1
2
![[hu]](/mwf/v12/flags/hu.png "Hungary")
Hi everyone,
Grandmaster Jaan Ehlvest will take on Rybka in a match at pawn odds. We are still looking for an online transmission method - if you'd like to help, or suggest something, please let me know.
# of games: 8
Dates: March 6-8 (start around noon EST)
Play schedule: start around noon EST, 3 games March 6, 3 games March 7, 2 games March 8
Time Control: 45' + 10"
Starting positions: Rybka has white and starts each game without one of her pawns
Prize: significant, but Ehlvest must win to get it
Method of play: live (in-person) games
Rybka hardware: QX6700
Rybka version: private anti-human version of Larry Kaufman
Opening book: private handicap book of Larry Kaufman
Vas
Grandmaster Jaan Ehlvest will take on Rybka in a match at pawn odds. We are still looking for an online transmission method - if you'd like to help, or suggest something, please let me know.
# of games: 8
Dates: March 6-8 (start around noon EST)
Play schedule: start around noon EST, 3 games March 6, 3 games March 7, 2 games March 8
Time Control: 45' + 10"
Starting positions: Rybka has white and starts each game without one of her pawns
Prize: significant, but Ehlvest must win to get it
Method of play: live (in-person) games
Rybka hardware: QX6700
Rybka version: private anti-human version of Larry Kaufman
Opening book: private handicap book of Larry Kaufman
Vas
Bump :)
Vas
Vas
![[ee]](/mwf/v12/flags/ee.png "Estonia")
Jaan, anna tuld!
It would be nice to have online video stream of the event. VLC streaming server can be useful to produce a stream. See http://www.videolan.org/vlc/
If there was additional tool for visualizing moves on board over the Internet (maybe some chess server?) then it was a good start for show.
Lots of bandwidth will be required too, if word spreads and interest grows.
If there was additional tool for visualizing moves on board over the Internet (maybe some chess server?) then it was a good start for show.
Lots of bandwidth will be required too, if word spreads and interest grows.
![[be]](/mwf/v12/flags/be.png "Belgium")
in fritz interface, you can use file > print > publish game
(with option to automatically refresh). Just fill in the webserver details in the publish popup window,
and all you need is an internet connection from the PC connected to the DGT board.
But maybe there are better methods (maybe the TOMA from dgtprojects?)
(with option to automatically refresh). Just fill in the webserver details in the publish popup window,
and all you need is an internet connection from the PC connected to the DGT board.
But maybe there are better methods (maybe the TOMA from dgtprojects?)
The simplest way would be a website that updates automatically.
We would have to change it for every new move and we could add easily diagrams, comments, etc.
This has also the advantage that we wouldn't have to use flash, java,...
The disadvantage is that this isn't as comfortable as a transmission on a chess server
We would have to change it for every new move and we could add easily diagrams, comments, etc.
This has also the advantage that we wouldn't have to use flash, java,...
The disadvantage is that this isn't as comfortable as a transmission on a chess server
![[us]](/mwf/v12/flags/us.png "United States")
This is great to hear!When GM's get involved this will surely bring in sponsors but as you say a prize fund is already established so interest is already formed.In my opinion I hate to make predictions but with even a pawn advantage a human (Even a GM)does not have much of a chance to at least tie the match.I predict a lopsided victory for Rybka of 6.5-1.5
FWCC
FWCC
I expect some draws but I doubt that he can win a game...
Assuming GM Ehlvest knows which pawn Rybka will be playing without and assuming he is willing to develop a lot of opening lines that get Rybka out of book early (at which point he can predict her moves fairly well a priori with his own Rybka), he might have some chances. Under the same conditions but with Kamsky as the opponent, I would be inclined to bet on him to win the match...
I just had an interesting thought.Why don't we make it PAWN AND MOVE odds.Remove one of her pawns maybe either c7 or f7 pawn,but to make it EVEN more interesting how about KNIGHT ODDS!Remove one of her knights maybe the B8 knight and lets go from there.I think then with a GM having a 3 pawn and move advantage would almost even the playing field,but again I would give the match win to Rybka.So then lets have a KNIGHT ODDS match like in the days of Morphy when he offered this to his opponents,now that would be a match! It's really something that now these monsters have to give our top players odds to make it interesting at least.I predict in less than ten years we will have datasets with SOLVED positions that you can download to make available to your engine much like our tablebases today.Are we really approaching a time when chess will be solved by these beast,time will tell.But that is a LONG WAY off-hopefully.
FWCC
FWCC
I played some knight-odds games against rybka. that was blitz and I didn't won every game but I won most of them.
