Gambit Openings

Ed is energetically investigating which programs are good in gambit (openings in which one side sacrifices material in exchange for a good position) - link.

In a competition like this, for every win there must be a loss. It might be that the information is in the data provided and I have missed it, but my questions would be:

1. In strong human games, roughly what is the win/draw/lose percentage for the side that sacrifices material for position?

2. Same question again for weak and strong computers

Thanks for any enlightenment on this!

You are asking the wrong question. And the quality of the gambit. And the strength of the players. Chess is chess, and we like to name things. But it always boils down to the true evaluation of the position for computer chess.

1. For strong human players. Knowing the correct line of play is everything to refute a bad gambit. If you don't know the correct line to refute a gambit or a bad gambit. This will result in a loss for even the best players. That is why gambits are played. You are hoping the other human player does not know the correct line of play. We are playing for tricks.

2. Again for computers. The question comes down to the Elo difference of the programs. And the quality of the gambit. Again we are lumping all gambit together. The true evaluation is everything. Stockfish for example being the best computer play. Has the strength to draw, and outplay other engines. Because most engine are weaker.

3. Generally the best strongest engines, with the fastest tactics will perform better in gambit openings. This is why I was not shocked to see Lc0 perform poorly. As with Lc0, it only knows wins, losses and draws. And knows nothing of resistance. Meaning if it sees it is losing. It will not play moves to prolong the game. And can play stupidly in such positions.

For computer chess the quality of the gambits matter, and the Elo difference of the engines. For humans it comes down to if the human knows the gambit, and the correct line of play.

But not all gambits are of the same quality!

Well said @mwyoung, it is important to say that the gambit openings that Ed picked are pretty much equal evaluation at the end line, there are no dubious or crazy lines (which is fair because even the slightest difference will result in win or lose when it comes to engine matches)

Thank you both for the enlightening feedback!