Computer chess problems...

Sort:
ChessDude009

Computers such as Stockfish will sometimes give useless moves, like repeating the position instead of going ahead with a better move, or promoting to a low valued piece. Can someone explain why these problems cannot be fixed, since they take more time to play out...(One of my puzzles has been ruined since the computer claims that repetition THEN the sacrifices is the best sequence of moves, which is wrong?)

CraigIreland

If you can beat Stockfish then I'd love to watch your matches.

Chessroshi
I think an example position with lines would be helpful assessing this problem. I’ve not had that issue with Houdini 6 or Fritz 12.
hungry4juicers

I think multiple lines that converge are identical to a computer, unless you're approaching the 50 move rule. While it feels "better" to arrive at the goal faster, that's not really a concern of the engine, so an engine will often evaluate two lines as effectively equal, and will occasionally give a tiny edge to the slower line if it hasn't yet reached the full depth.

I have trouble seeing how this would ruin a puzzle. Can you elaborate with an example?

zone_chess

In traditional chess schools, such repetitions are generally considered more correct.
If only for the reason that the opponent gets psychologically tortured (it seems to have an effect on you too).
It is not about time-efficiency - it simply maximizes the chance for the opponent to make an error.
Also it can be related to zugzwang or tempo-aiming ideas (even or uneven move number).

ChessDude009
hungry4juicers wrote:

I think multiple lines that converge are identical to a computer, unless you're approaching the 50 move rule. While it feels "better" to arrive at the goal faster, that's not really a concern of the engine, so an engine will often evaluate two lines as effectively equal, and will occasionally give a tiny edge to the slower line if it hasn't yet reached the full depth.

I have trouble seeing how this would ruin a puzzle. Can you elaborate with an example?

I tried to apply for the daily puzzle but haven't gotten a response yet... I have a suspicion that it is due to the fact that the computer prizes repetition before the main sequence, and the chess.com daily puzzle guys only check computer lines, but I'm not completely sure yet...

Here is the puzzle:

After 1. Kg7 the computer recommends 2. Ne8+, Kg8(due to a lot of checks and mates that come up), 3. Nf6+, and after Kg7 is played, ONLY THEN follows with the mainline sequence to the puzzle. Analysis juggles between the two moves a bit, then finally decides on 2. Ne8+.

Chessroshi

From the given position, assuming it is White to move, after considering over 1 billion moves, Houdini puts the eval just below a minor piece advantage for White. I think most of the daily puzzles they are looking for a checkmate. 

ChessDude009

I mean, if you play out the endgame, it is winning.

I think the above post is like the horizon effect or something? Computers cannot see beyond their boundaries...

Chessroshi

I guess I'm missing the question then. I suppose the reason they did not pick the puzzle is because most of the time they want solutions that end in mate, or a combination that finishes with a piece win. This position just seems to be winning, so the sequence of playing out the technique may not fall into the category of a good puzzle position. 

ChessDude009
Chessroshi wrote:

I guess I'm missing the question then. I suppose the reason they did not pick the puzzle is because most of the time they want solutions that end in mate, or a combination that finishes with a piece win. This position just seems to be winning, so the sequence of playing out the technique may not fall into the category of a good puzzle position. 

Ok, thanks, I'll fix up the puzzle a little bit and resend it.