I am only rated about 2100 elo and for a GM piece odds is a joke!
I am only rated about 2100 elo and for a GM piece odds is a joke!
![[dk]](/mwf/v12/flags/dk.png "Denmark")
anyway, a pawn is not a piece.
I just awnsered to the previous poster. didn't you notice?
I did. Don't feel offended. I noticed only after I replyied to you.
If you feel like it, I can alway help you get rit of posts of your own, that seem redundant.
Best,
If you feel like it, I can alway help you get rit of posts of your own, that seem redundant.
Best,
By Michael Waesch ![[de]](/mwf/v12/flags/de.png "Germany")
Date 2007-02-25 12:46
On CCC people already laugh about this "event". Mostly people wanna see real chess with both sides having a complete set of pieces on the board. So it seems that only the hardcore Rybka worshippers will watch. The majority of people who stated their opinions publicly simply don´t take it seriously.
Mike
Mike
Well, we already know about what you think about the standards at CCC, so why do you bring forth those statements here? :-)
Personally, I think it is a step in a better direction than the one used in the Kramnik-Fritz match. The main problem there was that Kramnik didn't play as humans have always played, that is, without assistance (from the Fritz book). Plus, he had access to the exact Fritz engine version being used months in advance.
Actually, back then I suggested here that Vas and company should go one better and offer pawn odds to any human, under otherwise equal, non-handicapped conditions. Seems like I was not the only one considering it to be a great idea :-)
Personally, I think it is a step in a better direction than the one used in the Kramnik-Fritz match. The main problem there was that Kramnik didn't play as humans have always played, that is, without assistance (from the Fritz book). Plus, he had access to the exact Fritz engine version being used months in advance.
Actually, back then I suggested here that Vas and company should go one better and offer pawn odds to any human, under otherwise equal, non-handicapped conditions. Seems like I was not the only one considering it to be a great idea :-)
Why would anybody laugh about this? You can find it more or less interesting, but why the despising?
Would these people find it more interesting if ordinary GMs played Rybka or any other top program on equal odds and managed a draw or two out of 10 games?
Regarding the significance, I think the day - if it comes - GMs can't win against knight-odds, a lot of people will have to revise their view about chess.
Would these people find it more interesting if ordinary GMs played Rybka or any other top program on equal odds and managed a draw or two out of 10 games?
Regarding the significance, I think the day - if it comes - GMs can't win against knight-odds, a lot of people will have to revise their view about chess.
![[il]](/mwf/v12/flags/il.png "Israel")
I think that day when GM's cannot win with knight odd not in blitz never comes
Knight is simply too much handicap to score 50% or more than it at slow time control even against IM's
In one game humans can blunder but in most game they are going to win.
Uri
Knight is simply too much handicap to score 50% or more than it at slow time control even against IM's
In one game humans can blunder but in most game they are going to win.
Uri
Uri, you sound like Kasparov about 12-15 years ago when he said a computer could never beat him at chess. Of course he made that statement out of complete ignorance. You should know better. I'm not sure of the value of Knight odds but I did read somewhere about 30 years (or more) ago that Knight odds in GM play was worth about 400 Elo. No matter what the advantage is, time will erase that advantage as computer programs get better and hardware gets faster. About the only thing that can prevent that is the end of the world.
I think that a knight worth clearly more than 400 elo at the high level at slow time control of 90+30.
My guess is that
2000 player may perform at level of 1600 if he starts without a knight but
2600 player may perform only at level of 1900 if he starts without a knight.
It is clear that there is an handicap that is clearly too much against GM's regardless of playing strength of the opponent.
My opinion is that knight handicap is simply too much handicap to win against GM's at the high level.
Uri
My guess is that
2000 player may perform at level of 1600 if he starts without a knight but
2600 player may perform only at level of 1900 if he starts without a knight.
It is clear that there is an handicap that is clearly too much against GM's regardless of playing strength of the opponent.
My opinion is that knight handicap is simply too much handicap to win against GM's at the high level.
Uri
Actually, this is an interesting topic.
Larry has been running a lot of handicap matches. Of course, the level of the side receiving the handicap is the main factor in the outcome, but I've been surprised by two things:
1) It seems that handicaps can be given to much stronger players than I would have ever thought possible. (Two months ago, I would have told you that I can beat God at tournament time-control with an extra knight.)
2) The level of the side giving the handicap is a lot more important than I thought.
Of course, the counter-argument is that engines just don't play well when they receive a handicap. That's why the games against Ehlvest will be interesting.
