How chess engines lie to you (and how you can stop falling for it) • page 1/1 • Community Blog Discussions • lichess.org

Comments on https://lichess.org/@/flash3s/blog/how-chess-engines-lie-to-you-and-how-you-can-stop-falling-for-it/JgyNCWH6

mkubecek

It's wrong to call it a lie. When Stockfish (with sufficient depth) evaluates a position as equal, all it means that it is confident it would hold a draw against (another instance of) itself and does not see a way to force a win against (another instance of) itself. The fact that a particular human player may not be able to hold a draw in such position does not change the fact that it's (almost certainly) objectively equal.

After all, there are equal positons where I fully believe Stockfish that they are equal but I'm still pretty sure I would lose them against sufficiently strong opponent for either side.

It's wrong to call it a *lie*. When Stockfish (with sufficient depth) evaluates a position as equal, all it means that it is confident **it** would hold a draw against (another instance of) itself and does not see a way to force a win against (another instance of) itself. The fact that a particular human player may not be able to hold a draw in such position does not change the fact that it's (almost certainly) objectively equal. After all, there are equal positons where I fully believe Stockfish that they are equal but I'm still pretty sure I would lose them against sufficiently strong opponent for either side.

flash3s

edited

@mkubecek said in #2:

It's wrong to call it a lie. When Stockfish (with sufficient depth) evaluates a position as equal, all it means that it is confident it would hold a draw against (another instance of) itself and does not see a way to force a win against (another instance of) itself. The fact that a particular human player may not be able to hold a draw in such position does not change the fact that it's (almost certainly) objectively equal.

After all, there are equal positons where I fully believe Stockfish that they are equal but I'm still pretty sure I would lose them against sufficiently strong opponent for either side.

That's a really good point. Will change the blog post title, as I agree with you that calling it a "lie" is indeed wrong. Thanks :)

@mkubecek said in #2: > It's wrong to call it a *lie*. When Stockfish (with sufficient depth) evaluates a position as equal, all it means that it is confident **it** would hold a draw against (another instance of) itself and does not see a way to force a win against (another instance of) itself. The fact that a particular human player may not be able to hold a draw in such position does not change the fact that it's (almost certainly) objectively equal. > > After all, there are equal positons where I fully believe Stockfish that they are equal but I'm still pretty sure I would lose them against sufficiently strong opponent for either side. That's a really good point. Will change the blog post title, as I agree with you that calling it a "lie" is indeed wrong. Thanks :)