Chess position computers cannot solve

  • Thread starter david2
  • Start date
  • #1
19
69
there was a post here on PF about a chess position created by mathematics professor Sir Roger Penrose which computers could not solve. unfortunately the thread was locked so I made a new thread.

< Mentor Note -- this thread is different, so the re-post is okay >

it was the following position.

Naamloos.jpg

white to move and make a draw.


although computers give a huge evaluation for black they will defend perfectly with white and eventualy make a draw. so they actually do ''solve'' the position even if they think black is winning.


if instead you feed the computer the following position, it will choose the wrong plan and actually lose.

humans are able to make a draw.

can you see the right plan?

Naamloos2.jpg


white to move and make a draw.
 

Attachments

  • Naamloos.jpg
    Naamloos.jpg
    121.2 KB · Views: 353
  • Naamloos2.jpg
    Naamloos2.jpg
    134.3 KB · Views: 439
Last edited by a moderator:

Answers and Replies

  • #2
12,821
6,696
Is the reason the computer has a problem due to piece points? One move is better strategically but some other move gains more points.

I remember one time winning a chess game against an experienced player after having lost 50+ games previously by sacrificing my queen. The other player took the bait and a checkmate opened up shortly thereafter.

He couldn't believe what just happened and this was the one game he forgot to record. In truth, I didn't see the checkmate before I made my move, instead I tried some unorthodox moves to see what would happen and after a move or two later actually saw the checkmate open up.

He also had a tell where he'd stop flipping his pen when he was perplexed by some position. Once he started flipping I knew he had a plan.
 
  • Like
Likes russ_watters and berkeman
  • #3
19
69
The variation is just too long for a computer to see and in the end of that variation the computer still thinks it is losing. That is why the computer cannot solve it.

Humans can see patterns on the board. That is why this position is rather easy for us to solve.
 
  • #4
12,821
6,696
How long is the variation? Do you have an analysis of it? or how the position was constructed to fool the computer?
 
  • #5
19
69
the main variation is 6 moves deep for each side. Actually not that long.

the position is constructed in a way that the computer needs to sacrefice a piece (losing even more material) and after that it is possible to forcefully close the position.

Computers can't really appreciate closed or semi closed positions, because they do not think like: ''Hey, the position is now closed, both sides can't tough the enemy king anymore''

It is a weapon humans used and use to win against the computer or at least make a draw. But today it is very hard to reach such positions, because the opening book of the computer tries to avoid closed positions. And most of the time the computer has crushed you before you even can think about this concept.
 
  • #6
4,465
72
1. c4+ Kc6
2. d5+ Kd7
3. e6+ Kxd8
4. b3 job done

4 ...... gxf4
5. Kxf4 ....
6. Bf5 fortress closed and locked
 
  • #7
4,465
72
no it's
  1. Ba4+ Kxa4
  2. b3+ Kb5
  3. c4+ Kc6
  4. d5+ Kd7
  5. e6+ Kxd8
  6. f5 fortress ready
 
  • Like
Likes david2 and cnh1995
  • #8
19
69
well done Andre. that is the variation
 
  • #9
19
69
here is an interesting position which the leading chess engine (stockfish) cannot solve. Another leading chess engine (houdini) sees it in a matter of seconds.

I thought chess engines are coded in a very similar way but it seems I am wrong.

 
  • #10
19
69
here is a game played by a top class GM Nakamura against Rybka (which was the best chess engine some years ago)

notice how Nakamura managed to close the position which is a well known tactic to beat a computer.

the end is just hilarious....Nakamura was black.

Naamloos000.jpg


the whole game can be seen here. Nakamura exploits the 50 move rule. the computer thinks it is ahead and does not want a 50 move draw (50 moves played without a capture or pawn move is a draw) so it started to push pawns.it goes downhill fast after that.

the only weird thing is that the computer played the grob opening which is not a wise thing to do.but the game is legit. btw i play the grob sometimes in blitz. it is still a fun opening to play on my chess level.

 
Last edited:
  • #11
19
69
another game played by Nakamura against a computer.


the ending is again hilarious as you can see.
 

Related Threads on Chess position computers cannot solve

Replies
8
Views
5K
Replies
9
Views
1K
  • Last Post
2
Replies
31
Views
9K
  • Last Post
Replies
11
Views
2K
  • Last Post
Replies
11
Views
2K
  • Last Post
Replies
11
Views
2K
  • Last Post
Replies
11
Views
4K
  • Last Post
Replies
10
Views
2K
  • Last Post
2
Replies
46
Views
5K
  • Last Post
Replies
12
Views
868
Top