Vas
Larry has been running a lot of handicap matches. Of course, the level of the side receiving the handicap is the main factor in the outcome, but I've been surprised by two things:
1) It seems that handicaps can be given to much stronger players than I would have ever thought possible. (Two months ago, I would have told you that I can beat God at tournament time-control with an extra knight.)
2) The level of the side giving the handicap is a lot more important than I thought.
Of course, the counter-argument is that engines just don't play well when they receive a handicap. That's why the games against Ehlvest will be interesting.
Vas
I am looking forward to the pawn-odds games, but about knight-odds I have to say: I don't believe there will ever be a comp, that can win a match against a GM :-)
I tested against rybka at 10 min games and won convincingly, then I tried at 5 min games. here are the games I played. I even annotated them a bit, haha.
result: 2.5-2.5
I feel I played badly, but against a comp it's not that easy at 5'.
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "lem"]
[Black "Rybka 2.2 32-bit"]
[Result "1-0"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "85"]
1. Nf3 {0} Nf6 {12} 2. d4 {2} g6 {9} 3. c4 {1} Bg7 {10} 4. Nc3 {1} d5 {14} 5.
cxd5 {2} Nxd5 {2} 6. e4 {2} Nxc3 {6} 7. bxc3 {0} O-O {11} 8. Rb1 {3} c5 {10} 9.
d5 {
of course this makes more sense with a knight on c6, but I can afford this 25}
Bxc3+ {6} 10. Bd2 {4} Bxd2+ {2} 11. Qxd2 {1} a6 {7} 12. Be2 {10} b5 {0} 13. h4
{6} Bg4 {7} 14. h5 {22} Bxf3 {5} 15. Bxf3 {2} Qd6 {6} 16. Qh6 {14} Qe5 {6} 17.
hxg6 {queentrade is forced 47} Qc3+ {2} 18. Qd2 {2} Qxd2+ {0} 19. Kxd2 {1} fxg6
{6} 20. Ke3 {3} Kg7 {4} 21. Rhc1 {21} c4 {
as this was the first game at this timecontrol I thought: man that's easy! 1}
22. Kd4 {3} h5 {9} 23. a4 {8} Rab8 {10} 24. axb5 {3} axb5 {2} 25. Rc2 {4} h4 {5
} 26. Rcb2 {21} h3 {1} 27. Rxb5 {4} Rxb5 {3} 28. Rxb5 {0} c3 {10} 29. Kxc3 {7}
hxg2 {0} 30. Bxg2 {12} Rxf2 {0} 31. Rb2 {5} Rf8 {0} 32. Kd4 {2} g5 {24} 33. Bh3
{1} Rh8 {5} 34. Bf5 {2} Rh1 {2} 35. Rb6 {1} Kf7 {3} 36. Rg6 {2} Rg1 {3} 37. Ke5
{1} Rh1 {4} 38. Rxg5 {2} Kf8 {2} 39. Ke6 {1} Ra1 {3} 40. Ke5 {2} Rh1 {1} 41.
Be6 {1} Rh8 {1} 42. Rg8+ {7} Rxg8 {0} 43. Bxg8 {1} 1-0
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "lem"]
[Black "Rybka 2.2 32-bit"]
[Result "0-1"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkb1r/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "230"]
1. Nf3 {0} d5 {11} 2. d4 {2} Nc6 {10} 3. c4 {2} e6 {12} 4. Nc3 {2} Bb4 {11} 5.
cxd5 {3} exd5 {8} 6. Bg5 {2} f6 {14} 7. Bf4 {16} g5 {10} 8. Be3 {6} Bf5 {6} 9.
g3 {6} Be4 {0} 10. Bg2 {2} Bxc3+ {4} 11. bxc3 {2} Qd7 {18} 12. O-O {4} O-O-O {
12} 13. a4 {12} b6 {0} 14. a5 {13} Nxa5 {4} 15. Rxa5 {just to play it save 3}
bxa5 {1} 16. Qa1 {2} a4 {7} 17. Qa3 {2} h5 {4} 18. Ra1 {6} h4 {0} 19. Qxa4 {11}
Qxa4 {2} 20. Rxa4 {2} hxg3 {3} 21. hxg3 {3} Kb7 {9} 22. Nd2 {4} Bxg2 {0} 23.
