Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old July 2nd 04, 03:16 PM
Antonio Torrecillas
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

Hello,

I'm searching for positions where computers play or evaluate wrongly.

Some days ago "magasin DSL" proposed here a position which was "solved"
by our friend Claus-Jurgen.
Curiously the king walk Ke4-d5 is not in the first 13 options for
Fritz-8. F8 prefers to wait with bishop or rook moves and F8 prefers too
Kg2 or Ke2 (with the idea of playing Kf1 soon) and when consider Ke4 is
with the idea of returning to f3 to f1 in the next moves.

---------------------------------
Magasin DSL wrote:

8/8/B2k4/2p4R/3b4/4pK2/5p2/8 w - - 0 9
White : Kf3,Rh5,Ba6 / Black : Kd6, Bd4,c5,e3,f2


Claus-Jurgen wrote:
"I think White wins. White forces the Black king to the edge and
develops mating threats which Black can delay only by sacrificing the
pawns. The white king or rook are placed in a way that Black will lose
his pawns if he moves the bishop.
Example: 1. Ke4 Kc7 2. Rh6 Kd7 3. Kd5 Ke7 ... "
--------------------------------

I know too "almost bloqued positions" where computer do not see it's
draw if completely bloqued and it's needed to open some line no matter
how big is the sacrifice needed.

Anyone can give me more examples?

thanks
Antonio T.

  #2   Report Post  
Old July 2nd 04, 06:01 PM
Mark S. Hathaway
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

Antonio Torrecillas wrote:

Hello,

I'm searching for positions where computers play or evaluate wrongly.
Anyone can give me more examples?


There's a good one from the recent (July 2004) Chess Life magazine.

Faris Gabbara - IM Ben Finegold, page 15

White gets a rook stuck in what looks like an active location,
but it can't move beyond that point. The computer thinks it's
equal until just a few moves from the end and then it suddenly
"realizes" White is lost.

1. d4 Nf6 2. Bg5 e6 3. Nf3 c5 4. c3 Be7 5. Nbd2 b6 6. Bd3 Bb7
7. O-O O-O 8. c3 cxd4 9. exd4 Nd5 10. Bxe7 Qxe7 11. Re1 d6
12. Nc4 Nd7 13. Ne3 Nxe3 14. Rxe3 Nf6 15. Ng5 h6 16. Ne4 Nd5
17. Rg3 f5 18. Nd2 Qf6 19. Nc4 e5 20. Ne3 Nxe3 21. fxe3 e4

"Now the white rook is shut out of the game. Fritz 7 thinks
White is fine till the end, showing how dumb computers still
are." -- Finegold

22. Bb5 Kh7 23. Qc2 g5 24. Rf1 Rf7 25. Be2 Raf8 26. c4 Qg6
27. Qd1 Qe6 28. Qc2 d5 29. b3 g4

"Fritz 7 thinks White has a great game here!" -- Finegold

30. h3 h5

"Now, Fritz finally thinks Black has equalized!" -- Finegold

31. hxg4 fxg4 32. Rxf7+ Qxf7

"And finally, Fritz prefers Black." -- Finegold

33. Qd1 Bc8 34. cxd5 Qc7 35. Qe1 a5 36. Rxg4 hxg4
37. Bxg4 Kg7!

0-1
  #3   Report Post  
Old July 2nd 04, 09:39 PM
Antonio Torrecillas
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

En/na Mark S. Hathaway ha escrit:
Antonio Torrecillas wrote:
I'm searching for positions where computers play or evaluate wrongly.
Anyone can give me more examples?


There's a good one from the recent (July 2004) Chess Life magazine.
Faris Gabbara - IM Ben Finegold, page 15 (...)


Thanks Mark,

It is just the kind of samples I'm searching.

I suposed Mr "Magasin DSL" put his position in engines and did not trust
what was reading. Those days, looking at ko world ch, I have seen people
who trust engine evaluations and I'm recopilating examples showing how
dangereous is to trust computers without any own reasoning.

Here you have an example I saw recently abouy a similar situation.

[Event "Bruxelles"]
[Site "?"]
[Date "1987.??.??"]
[Round "?"]
[White "Larsen, B."]
[Black "Torre, E."]
[Result "1-0"]
[ECO "C24"]
[WhiteElo "2565"]
[BlackElo "2540"]
[PlyCount "111"]
[EventDate "1987.??.??"]

