Stick that position in Fritz either side to move it doesn't matter,have it analyse it for a few minutes or even hours and it'll tell you Black is crushing White.It will even play 1.Rb7 or 1...,Rb7 after which White couldn't lose even if he wanted to(except by resigning of course)!
I just played a 5 minute game against Fritz8 from this position(me as White) and Fritz refused my draw offers at move 20,30 and 40 and only evaluated the position as drawn at move 43.
It's amazing what these machines can do but sometimes it's more amazing what they cannot do.
I've only tested it with Fritz8,crafty 19.01 and cometB50 so I don't know how later versions or other engines evaluate the position.
Originally posted by Gorgarthe computer doesn't actually understand there is no way to breakthrough. -- it just see's the 2 rooks, and bishop and can't understand why that isn't enough to win...
[fen]1r1k4/5b2/prp1p1p1/PpPpPpPp/1P1P1P1P/8/8/4K3 w - - 0 1[/fen]
Stick that position in Fritz either side to move it doesn't matter,have it analyse it for a few minutes or even hours and it'll tell you Black is crushing White.It will even play 1.Rb7 or 1...,Rb7 after which White couldn't lose even if he wanted to(except by resigning of course)!
I just pl 1 and cometB50 so I don't know how later versions or other engines evaluate the position.
I have actually played against a computer and reached a very similair position, to win a draw -- I'll post the PGN.
[Event "rated standard match"]
[Site "freechess.org"]
[Date "2006.06.18"]
[Round "?"]
[White "BugZH"]
[Black "Shinidoki"]
[Result "1/2-1/2"]
[WhiteElo "2265"]
[BlackElo "1741"]
[ECO "A04"]
[TimeControl "60+21"]
1. Nf3 c5 2. e4 Nc6 3. Bb5 d5 4. exd5 Qxd5 5. O-O Bg4 6. Nc3 Bxf3 7. Nxd5
Bxd1 8. Bxc6+ bxc6 9. Nc7+ Kd7 10. Nxa8 Bxc2 11. d3 e6 12. Be3 Nf6 13. Rac1
Ba4 14. Rc3 Bd6 15. Ra3 Rxa8 16. Rxa4 Nd5 17. Rc1 Nxe3 18. fxe3 e5 19. Kf2 f5
20. Kf3 g5 21. g4 f4 22. Rac4 Ke6 23. h3 Rf8 24. Ke4 Rf7 25. R4c2 a5 26. Rf2
Rb7 27. Kf3 Rb4 28. a3 Rb3 29. Ke4 h6 30. Rc4 Rb5 31. Rc3 Be7 32. b3 Rb6 33.
Rfc2 Bd6 34. Ra2 Be7 35. Rac2 Bd6 36. Re2 Be7 37. Ra2 Bd6 38. Ra1 Be7 39. Ra2
Bd6 40. Ra1 Be7 41. Rac1 Bd6 42. Rf1 Be7 43. Rfc1 Bd6 44. Re1 Be7 45. a4 Bd6
46. Rb1 Rb4+ 47. Rc4 Rb6 48. Rc3 Rb4+ 49. Rc4 Rb6 50. Rb2 Be7 51. Kf3 Kd5 52.
Rc3 Bd6 53. Ra2 Rb4 54. Rd2 Rb6 55. Rh2 Be7 56. Rf2 Rb4 57. Rd2 Rb6 58. Rdc2
Rb4 59. Re2 Rb6 60. Re1 Rb4 61. Re2 Rb6 62. Rd2 Rb4 63. Rd1 Rb6 64. Rf1 Rb4
65. Re1 Rb6 66. Re2 Rb4 67. Rd2 Rb6 68. Rd1 Rb4 69. Rb1 Rb6 70. Re1 Rb4 71.
Rd1 Rb6 72. Rb1 Rb4 73. Rc2 Rb6 74. Rc3 Rb4 75. Rc4 Rb6 76. Rcc1 Rb4 77. Rb2
Rb6 78. Rc4 Bd6 79. Rcc2 Rb4 80. Rc4 Rb6 81. Rc3 Rb4 82. Rd2 Rb6 83. Rg2 Rb4
84. Rb2 Rb6 85. Rb1 Rb4 86. Rc4 Rb6 87. Rc3 Rb4 88. Rc4 {Game drawn by mutual
agreement} 1/2-1/2
I just did a quick search on FICS for "BugZH", then looked at its profile.
according to the finger notes, It is "Sjeng 11.2" on a 233mz machine and has a standard rating of 2300, 600pts above me.
it absolutely crushed me in the opening, but then somehow got very confused and just started aimlessly moving. It was unwilling to open files or push pawns, so i got a draw...because the computer had a warped way of thinking.
^unlike your position though, I should imagine that was won.
Oh, and another interesting thing on fritz 8
Enter this position: -
White to move.
the solution should seem obvoius, Kxd3, yet, if I only let Fritz 8 run for 20 secounds or less (on my penIII pc) it seems to play Kd2! And after lets say Kd6, it still may not even capture!, but instead prefer to just aimlessly move the king around. (not that this will make any difference to the gamescore)
anyone else with fritz 8 get the same thing or is it just me?
