5

As I understand it, current positional assessments by computers evaluate a position based on optimal play by both players from this point.

i.e. a trap moves that bait an opponent into making a mistake (even a really subtle one!) have no additional value because the computer assesses the position on the assumption that the opponent doesn't make that mistake.

That seems reasonable; playing in the hope that the opponent makes bad moves isn't strong tactical play.

But I assume that means that the converse is also true:

If I have a single winning line full of incredibly obscure counter-intuitive moves, that only a GM would have any hope of finding, and any deviation from that line leads to the whole thing immediately collapsing ... then that position is evaluated as winning.

Are there any evaluation programs/frameworks that attempt to account for that "you're ahead but only if you make a series of very specific moves" concept? Essentially "you're ahead, but you've made the position very unstable, and thus hard to play".

Improve this question
2
  • Several years ago, I think a similar question was discussed in rec.games.chess.computers (?), but it dealt with endgames: if a computer found itself in a theoretically lost endgame against a human player, was there any way to use endgame tables to go for highest degree of complexity (even if the complex path was shorter), and hope the opponent made a mistake along the way. Endgames would probably be easier to identify, and to analyze as to 'what made the player lose his way'. I don't recall that any definite factors were identified. W. Hartston suggested 'long moves' somewhere.
    – user30536
    Mar 28 at 15:20
  • The difficulty of the position depends on the strength of the player. One player may think that certain position is difficult, but the other is not. There are players who likes tactical game, and others like positional grinding. There are players who does not like to be attacked. There are players who are comfortable even if under attack. There are players who does not like to be pinned. If you know your opponents, you will do things to create situations where they do not like their position. You study games where they lose.
    – ferdy
    Mar 28 at 23:39

1 Answer 1

Reset to default
1

When it's hard for humans to find a very specific series of moves, it's simply because they lack strength in tactical aspect of chess. There is no such thing as "you're ahead, but you've made the position very unstable, and thus hard to play".

That being said, if I understand what you are saying, you want to understand when it's hard for humans to play. To find the answer, we need to understand how humans generally think in chess.

Humans mostly evaluate positions based on material imbalance and King safety. There are also other factors which may influence a human's understanding of a position, such as the amount of space being controlled, bishop pair, passed pawns/connected passed pawns, isolated pawns, doubled/tripled pawns, minor piece activity, how well the rooks are positioned, etc. When it's hard for humans to understand a position is when none (or barely any) of these imbalances are present despite one side being significantly better (based on computers). This situation arises when one side simply has no useful moves. In such positions, if the winning side fails to understand how helpless his opponent is, he may fail to punish his opponent. I believe this case relates to the strategical aspect of chess.

To finally answer your question, no, there are no tools that detect how hard it is for humans to properly evaluate a position (it would be very interesting to see this functionality by chess engines, though!)

In regard to tactical pattern difficulty, I'm not aware of any tool that has the capability to understand how hard the moves are to find for a human. Chess.com has an interesting algorithm that highlights hard-to-find moves with blue exclamation marks, but I don't know how relevant or helpful that would be for you, and we have no access to the code of this algorithm as chess.com is not open-source, so we can't know exactly how it works.

Improve this answer
2
  • "you're ahead, but you've made the position very unstable, and thus hard to play" This is correct from a game theory perspective, but when discussing human chess there are definitely easy and hard positions........... "When it's hard for humans to understand a position is when none (or barely any) of these imbalances are present" I disagree. The hardest positions are those where many of those factors are present at the same time, with some favoring one side and some favoring the other.
    – David
    Mar 30 at 9:56
  • @David those are valid arguments. I tend to analyze every online game I play, and I find that sometimes one side gains huge advantage when nothing apparent happened, which is the situation I described. I find that those are the positions I struggle though most to understand, because I fail to realize that the other side has no real moves. But of course, everything I said is my opinion, and I don't believe it's possible to present real facts here.
    – Emre Bener
    Mar 31 at 6:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.