Forums

Who wins from this (Puzzle)?

Sort:
Can anyone let me know who wins from the following position (or is it a draw)?
[White to move]
Clue: This puzzle is beyond the capability of a human to solve. But if you have a guess let me know. I'll reveal the answer after a few people have guessed.
I asked my dog and she said take the queen on H1
The queen on h1 is the first piece to be taken. Speaking of dogs, check out my new "Bulldog" chess piece I'm working on (2 versions).
I'll tell you the answer to the puzzle after one or two more people make a guess

It depends on whether or not the 50-move rule applies in the context of the puzzle. If it does, then the result [with best play] will be a draw after the obvious 1.Qxh1 is played. On the other hand, if the 50-move rule is being disregarded in this situation, then the endgame after Qxh1 is VERY far from trivial - so deep, in fact, that the assistance of Lomonosov tablebases would probably be required for a definitive solution.

Black wins
I played my dad and I had a rook and a knight and I won
MSurkont gets the prize for best answer from another species, cobra91 for best technical answer, and foofooes for best attempt at actual play.
To see the answer play the solution here:
As cobra91 says, with best play white wins but it takes more than 50 moves. In fact it takes 546 moves!
It's the longest known forced checkmate in a chess puzzle. I did get the puzzle from the Lomonosov 7-piece tablebases. I added the extra queen at the start to make it an 8 piece puzzle (so if you plan to use the tablebase to solve you at least need to count the pieces and remove the easy first capture).
Thanks to all who guessed or considered the puzzle!

wow

In this puzzle, all moves are "perfect" - both sides make the best possible move to checkmate the other, or if not possible to survive as long as possible.

I should say thanks to the Lomonosov website ('http://tb7.chessok.com/probe) for making the .pgn available

I was curious to see what a SF analysis would say about this perfect play. Adding together the two sides, SF says there were 1037 excellent moves, 12 good moves, 29 inaccurate moves, and one blunder. At 535 it says there was a "faster mate in 26" (white to mate at 561). But best play actually finishes at 546 (15 moves faster)!

Btw, if anyone wants to join me in some variant-chess games, come visit us at one of the variant threads at chess.com. I've played a few games of Bulldog Chess (example here).

I'm also trying to find someone to play me in Amsterdam Medieval Chess. Rules are (here). It is a training exercise for a chess game played on a 10 x 10 board!

Don't worry - I won't calculate a checkmate from 546 moves away!
vickalan wrote:
MSurkont gets the prize for best answer from another species, cobra91 for best technical answer, and foofooes for best attempt at actual play.
To see the answer play the solution here:

As cobra91 says, with best play white wins but it takes more than 50 moves. In fact it takes 546 moves!
It's the longest known forced checkmate in a chess puzzle. I did get the puzzle from the Lomonosov 7-piece tablebases. I added the extra queen at the start to make it an 8 piece puzzle (so if you plan to use the tablebase to solve you at least need to count the pieces and remove the easy first capture).
Thanks to all who guessed or considered the puzzle!

I saw all that...

@-@ Seriously though, 546...

Holy cow!

Thanks for that amazing endgame, vickalan!

Good work

vickalan wrote:
As cobra91 says, with best play white wins but it takes more than 50 moves. In fact it takes 546 moves!
It's the longest known forced checkmate in a chess puzzle. I did get the puzzle from the Lomonosov 7-piece tablebases. I added the extra queen at the start to make it an 8 piece puzzle (so if you plan to use the tablebase to solve you at least need to count the pieces and remove the easy first capture).
Thanks to all who guessed or considered the puzzle!

Heh... based on what little theoretical knowledge I have concerning QN vs. RBN endings, I suspected White may be winning (without the 50-move rule) after some inordinate number of moves. Without tablebase analysis, though, I could never have confirmed that suspicion. Such endings are well beyond the scope of what humans can currently understand.

vickalan wrote:
Btw, if anyone wants to join me in some variant-chess games, come visit us at one of the variant threads at chess.com. I've played a few games of Bulldog Chess (example here).

I'm also trying to find someone to play me in Amsterdam Medieval Chess. Rules are (here). It is a training exercise for a chess game played on a 10 x 10 board!

Don't worry - I won't calculate a checkmate from 546 moves away!

As far as chess variants go, I'm usually happy to "trade" games. When no one is willing to try the variants I find most fun/interesting, however, I am much less inclined to play a new and randomly chosen game which I'm not particularly enthusiastic about.

#546 move rule

No problem urk, glad you studied it.
I guess a caveman can make as much from this puzzle as anyone!

cobra91: do you have a favorite variant? I'm already playing a few, but if there's one we both like maybe a game in the future?