Kxg2 {1} Rb8 {6} 24. Nb3 {2} Kc6 {1} 25. Nc5 {1} a5 {1} 26. Rxa5 {3} Ra8 {0}
27. Rxa8 {6} Rxa8 {1} 28. Kf3 {I couldn't imagine, that I wouldn't win this 6}
Kd6 {1} 29. Bd2 {28} Ra2 {0} 30. Ke3 {1} c6 {1} 31. f4 {7} gxf4+ {1} 32. gxf4 {
2} Ra1 {2} 33. Kd3 {2} f5 {1} 34. Be3 {11} Rd1+ {12} 35. Kc2 {1} Re1 {5} 36.
Kd2 {1} Ra1 {0} 37. Bf2 {2} Rh1 {1} 38. Kd3 {10} Rh3+ {6} 39. Kd2 {8} Rh1 {1}
40. Nd3 {3} Rh2 {1} 41. Bg3 {9} Rh3 {2} 42. Be1 {3} Rh7 {0} 43. Bf2 {
in a blitzgame this is not easy to win 11} Rh2 {2} 44. Be3 {1} Rg2 {5} 45. Ne5
{1} Rg7 {0} 46. Kd3 {1} Ra7 {3} 47. Bf2 {1} Ra1 {4} 48. Bh4 {1} Ra8 {1} 49. Bf6
{4} Rf8 {0} 50. Bg7 {1} Rg8 {3} 51. Bh8 {
go never for such cheap tactics against a comp. ahh also against a human 3} Ke6
{2} 52. Nf7 {16} Kxf7 {1} 53. Be5 {1} Ke6 {1} 54. e3 {1} Ra8 {1} 55. Kd2 {2}
Ra2+ {2} 56. Kd3 {0} Ra1 {1} 57. Kd2 {2} Rb1 {0} 58. Kd3 {0} Rb2 {2} 59. Bc7 {2
} Rb5 {0} 60. Be5 {1} Rb1 {1} 61. Bc7 {1} Rd1+ {2} 62. Ke2 {1} Ra1 {0} 63. Kd3
{1} Re1 {1} 64. Be5 {1} Rf1 {1} 65. Bc7 {1} Ra1 {1} 66. Be5 {1} Rd1+ {2} 67.
Ke2 {1} Rh1 {1} 68. Kd3 {1} Rg1 {0} 69. Bc7 {3} Rg3 {1} 70. Be5 {1} Rg2 {1} 71.
Bc7 {1} Rg1 {0} 72. Be5 {1} Rh1 {1} 73. Bc7 {1} Kd7 {2} 74. Be5 {1} Ra1 {1} 75.
Bf6 {0} Kd6 {0} 76. Be5+ {1} Ke7 {0} 77. Bg7 {1} Rb1 {1} 78. Be5 {1} Rd1+ {0}
79. Ke2 {1} Rh1 {1} 80. Kd3 {0} Rb1 {1} 81. Bg7 {1} Kd7 {1} 82. Be5 {1} Rb2 {2}
83. Bf6 {1} Kd6 {0} 84. Be5+ {1} Ke7 {1} 85. Bc7 {1} Ra2 {1} 86. Be5 {1} Ra8 {1
} 87. Bg7 {1} Ra1 {1} 88. Be5 {0} Rf1 {2} 89. Bg7 {1} Kd6 {0} 90. Be5+ {1} Kd7
{0} 91. Bg7 {0} Rd1+ {4} 92. Ke2 {1} Rb1 {3} 93. Be5 {1} Rb7 {1} 94. Kd3 {0}
Rb3 {7} 95. Bf6 {1} Ra3 {4} 96. Be5 {0} Ke6 {1} 97. Bg7 {1} Kf7 {1} 98. Be5 {0}
Rb3 {2} 99. Bc7 {0} Ke6 {0} 100. Be5 {0} Rb7 {3} 101. Bh8 {1} Rb2 {1} 102. Be5
{1} Rb5 {0} 103. Bg7 {1} c5 {1} 104. Be5 {
couldn't react to c5, cos I constantly had to premove 0} c4+ {0} 105. Kd2 {2}
Rb2+ {0} 106. Kd1 {1} Ra2 {0} 107. Bg7 {1} Ra3 {0} 108. Kd2 {1} Kd7 {0} 109.