**NB, other engines like crafty will play Kxd3 immediatly**
Even more amazing.From the position I gave,Fritz8 as white after 25 minutes of evaluating it gives 1.cxb6 as the best move yet discards 1.axb6!So it realises axb6 is losing but not cxb6?? 😲😕🙄
The damn thing is just clueless.I guess shredder is the better engine.Now I just gotta find a way to reach such a position each time vs Fritz 😉
Fritz on 20 secs a move from shinidoki's diagram and indeed,it doesn't capture the pawn.Really odd LOL
Computer programs are very one minded things, they have not 1 ounce of logic or common sense, all they have is a set of rules, guidlines and an exceptions list.
If fritz looks at that position and evalutes it as -15, but capturing the rook brings it down to -10 it considers that an improvement on its position -- ignoring the fact that -10 is still losing.....
basically, it assumes 2+2=4, when infact the awnser is 5
I shouldn't critise Fritz 8's ability from a single position to harshly however, because all you have found is one example where its evaluation system is flawed (if it wasn't flawed, then it would play near to perfect) -- If a 'flaw' in its evaluation was from move 3 in the ruy lopez then we would have cause for concern, but this position has no practical relevance, thus, its strength or rating won't be effect by the fact it can't solve a single puzzle.
^^ and yes, the position i gave is a complete draw. ((provided the pawn doesn't promote safely...which could only be done by blunder.))
and so, Kd2 is just as logical as Kxd3, but I am 99% certain almost all human players will play Kxd3 99% of the time, simply because the less objective "human logic" tells us to rid of the pawn. "machine logic" says it makes no difference, and so, it plays what ever it 'feels' like
Originally posted by ShinidokiTrue enough,this position has no real practical value.It's just something funny,nothing to draw any conclusions towards Fritz' playing strength.
Computer programs are very one minded things, they have not 1 ounce of logic or common sense, all they have is a set of rules, guidlines and an exceptions list.
If fritz looks at that position and evalutes it as -15, but capturing the rook brings it down to -10 it considers that an improvement on its position -- ignoring the fact that -10 is still losin ...[text shortened]... ine logic" says it makes no difference, and so, it plays what ever it 'feels' like
I can see why it would think -10 is better than -15 and thus takes the rook.But then how to explain it discards axb6?
I played the position against Fritz 9, with Fritz as white. It offered the evaluation that black was winning (-0.4), but would move only its king. It took some time, but we eventually repeated a position three times, and Fritz claimed the draw.
As black's material advantage is +13, the 4 pawn difference in evaluation reveals Fritz's relative contempt for the additional material. it clearly recognizes the positional factors that render such overwhelming force nearly impotent.
You can change the setting of fritz so that it offers a draw earlier.
Also what I find is that if you set up the position and then go back couple of moves and then make some moves to get that position back again, then fritz understand the position better.
So instead of setting up the position, set up a similar position and play some moves to reach the desired position.
"I can see why it would think -10 is better than -15 and thus takes the rook.But then how to explain it discards axb6?"
I have no clue, but at a guess, i would be some sort of odd "positional" consideration ....
And RahimK, Fritz would play the position better secound time round, it has a bunch of hash tables helping it.
Originally posted by ShinidokiI don't mean making it play again. I mean helping it understand the position better.
"I can see why it would think -10 is better than -15 and thus takes the rook.But then how to explain it discards axb6?"
I have no clue, but at a guess, i would be some sort of odd "positional" consideration ....
And RahimK, Fritz would play the position better secound time round, it has a bunch of hash tables helping it.
Say I give you a position and give you 4 minutes. You might not have a clue.
Now I show you the 4 moves that led up to the position. You might have a better understanding.
So i'm saying rewind the position couple of moves back and then move forward to the desired position.
So instead of setting up the normal starting position with 1.e4 e5 2.Nf3 Nc6
Start with the normal position and play the moves 1.e4 e5 2.Nf3 Nc6.
Originally posted by RahimKIn that case I apoligise, I misunderstood.
I don't mean making it play again. I mean helping it understand the position better.
Say I give you a position and give you 4 minutes. You might not have a clue.
Now I show you the 4 moves that led up to the position. You might have a better understanding.
So i'm saying rewind the position couple of moves back and then move forward to the desired ...[text shortened]... n with 1.e4 e5 2.Nf3 Nc6
Start with the normal position and play the moves 1.e4 e5 2.Nf3 Nc6.
Originally posted by Wulebgr-0.4 would be a fourth of a pawn.
I played the position against Fritz 9, with Fritz as white. It offered the evaluation that black was winning (-0.4), but would move only its king. It took some time, but we eventually repeated a position three times, and Fritz claimed the draw.
As black's material advantage is +13, the 4 pawn difference in evaluation reveals Fritz's relative contempt for ...[text shortened]... clearly recognizes the positional factors that render such overwhelming force nearly impotent.