This post is inspired by some of the comments in the "Analysis program woes". The position below comes from a game played on another site where engine use is permitted. The purpose of the post is to show that the right combination of engine and software (and a certain amount of box use skill) can tell you something about a position that is not limited to just a number.
Black to move.
The combination of engine and software used gives four moves with a dead zero score (it reckons this is a draw) but favours one of those moves very strongly. Black know it prefers one of these moves because more time is spent analysing the consequences of that move, there are more positions in the tree after that move than any other move and the box has made that move the main line.
Which move do you think the box likes and why?
I'm sorry I don't understand why it favours one move if it has given them all the same score.
From my point of view I would have thought that Qb5 seems to win the pawn back after say 2.Ng3 Qb1+
In the Bxe2 2.Kxe2 Qxd4 line black has to be very careful about his b pawn. I cannot then see if either side has a breakthrough without getting a board out to check and at the moment I don't have the time.
Originally posted by RagwortBc4 gets the pawn back with a superior position for black if white trades queens
I'm sorry I don't understand why it favours one move if it has given them all the same score.
From my point of view I would have thought that Qb5 seems to win the pawn back after say 2.Ng3 Qb1+
In the Bxe2 2.Kxe2 Qxd4 line black has to be very careful about his b pawn. I cannot then see if either side has a breakthrough without getting a board out to check and at the moment I don't have the time.
that leaves .. Qa5 or .. Qb5. The more difficult seems .. Qa5.
.. Qa5 attacks three pawns, one is even bound. If Qd2, it gives a cute symmetrical square, with both queens guarding/attacking three pawns at same time. Then .. Bd4 and pawn is lost.
Instead of Qd2 maybe N to somewhere makes it all complicated...
Sorry, got no real analysis. Is difficult position if thinking beyond a draw
Originally posted by tharkeshIf (that word again) both players don't do anything stupid all four moves lead to a draw. You are not looking for the best move here as I realised after a while, but there is a perfectly good reason to favour the move the software wants to play over the other three. The worrying thing is that I would play the move (in fact I did play it) for the same reason.
that leaves .. Qa5 or .. Qb5. The more difficult seems .. Qa5.
.. Qa5 attacks three pawns, one is even bound. If Qd2, it gives a cute symmetrical square, with both queens guarding/attacking three pawns at same time. Then .. Bd4 and pawn is lost.
Instead of Qd2 maybe N to somewhere makes it all complicated...
Sorry, got no real analysis. Is difficult position if thinking beyond a draw
Originally posted by tharkeshor ... Qc4
that leaves .. Qa5 or .. Qb5. The more difficult seems .. Qa5.
.. Qa5 attacks three pawns, one is even bound. If Qd2, it gives a cute symmetrical square, with both queens guarding/attacking three pawns at same time. Then .. Bd4 and pawn is lost.
Instead of Qd2 maybe N to somewhere makes it all complicated...
Sorry, got no real analysis. Is difficult position if thinking beyond a draw
followed by Qh4+ winning the h-pawn
would an engine give that extra weight?
...edit .... that makes 5 moves so I guess that fails miseraably!!
Originally posted by wolfgang59It's not the engine that is giving the move extra weight, it is the software the engine is running inside. If I run the engine using infinite analysis I don't get exactly zero for any of the moves. Use this particular software and I get exactly zero, 0.00, for four moves. That is odd in itself, and is what attracted my attention in the first place. Engines just don't see draws unless they are forced stalemates or repetitions. Now imagine that happening no matter what engine you use in the software, and every engine results in the same move being preferred. It's just weird.
or ... Qc4
followed by Qh4+ winning the h-pawn
would an engine give that extra weight?
...edit .... that makes 5 moves so I guess that fails miseraably!!
Originally posted by atticus2The actual move preferred is Qb5. The reason? White has three safe moves after that (Qd2, Ng1, Qd1) and a whole host of ways to get mated. The software (not the engine, it doesn't matter what engine I use) seems to be choosing the move that gives white the greatest opportunity to cock up. I have now seen this happen three times in the past two weeks.
1...Qc4 (hitting e2); if 2. Qxc4 Bxc4 wins d5 or a2 pawn (...Qh4 is not relevant). In the ending, Black could be better.
If 2. Qe3, it's back to the line(s) I gave before (after ...Qxa2)