Be5 {1} Ra2+ {0} 110. Kd1 {1} Ke7 {0} 111. Bg7 {1} Ra6 {1} 112. Be5 {1} Kf7 {0}
113. Bc7 {6} Kg6 {0} 114. Be5 {1} Ra2 {0} 115. Bd6 {3} Kh5 {
time - and lost anyway 0} 0-1
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "Rybka 2.2 32-bit"]
[Black "lem"]
[Result "1/2-1/2"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[PlyCount "95"]
1. Nf3 {9} d5 {12} 2. e3 {9} Nf6 {54} 3. b3 {5} g6 {3} 4. Bb5+ {8} Bd7 {3} 5.
Qe2 {5} Bg7 {7} 6. Bb2 {7} a6 {6} 7. Bxd7+ {17} Nbxd7 {5} 8. O-O {14} O-O {1}
9. c4 {8} dxc4 {2} 10. Qxc4 {7} c5 {11} 11. Rac1 {9} b5 {2} 12. Qh4 {8} Qb6 {5}
13. Bc3 {8} Rfe8 {37} 14. d3 {7} Nh5 {14} 15. d4 {5} cxd4 {6} 16. Bxd4 {2} Bxd4
{6} 17. Qxd4 {3} Qxd4 {1} 18. Nxd4 {3} Rac8 {2} 19. Rfd1 {2} Nb6 {5} 20. Nc6 {5
} Kg7 {8} 21. Kf1 {0} Rc7 {9} 22. Nb4 {9} Rxc1 {4} 23. Rxc1 {3} Rc8 {13} 24.
Rxc8 {2} Nxc8 {0} 25. Nxa6 {1} Kf6 {14} 26. a4 {4} bxa4 {2} 27. bxa4 {0} Ke6 {2
} 28. Ke2 {2} Kd6 {2} 29. Nb4 {1} Ng7 {6} 30. Kd3 {3} Ne8 {1} 31. Kc4 {6} Nc7 {
1} 32. e4 {25} Nb6+ {3} 33. Kb3 {0} e6 {5} 34. Nd3 {3} f5 {3} 35. a5 {2} Nd7 {5
} 36. exf5 {0} gxf5 {2} 37. Kc4 {2} Ne5+ {and white won't lose. I think
against any human I would have won even if he played perfectly. against a comp
there's a significant psycholgical disadvantage :-) 5} 38. Nxe5 {1} Kxe5 {0}
39. Kc5 {4} h5 {6} 40. Kc6 {3} Na6 {6} 41. Kb6 {2} Nb4 {3} 42. Kb5 {2} Nd5 {3}
43. Kc5 {1} Nc7 {9} 44. Kc6 {0} Na6 {1} 45. Kb6 {1} Nb4 {1} 46. Kb5 {1} Nd5 {1}
47. Kc5 {0} Nc7 {1} 48. Kc6 {7} 1/2-1/2
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "Rybka 2.2 32-bit"]
[Black "lem"]
[Result "1/2-1/2"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[PlyCount "79"]
1. Nc3 {16} d5 {4} 2. d4 {7} Nf6 {3} 3. Bf4 {19} e6 {2} 4. g3 {10} Bd6 {1} 5.
Bg2 {14} Bxf4 {2} 6. gxf4 {7} O-O {10} 7. Qd2 {6} b6 {1} 8. O-O-O {7} Bb7 {4}
9. Rhg1 {6} Nbd7 {4} 10. Kb1 {5} g6 {2} 11. a3 {8} c5 {1} 12. Bf3 {1} c4 {11}
13. h4 {9} a6 {2} 14. h5 {6} b5 {9} 15. Rg5 {6} Nb6 {22} 16. Rdg1 {8} Qd7 {13}
17. Qe3 {6} Qd6 {14} 18. f5 {3} exf5 {6} 19. Rxf5 {2} Ne4 {14} 20. Nxe4 {7}
dxe4 {1} 21. Bxe4 {3} Bxe4 {1} 22. Qxe4 {4} Rae8 {4} 23. Re5 {2} Rxe5 {1} 24.
dxe5 {1} Qd5 {2} 25. Qe3 {3} Na4 {8} 26. f4 {0} Rd8 {5} 27. hxg6 {7} hxg6 {2}
28. Qh3 {damn- what to do against the perpetual after Rxg6? 2} Qd4 {the other t
hreat was maybe Rh1 - so I secured the draw and felt like an idiot again. 40}
29. Rxg6+ {1} fxg6 {1} 30. Qe6+ {0} Kg7 {1} 31. Qf6+ {0} Kh7 {2} 32. Qe7+ {1}
Kh6 {2} 33. Qh4+ {2} Kg7 {1} 34. Qe7+ {1} Kg8 {2} 35. Qe6+ {1} Kg7 {1} 36. Qf6+
{0} Kh7 {1} 37. Qe7+ {1} Kh6 {2} 38. Qh4+ {1} Kg7 {0} 39. Qe7+ {2} Kh6 {0} 40.