1. e4 e5 2. Bc4 Nf6 3. d3 c6 4. Nf3 Be7 5. O-O d6 6. Re1 O-O 7. Nbd2
Nbd7 8. a3 h6 9. Ba2 Re8 10. Nf1 Nf8 11. Ng3 Be6 12. Bxe6 Nxe6 13. d4
Qc7 14. c4 Nh7 15.d5 Neg5 16. Nf5 Nxf3+ 17. Qxf3 Kh8 18. Be3 c5 19. Rac1
Bg5 20. h4 Bxe3 21. Rxe3 Rac8 22. b4 Qd8 23. Qg3 Rg8 24. f4 f6 25. b5
Qd7 26. fxe5 fxe5 27. Qg6 Rcd8 28.Rg3 Nf8 29. Qh5 Nh7 30. Re1 Rdf8
31.Rg6 Rf6 32. Rxf6 Nxf6 33. Qg6 Rf8 34. g4 Ne8 35. g5 Qf7 36. Qxf7 Rxf7
37. g6 $2 {(Fritz 8 thinks is "Best move", ... but that is not true)}
37...Rd7 $2 {(Fritz8 prefers any other rook move different to Re7 in the
place of playing ...)} (37... Rxf5 $3 38. exf5 Nf6 $11 {and this
position is an easy draw despite Fritz evaluates as +- (2.53).}) 38.Rf1
Kg8 39. a4 h5 40. Kf2 Nf6 41.Ke3 Kf8 42. a5 Rd8 43. Kd3 Rd7 44. a6 b6
45.Rg1 Ng4 46. Rxg4 $3 {(For Fritz8 is 12th option. At least it
evaluates as +- ... +1.97)} 46... hxg4 47. Ke3 Rd8 48. Kf2 Kg8 49. Kg3
Rd7 50. Kxg4 Kf8 51. Kg5 Kg8 52. h5 Kh8 53. h6 gxh6+ 54. Nxh6 Kg7
55.Nf5+ Kf8 56. Kf6 1-0

Antonio T.

Ps: Remember that "$3" mean !! (good move) and $2 mean ? (bad move)

  #4   Report Post  
Old July 2nd 04, 11:36 PM
Antonio Torrecillas
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

En/na Antonio Torrecillas ha escrit:

Hello,

I'm searching for positions where computers play or evaluate wrongly.


I remembered this one and I have found it in Tim Krabbé page:
http://www.xs4all.nl/~timkr/chess2/diary_6.htm

item 114 / Computers can't play chess (ctd.)

----------------------------------------------------
r7/8/5R2/2k5/8/3R1K2/6P1/8 w - - 0 70
wKf3,Rd3,f6,Pg2/bKc5,Ra8

In Cadaquès, Spain, there is a match going on between Deep Fritz and
Deep Junior for the right to play Kramnik. In the fourth game of this
match, Deep Junior - Deep Fritz, there was a strange case of chess
blindness. In the diagrammed position, there followed: 70.Rc6+! White
wants to get rid of his Rook, so he can read his tablebases, instead of
think. 70...Kb5! Of course, Black doesn't want to let him; 'tablebase'
is just as bad as 'mate'. 71.Rc5+?? Incredible - he misses a tablebase
in two with 71.Rb3+ 71...Kb4! Setting a cunning trap. 72.Rb5+! Because
after 72.Rb3+ Kxc5! there is no tablebase, hence: draw. 72...Kc4 73.Rd4+
Kc3 74.Rc5+ Kxd4? For some strange reason, Black suddenly acknowledges
he's in a tablebasing net. He could have prolonged the struggle with
Kb3. Now, there followed 75.Rf5 tablebase.
Compare item 61 in this diary. It's like buying a can of beer, then
taking the plane to Zimbabwe because you have a friend there who knows
how to open them.
----------------------------------------------------

This item 61 ican be found in:
http://www.xs4all.nl/~timkr/chess2/diary_4.htm

AT

  #5   Report Post  
Old July 3rd 04, 10:52 PM
bruno de baenst
 
Posts: n/a
Default About positions where computer play or evaluate wrongly


I had such a position in one of my games where I could find the fastest
winning move pretty fast but my computer doesn't and often isn't able to win
the game when you blitz it out, and if it does manage to win, it first
wastes like 30 moves with tons of useless rook moves.