Bowlinggreen: the #546 move rule? That's a great point. There's legitimate chess endings (at least theoretical) where one side can win, but if the 50 move rule stops the game, the "winning" side only gets a draw.
In real play, I wonder if the 50-move rule ever stopped a good game with one side having a winning strategy (rather than just pointless moving around of pieces). This 546 puzzle shows it's possible, but not sure if it has ever happened?

it probably  never happened

the most moves in a game is probably in the 300s

cobra91 wrote:
Heh... based on what little theoretical knowledge I have concerning QN vs. RBN endings, I suspected White may be winning (without the 50-move rule) after some inordinate number of moves. Without tablebase analysis, though, I could never have confirmed that suspicion. Such endings are well beyond the scope of what humans can currently understand.

I don't know anything about QN vs. RBN endings, but by points white is ahead at the start (12 vs. 11). White stays ahead all the way to move 514 where Black makes the first capture. But then white's King of all pieces captures the black rook! White stays ahead with this material for the next 30 moves, until 545 where white finishes off Black and puts him in check.

Lesson: Not sure if always true but in this case the points system of scoring works. White starts ahead in points, and wins the game!
vickalan wrote:
cobra91 wrote:
Heh... based on what little theoretical knowledge I have concerning QN vs. RBN endings, I suspected White may be winning (without the 50-move rule) after some inordinate number of moves. Without tablebase analysis, though, I could never have confirmed that suspicion. Such endings are well beyond the scope of what humans can currently understand.

I don't know anything about QN vs. RBN endings, but by points white is ahead at the start (12 vs. 11). White stays ahead all the way to move 514 where Black makes the first capture. But then white's King of all pieces captures the black rook! White stays ahead with this material for the next 30 moves, until 545 where white finishes off Black and puts him in check.

Lesson: Not sure if always true but in this case the points system of scoring works. White starts ahead in points, and wins the game!

Interesting take!   However, I do think it's also important to understand that the "stronger" side (White, in this case) walks a razor-thin tightrope with QN vs. RBN. In a lot of the positions that arise along the narrow path to victory, many of the wrong moves don't even just draw - they lose!!   In fact, I'd be willing to bet that a tablebase-assisted player could defeat any modern chess engine in such endings, even if forced to play the "weaker" side, and even if the program was backed up by the best available hardware.

Since you asked whether I have a favorite chess variant, the most accurate answer would be that I have a bunch of favorites. The ones that immediately come to mind are Chess with Different Armies (Betza), Ultima (Abbott), Grasshopper Chess (Boyer), War of Mates (unknown), and Retroduels (unknown). However, most games found within the CVP archive (chessvariants.org) are of very high quality. Spartan Chess (Streetman), for instance, is yet another exceptional game which is also fairly well known. Cetran Chess 2 (Cetina), meanwhile, is an outstanding but quite obscure example (and another of my personal favorites).

cobra91 wrote:
Interesting take!   However, I do think it's also important to understand that the "stronger" side (White, in this case) walks a razor-thin tightrope with QN vs. RBN. In a lot of the positions that arise along the narrow path to victory, many of the wrong moves don't even just draw - they lose!!   In fact, I'd be willing to bet that a tablebase-assisted player could defeat any modern chess engine in such endings, even if forced to play the "weaker" side, and even if the program was backed up by the best available hardware.

Since you asked whether I have a favorite chess variant, the most accurate answer would be that I have a bunch of favorites. The ones that immediately come to mind are Chess with Different Armies (Betza), Ultima (Abbott), Grasshopper Chess (Boyer), War of Mates (unknown), and Retroduels (unknown). However, most games found within the CVP archive (chessvariants.org) are of very high quality. Spartan Chess (Streetman), for instance, is yet another exceptional game which is also fairly well known. Cetran Chess 2 (Cetina), meanwhile, is an outstanding but quite obscure example (and another of my personal favorites).

That is really intersting too! And it's probably one of the reasons this endgame goes on for 546 moves. I'm sure you're right that a tablebase-assisted player could beat any chess engine in this ending (unless the engine itself had access to a tablebase).

Thanks for your analysis, I didn't know for sure if any wrong moves can actually lead to a loss. Since most moves look like guess-work to me, I probably would have 500 chances of losing if I played this endgame.

Btw, I read a little about some of the variants you listed. One game has a piece called the "Forfnifurlrurking"? (I copied that to make sure I spelled it right). Really??? Well except for the piece name(s), some of those games have interesting pieces!