Qh4+ {2} 1/2-1/2
[Event "Blitz:5'"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "lem"]
[Black "Rybka 2.2 32-bit"]
[Result "1/2-1/2"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkb1r/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "94"]
[TimeControl "300"]
1. Nf3 {0} d5 {10} 2. d4 {7} Nc6 {8} 3. c4 {2} e6 {12} 4. Nc3 {3} Bb4 {12} 5.
Bg5 {4} f6 {3} 6. Bf4 {3} g5 {28} 7. Be3 {2} dxc4 {7} 8. g3 {7} Bd7 {11} 9. Bg2
{12} Ne7 {7} 10. Nd2 {30} Bc6 {0} 11. Bxc6+ {6} Nxc6 {1} 12. O-O {4} Bxc3 {2}
13. bxc3 {1} Qd5 {9} 14. f3 {15} Rd8 {7} 15. Bf2 {5} O-O {4} 16. e4 {2} Qb5 {12
} 17. Rb1 {4} Qa5 {4} 18. Qc2 {8} b5 {5} 19. a4 {15} Qxa4 {6} 20. Qxa4 {16}
bxa4 {0} 21. Nxc4 {2} Rb8 {5} 22. Na3 {15} Rb3 {5} 23. Rxb3 {12} axb3 {2} 24.
Rb1 {4} Rb8 {4} 25. Kf1 {4} a5 {7} 26. Ke2 {3} a4 {9} 27. Kd3 {2} Kf7 {3} 28.
f4 {1} gxf4 {2} 29. gxf4 {0} Ne7 {5} 30. c4 {6} f5 {3} 31. e5 {1} Ng6 {3} 32.
Be3 {3} b2 {1} 33. Nb5 {17} Rxb5 {noooooooooooo 0} 34. cxb5 {2} a3 {0} 35. Kc2
{7} a2 {0} 36. Kxb2 {2} axb1=Q+ {0} 37. Kxb1 {1} Ne7 {2} 38. Kc2 {5} Ke8 {0}
39. Kd3 {6} Kd7 {0} 40. Kc4 {2} c6 {3} 41. bxc6+ {9} Kxc6 {2} 42. Bd2 {11} Nd5
{0} 43. Kd3 {2} Nb6 {2} 44. Kc3 {2} Kd5 {1} 45. Kd3 {1} Kc6 {1} 46. Kc3 {1} Kd5
{1} 47. Kd3 {0} Kc6 {if it plays on it'll win on time 2} 1/2-1/2
I tested against rybka at 10 min games and won convincingly, then I tried at 5 min games. here are the games I played. I even annotated them a bit, haha.
result: 2.5-2.5
I feel I played badly, but against a comp it's not that easy at 5'.
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "lem"]
[Black "Rybka 2.2 32-bit"]
[Result "1-0"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "85"]
1. Nf3 {0} Nf6 {12} 2. d4 {2} g6 {9} 3. c4 {1} Bg7 {10} 4. Nc3 {1} d5 {14} 5.
cxd5 {2} Nxd5 {2} 6. e4 {2} Nxc3 {6} 7. bxc3 {0} O-O {11} 8. Rb1 {3} c5 {10} 9.
d5 {
of course this makes more sense with a knight on c6, but I can afford this 25}
Bxc3+ {6} 10. Bd2 {4} Bxd2+ {2} 11. Qxd2 {1} a6 {7} 12. Be2 {10} b5 {0} 13. h4
{6} Bg4 {7} 14. h5 {22} Bxf3 {5} 15. Bxf3 {2} Qd6 {6} 16. Qh6 {14} Qe5 {6} 17.
hxg6 {queentrade is forced 47} Qc3+ {2} 18. Qd2 {2} Qxd2+ {0} 19. Kxd2 {1} fxg6
{6} 20. Ke3 {3} Kg7 {4} 21. Rhc1 {21} c4 {
as this was the first game at this timecontrol I thought: man that's easy! 1}
22. Kd4 {3} h5 {9} 23. a4 {8} Rab8 {10} 24. axb5 {3} axb5 {2} 25. Rc2 {4} h4 {5
} 26. Rcb2 {21} h3 {1} 27. Rxb5 {4} Rxb5 {3} 28. Rxb5 {0} c3 {10} 29. Kxc3 {7}
hxg2 {0} 30. Bxg2 {12} Rxf2 {0} 31. Rb2 {5} Rf8 {0} 32. Kd4 {2} g5 {24} 33. Bh3
{1} Rh8 {5} 34. Bf5 {2} Rh1 {2} 35. Rb6 {1} Kf7 {3} 36. Rg6 {2} Rg1 {3} 37. Ke5
{1} Rh1 {4} 38. Rxg5 {2} Kf8 {2} 39. Ke6 {1} Ra1 {3} 40. Ke5 {2} Rh1 {1} 41.