Volckaert,Marc (1750) - De Baenst,Bruno (1933) [C44] 14.02.2003
1.e4 e5 2.Nf3 Nc6 3.Be2 Nf6 4.d3 Bc5 5.Be3 d6 6.Nc3 Be6 7.h3 Qd7 8.Bxc5 dxc5
9.Ng5 Nd4 10.Nxe6 Qxe6 11.g3 b5 12.Bg4 Qb6 13.Nd5 Nxd5 14.exd5 b4 15.Bf3
Nxf3+ 16.Qxf3 0-0 17.0-0 Rad8 18.Rfe1 Qd6 19.Qe4 f6 20.Re2 Qxd5 21.Qxd5+
Rxd5 22.Rae1 Rd7 23.Kg2 Re8 24.Re4 Rd4 25.R4e2 Red8 26.b3 g5 27.g4 R4d6
28.Kg3 Ra6 29.Ra1 Kf7 30.Re4 Ke6 31.Kf3 Ra3 32.Ree1 Rd6 33.Reb1 Rda6 34.Rb2
Kd5 35.Ke3 R6a5 36.f3 Ra6 37.Kd2 Kd4 38.Rh1 f5 39.h4 Rh6 40.h5 f4 41.Re1
Rha6 42.Re4+ Kd5 43.Kc1 Rxa2 44.Rxa2 Rxa2 45.Kb1 Ra6 46.Kb2 Rc6 47.Rc4 Rd6
48.Ka2 e4! This nice breaking move that wins easily my fritz 8 can't find
(not in first 20 minutes anyway , didn't test longer) 49.Rxe4 Re6 50.Kb2
Rxe4 51.dxe4+ Kd4 52.Kc1 c4 53.Kb2 cxb3 54.cxb3 c5 55.Kc2 c4 56.bxc4 Kxc4
57.Kb2 a5 58.Kc2 a4 59.Kb1 b3 60.Kb2 Kb4 61.Kb1 a3 62.h6 Kc3 63.e5 a2+
64.Ka1 Kc2 0-1







"Antonio Torrecillas" schreef in bericht
...
Hello,

I'm searching for positions where computers play or evaluate wrongly.

Some days ago "magasin DSL" proposed here a position which was "solved"
by our friend Claus-Jurgen.
Curiously the king walk Ke4-d5 is not in the first 13 options for
Fritz-8. F8 prefers to wait with bishop or rook moves and F8 prefers too
Kg2 or Ke2 (with the idea of playing Kf1 soon) and when consider Ke4 is
with the idea of returning to f3 to f1 in the next moves.

---------------------------------
Magasin DSL wrote:

8/8/B2k4/2p4R/3b4/4pK2/5p2/8 w - - 0 9
White : Kf3,Rh5,Ba6 / Black : Kd6, Bd4,c5,e3,f2


Claus-Jurgen wrote:
"I think White wins. White forces the Black king to the edge and
develops mating threats which Black can delay only by sacrificing the
pawns. The white king or rook are placed in a way that Black will lose
his pawns if he moves the bishop.
Example: 1. Ke4 Kc7 2. Rh6 Kd7 3. Kd5 Ke7 ... "
--------------------------------

I know too "almost bloqued positions" where computer do not see it's
draw if completely bloqued and it's needed to open some line no matter
how big is the sacrifice needed.

Anyone can give me more examples?

thanks
Antonio T.



---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.699 / Virus Database: 456 - Release Date: 4/06/2004




  #6   Report Post  
Old July 5th 04, 09:40 PM
Oliver Maas
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

Hi,

one position from my own games: i had white and i had to sacrifice a knight
to get a winning position.
The move 17.h4! was not found by Fritz, however, further analysis showed
that Black has no reasonable
defense. Fritz´ tame proposal 17.Nf3 leads only to equality.

Maas,O - Gnirk,T [A02]
Reutlingen - Urach

1.e4 b6 2.f4 Bb7 3.Nc3 e6 4.Nf3 Bb4 5.Bd3 Na6N
6.a3 Bxc3 7.dxc3 Nc5 8.Qe2 Nf6 9.e5 Nfe4 10.Be3 f5
11.0-0-0 Qe7 12.Bxc5 Qxc5 13.Bxe4 Bxe4 14.Ng5 Bd5 15.Rd4 0-0?

[necessary is 15...g6 16.Rhd1 h6 17.Nf3 0-0-0=]

16.Qh5!+- 16...h6

[Here Fritz8 considers 17.Nf3 with equal play. Later he prefers 17.Rg1 with
equal play,
still later he likes 17.Nh3 (-0.12). But i think that after 17.h4 Black is
already lost, Fritz8 does not come
to this conclusion at this point, however, if one follows into the
variations, Fritz8 changes its opinion.]

