A 2000+ rated puzzle.

I tried to solve the puzzle and I failed. I have though better things to do than using a chess engine to solve it. However it is my opinion that, if an engine fails to solve a puzzle in a reasonable time, then maybe the puzzle is too difficult; the chances of finding that moves in a game would be very small.

Originally posted by cotoi
I tried to solve the puzzle and I failed. I have though better things to do than using a chess engine to solve it. However it is my opinion that, if an engine fails to solve a puzzle in a reasonable time, then maybe the puzzle is too difficult; the chances of finding that moves in a game would be very small.

and thus its practical value, minimal.

Perhaps someone can tell my why an engine would have trouble solving this? And, for that matter, why a 1400 ranked player would have trouble with it? Qc5 is the obvious first choice and from there it doesn't appear to be too difficult to get a mate (even if two of black's moves aren't forced). What am i missing?

One question about this puzzle: Can you say how many moves this one is to get the winning position, or mate, whatever? 10 moves down the road? 5?

Originally posted by sonhouse
One question about this puzzle: Can you say how many moves this one is to get the winning position, or mate, whatever? 10 moves down the road? 5?

The question stated in the first posting was "White to move and win!" Not "mate in n moves" or anything. Just find a move that gives white a (somewhat) superiour position.

I would like very well now the best move, and the motivation for it. A program can certainly show a move that 'it' finds be the best. Perhaps it will not be the same as the best player in the world find being the best move. Neither of them may be right on this. Hence the interest for this particular puzzle.

What is the best move, btw?

Originally posted by Sever
Perhaps someone can tell my why an engine would have trouble solving this? And, for that matter, why a 1400 ranked player would have trouble with it? Qc5 is the obvious first choice and from there it doesn't appear to be too difficult to get a mate (even if two of black's moves aren't forced). What am i missing?

You're missing nothing since you say nothing. You just throw in a check. What is the follow-up?

I think that the idea is 1.Qc5+ Kg8 2.gxh7+ Kh8 3.Qd6 and now Black is in zugzwang. However the lines seem difficult to calculate. Let's say 3... a5 4.Bxa5 (zugzwang again) Qc8 5.Bd2 Nc7 6.Bc3. It looks like the solution would be to set up a sequence of zugzwang positions.

It was posted ages ago. (or one very similiar to it)

I actually found the winning idea but screwed up the play with the
Bishop and the a-pawn. If I recall I stumbled by not finding Ba1.

(then I did not know if I was on the right path or had fallen into
the composers trap)

Think it was Swiss G or Heinzkat who put me right.

Excellent study. (who by?? cannot remember)

Has a wee chuckle at posters who claimed it was unsound because their
box failed to find it. And saying it was unfair to post a puzzle a box could not solve.

Here is the solution - if you don't want to see it, then look away now.

Or do it a step at a time to see the first few moves.

Originally posted by cotoi
I don't see what is the purpose of posting an anti-engine problem.

It silences 80% of people who solve chesspuzzles on internetfora.

Excellent study indeed.I had seen it before and still couldn't solve it.Way over my head this one.

toet.

Fun exercise in brain stretching. Unfortunately, my elevator doesn't get anywhere near that floor. 😛

Originally posted by cotoi
I tried to solve the puzzle and I failed. I have though better things to do than using a chess engine to solve it. However it is my opinion that, if an engine fails to solve a puzzle in a reasonable time, then maybe the puzzle is too difficult; the chances of finding that moves in a game would be very small.

I'd have to disagree with that. Engines are programed in strange ways, and due to some programming glitches or other issues, they often cannot find mate in twos or win relatively elementary rook endings. So just because an engine can't find it, it's not necessarily too hard. Relating to this puzzle though, is there anyone that believes they could actually find it in a real game, OTB or otherwise?

Originally posted by Sever
Perhaps someone can tell my why an engine would have trouble solving this?

Top engines cannot afford to do a brute force search of all moves. It would limit their search depth too severly. So they try to use heuristics to dismiss/prune continuations that look irrelevant. Of course, these heuristics are indeed guidelines and hence are prone to exceptions.

One such heuristic is the "null move heuristic". This can backfire in positions where zugzwang plays a significant role, and this puzzle is of this type.

http://en.wikipedia.org/wiki/Null-move_heuristic

Originally posted by Varenka
Top engines cannot afford to do a brute force search of all moves. It would limit their search depth too severly. So they try to use heuristics to dismiss/prune continuations that look irrelevant. Of course, these heuristics are indeed guidelines and hence are prone to exceptions.

One such heuristic is the "null move heuristic". This can backfire in p ...[text shortened]... ant role, and this puzzle is of this type.

http://en.wikipedia.org/wiki/Null-move_heuristic

Thanks for the link. I'd heard "null-move" a million times but had never bothered to read up on it.

So far none of my engines have been able to crack it. Don't have Zappa Mexico - I did try SOS, which can be set on brute force enabled and null move disabled, but it was just taking forever. ChestUCI is cranking on it, currently working on move 14, so maybe it will find it any minute.

I saw the zugzwang with the Qc5+ move, and I thought the zugzwang was cool. Unfortunately, it doesn't go anywhere, just sort of peters out into a draw.

This was an interesting puzzle for me, both from trying to use my (defective) noodle first, then trying the engine crutches.

Originally posted by cotoi
You're missing nothing since you say nothing. You just throw in a check. What is the follow-up?

I was operating on the assumption that the solution posted by sbacat (which was the first good solution i reached before trying out different first moves) was the correct one.

Assuming that Greenpawn's solution is correct (and not a joke) then my question has been answered. As is usual in puzzles of all types, the devil was in the detail, or lack thereof, in the opening description. When Mariska said "white to move and win" what she should have said was "white to move and win in such a manner as no engine would and no sane person could!"

Cannot see anyone pulling this off in a game.

Some studies have been inspired from what happened in a game.

Possibly the most famous study The Saavedra came from an actual game
it was missed during the game and when the study was printed it was 'cooked'.

It's original task was White to play and draw.
Fernando Saavedra found the win.

White to play and win.