sees White's previous move on turn 1 and can therefore defend better.

You can do this on the mobile app version.

You are using expert computers rated around 2000. That is nowhere near conclusive.

If you have a real chess engine, please use that.

The engines you used aren't strong and you used an online version which is even weaker as far as I remember. If you analyze the game with Stockfish 10 you'll probably see that both "experts" have made quite many mistakes (and blunders). The first move for istance is already an imprecision (what exactly is that move meant to be anyway?)

This doesn't prove that White has no advantage at all, because white threw that advantage on the first move. Just my two cents on this

I just started a new tournament in Arena. Stockfish 10 against an identical clone, alternating sides every other game. No opening book. (I'm seeing about 99 percent French Tarrasch openings, lol.)

In theory, I think I wouldn't even have to alternate sides, as long as the two engines are identical in every way.

I could use an opening book, but it would need to be a neutral book. Even so, the book might slightly change the winning percentage for White vs. Black.

I also could have used a weaker engine, but the winning percentage for White would slightly drop as the strength drops.

After 40 games, White has scored 5 wins, 34 draws, and 1 loss. (A 55% score, which is very close to the expected score, according to Wikipedia.)

After 400 games, White has scored 32 wins, 360 draws, and 8 losses. (A 53% score.)

I started a second tournament, with the only change being the two Stockfish 10 engines use the same opening book. I think I was wrong when I previously said that the book had to be a neutral book. In fact, I now think it's better if the book isn't neutral. Anyway, I used the HS-Book (a Harry Schnapp book from 2010).

After 200 games, White has scored 30 wins, 160 draws, and 10 losses. (A 55% score.)

As far as showing that White has an advantage, the strength of the engine isn't important. You just have to make sure that you play two identical copies of the same engine.

Would love to see Alpha Zero battling against strong computers!

Update on the second tournament (Time control = 40 moves in 4 minutes) -

After 300 games, White has scored 53 wins, 231 draws, and 16 losses. (A 56.2% score.)

An update to the update for the 2nd tourney:

After 400 games, White has scored 66 wins, 311 draws, and 23 losses. (A 55.4% score.)

...and I think this is as far as I'm going with this little exercise.

White scores about 55% in my computer vs. computer database.