17.h4! Qe7

[17...Qb5 (Fritz8) 18.Qg6 hxg5 19.hxg5 Qe2 20.Qh7+ now Fritz8 says +-;
17... d6 18.Qg6 +-]

18.c4

[18.Qg6 is easier 18...hxg5 19.hxg5+-]

18...Bxg2 19.Rh2 Bc6 20.Qg6 hxg5

[Thats where Fritz´ evaluation turns from -/+ to +/-, now he sees the danger
for the Black king, but too late.]

21.hxg5 Qe8

[Other moves don´t help either
21...Qc5 22.c3 Bf3 23.Qh7+ Kf7 24.Rxd7+! (24.g6+ Ke8 25.Qxg7+-) 24...Ke8
25.Rhd2 Qg1+ 26.Kc2 Be4+ 27.Kb3+-; 21...d6 22.Rh8+!! Kxh8 23.Qh5+ Kg8
24.g6+-;
21...Bf3 22.Rh6! Rad8 23.Qh7+ Kf7 24.Rg6 Rg8 25.Rf6+! Qxf6 26.exf6+-]

22.Qh7+ Kf7 23.g6+ Ke7 24.Qh4+! Rf6 25.exf6+ gxf6 26.Qh7+ 1-0

kind regards

Oliver



  #7   Report Post  
Old July 5th 04, 10:41 PM
Antonio Torrecillas
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

Thanks Bruno and Oliver,

This last one reminds me "MR. Nemeth strategy" to beat computers.

AT

En/na Oliver Maas ha escrit:
Hi,

one position from my own games: i had white and i had to sacrifice a knight
to get a winning position.
The move 17.h4! was not found by Fritz, however, further analysis showed
that Black has no reasonable
defense. Fritz´ tame proposal 17.Nf3 leads only to equality.

Maas,O - Gnirk,T [A02]
Reutlingen - Urach

1.e4 b6 2.f4 Bb7 3.Nc3 e6 4.Nf3 Bb4 5.Bd3 Na6N
6.a3 Bxc3 7.dxc3 Nc5 8.Qe2 Nf6 9.e5 Nfe4 10.Be3 f5
11.0-0-0 Qe7 12.Bxc5 Qxc5 13.Bxe4 Bxe4 14.Ng5 Bd5 15.Rd4 0-0?
16.Qh5!+- 16...h6
17.h4! Qe7


  #8   Report Post  
Old July 6th 04, 03:55 PM
Doctor SBD
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

One of the paradoxes I have noticed about Fritz is that it rarely finds
long-term sacrificial combinations, but if you input the move starting the
combination,it quickly finds the win! It can sit there for an hour, wanting to
play something that leads to a slight plus for the other side, but when you
play the first move, it quickly assigns a plus to you...

Just somewhat strange....

SBD
  #9   Report Post  
Old July 6th 04, 04:25 PM
Kenny B.
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

Heh, yeah this has always been something of an annoyance for me as well.

It seems that all computers do it as well. I once took a game where Kasparov
made a brilliant sacrifice and tried to see if my PC would duplicate it, but it
wouldn't no matter how quick or slow I put it. As soon as I made the move, it
was up massive points. I don't understand fully what makes this happen either
since I'm no computer expert. I would like to hear from anyone who might know
how to go about changing this.

"Doctor SBD" wrote in message
...
One of the paradoxes I have noticed about Fritz is that it rarely finds
long-term sacrificial combinations, but if you input the move starting the
combination,it quickly finds the win! It can sit there for an hour, wanting to
play something that leads to a slight plus for the other side, but when you
play the first move, it quickly assigns a plus to you...

Just somewhat strange....

SBD


  #10   Report Post  
Old July 6th 04, 04:44 PM
Doctor SBD
 
Posts: n/a
Default About positions where computer play or evaluate wrongly

Exactly, Kenny. I also had a game once that I analyzed where both players had
castled on opposite sides, and we were using pawn levers to try to get to each
other's king. The center was also dynamic.

On each player's move, the computer assigned a fairly hefty plus. But you could
even carry out the recommended move, and the eval would switch back to a plus
for the other side!

Maybe this is why IM Ben Finegold says computers are stupid.

SBD
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On



All times are GMT +1. The time now is 07:08 PM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Copyright ©2004-2019 ChessBanter.
The comments are property of their posters.
 

About Us

"It's about Chess"

 

Copyright © 2017