Be6 {1} Rh8 {1} 42. Rg8+ {7} Rxg8 {0} 43. Bxg8 {1} 1-0
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "lem"]
[Black "Rybka 2.2 32-bit"]
[Result "0-1"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkb1r/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "230"]
1. Nf3 {0} d5 {11} 2. d4 {2} Nc6 {10} 3. c4 {2} e6 {12} 4. Nc3 {2} Bb4 {11} 5.
cxd5 {3} exd5 {8} 6. Bg5 {2} f6 {14} 7. Bf4 {16} g5 {10} 8. Be3 {6} Bf5 {6} 9.
g3 {6} Be4 {0} 10. Bg2 {2} Bxc3+ {4} 11. bxc3 {2} Qd7 {18} 12. O-O {4} O-O-O {
12} 13. a4 {12} b6 {0} 14. a5 {13} Nxa5 {4} 15. Rxa5 {just to play it save 3}
bxa5 {1} 16. Qa1 {2} a4 {7} 17. Qa3 {2} h5 {4} 18. Ra1 {6} h4 {0} 19. Qxa4 {11}
Qxa4 {2} 20. Rxa4 {2} hxg3 {3} 21. hxg3 {3} Kb7 {9} 22. Nd2 {4} Bxg2 {0} 23.
Kxg2 {1} Rb8 {6} 24. Nb3 {2} Kc6 {1} 25. Nc5 {1} a5 {1} 26. Rxa5 {3} Ra8 {0}
27. Rxa8 {6} Rxa8 {1} 28. Kf3 {I couldn't imagine, that I wouldn't win this 6}
Kd6 {1} 29. Bd2 {28} Ra2 {0} 30. Ke3 {1} c6 {1} 31. f4 {7} gxf4+ {1} 32. gxf4 {
2} Ra1 {2} 33. Kd3 {2} f5 {1} 34. Be3 {11} Rd1+ {12} 35. Kc2 {1} Re1 {5} 36.
Kd2 {1} Ra1 {0} 37. Bf2 {2} Rh1 {1} 38. Kd3 {10} Rh3+ {6} 39. Kd2 {8} Rh1 {1}
40. Nd3 {3} Rh2 {1} 41. Bg3 {9} Rh3 {2} 42. Be1 {3} Rh7 {0} 43. Bf2 {
in a blitzgame this is not easy to win 11} Rh2 {2} 44. Be3 {1} Rg2 {5} 45. Ne5
{1} Rg7 {0} 46. Kd3 {1} Ra7 {3} 47. Bf2 {1} Ra1 {4} 48. Bh4 {1} Ra8 {1} 49. Bf6
{4} Rf8 {0} 50. Bg7 {1} Rg8 {3} 51. Bh8 {
go never for such cheap tactics against a comp. ahh also against a human 3} Ke6
{2} 52. Nf7 {16} Kxf7 {1} 53. Be5 {1} Ke6 {1} 54. e3 {1} Ra8 {1} 55. Kd2 {2}
Ra2+ {2} 56. Kd3 {0} Ra1 {1} 57. Kd2 {2} Rb1 {0} 58. Kd3 {0} Rb2 {2} 59. Bc7 {2
} Rb5 {0} 60. Be5 {1} Rb1 {1} 61. Bc7 {1} Rd1+ {2} 62. Ke2 {1} Ra1 {0} 63. Kd3
{1} Re1 {1} 64. Be5 {1} Rf1 {1} 65. Bc7 {1} Ra1 {1} 66. Be5 {1} Rd1+ {2} 67.
Ke2 {1} Rh1 {1} 68. Kd3 {1} Rg1 {0} 69. Bc7 {3} Rg3 {1} 70. Be5 {1} Rg2 {1} 71.
Bc7 {1} Rg1 {0} 72. Be5 {1} Rh1 {1} 73. Bc7 {1} Kd7 {2} 74. Be5 {1} Ra1 {1} 75.
Bf6 {0} Kd6 {0} 76. Be5+ {1} Ke7 {0} 77. Bg7 {1} Rb1 {1} 78. Be5 {1} Rd1+ {0}
79. Ke2 {1} Rh1 {1} 80. Kd3 {0} Rb1 {1} 81. Bg7 {1} Kd7 {1} 82. Be5 {1} Rb2 {2}
83. Bf6 {1} Kd6 {0} 84. Be5+ {1} Ke7 {1} 85. Bc7 {1} Ra2 {1} 86. Be5 {1} Ra8 {1
} 87. Bg7 {1} Ra1 {1} 88. Be5 {0} Rf1 {2} 89. Bg7 {1} Kd6 {0} 90. Be5+ {1} Kd7
{0} 91. Bg7 {0} Rd1+ {4} 92. Ke2 {1} Rb1 {3} 93. Be5 {1} Rb7 {1} 94. Kd3 {0}
Rb3 {7} 95. Bf6 {1} Ra3 {4} 96. Be5 {0} Ke6 {1} 97. Bg7 {1} Kf7 {1} 98. Be5 {0}
Rb3 {2} 99. Bc7 {0} Ke6 {0} 100. Be5 {0} Rb7 {3} 101. Bh8 {1} Rb2 {1} 102. Be5
{1} Rb5 {0} 103. Bg7 {1} c5 {1} 104. Be5 {
couldn't react to c5, cos I constantly had to premove 0} c4+ {0} 105. Kd2 {2}
Rb2+ {0} 106. Kd1 {1} Ra2 {0} 107. Bg7 {1} Ra3 {0} 108. Kd2 {1} Kd7 {0} 109.
Be5 {1} Ra2+ {0} 110. Kd1 {1} Ke7 {0} 111. Bg7 {1} Ra6 {1} 112. Be5 {1} Kf7 {0}
113. Bc7 {6} Kg6 {0} 114. Be5 {1} Ra2 {0} 115. Bd6 {3} Kh5 {
time - and lost anyway 0} 0-1
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "Rybka 2.2 32-bit"]
[Black "lem"]
[Result "1/2-1/2"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[PlyCount "95"]
1. Nf3 {9} d5 {12} 2. e3 {9} Nf6 {54} 3. b3 {5} g6 {3} 4. Bb5+ {8} Bd7 {3} 5.
Qe2 {5} Bg7 {7} 6. Bb2 {7} a6 {6} 7. Bxd7+ {17} Nbxd7 {5} 8. O-O {14} O-O {1}
9. c4 {8} dxc4 {2} 10. Qxc4 {7} c5 {11} 11. Rac1 {9} b5 {2} 12. Qh4 {8} Qb6 {5}
13. Bc3 {8} Rfe8 {37} 14. d3 {7} Nh5 {14} 15. d4 {5} cxd4 {6} 16. Bxd4 {2} Bxd4
{6} 17. Qxd4 {3} Qxd4 {1} 18. Nxd4 {3} Rac8 {2} 19. Rfd1 {2} Nb6 {5} 20. Nc6 {5
} Kg7 {8} 21. Kf1 {0} Rc7 {9} 22. Nb4 {9} Rxc1 {4} 23. Rxc1 {3} Rc8 {13} 24.
Rxc8 {2} Nxc8 {0} 25. Nxa6 {1} Kf6 {14} 26. a4 {4} bxa4 {2} 27. bxa4 {0} Ke6 {2
} 28. Ke2 {2} Kd6 {2} 29. Nb4 {1} Ng7 {6} 30. Kd3 {3} Ne8 {1} 31. Kc4 {6} Nc7 {
1} 32. e4 {25} Nb6+ {3} 33. Kb3 {0} e6 {5} 34. Nd3 {3} f5 {3} 35. a5 {2} Nd7 {5
} 36. exf5 {0} gxf5 {2} 37. Kc4 {2} Ne5+ {and white won't lose. I think
against any human I would have won even if he played perfectly. against a comp
there's a significant psycholgical disadvantage :-) 5} 38. Nxe5 {1} Kxe5 {0}
39. Kc5 {4} h5 {6} 40. Kc6 {3} Na6 {6} 41. Kb6 {2} Nb4 {3} 42. Kb5 {2} Nd5 {3}
43. Kc5 {1} Nc7 {9} 44. Kc6 {0} Na6 {1} 45. Kb6 {1} Nb4 {1} 46. Kb5 {1} Nd5 {1}
47. Kc5 {0} Nc7 {1} 48. Kc6 {7} 1/2-1/2
[Event "?"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "Rybka 2.2 32-bit"]
[Black "lem"]
[Result "1/2-1/2"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[PlyCount "79"]
1. Nc3 {16} d5 {4} 2. d4 {7} Nf6 {3} 3. Bf4 {19} e6 {2} 4. g3 {10} Bd6 {1} 5.
Bg2 {14} Bxf4 {2} 6. gxf4 {7} O-O {10} 7. Qd2 {6} b6 {1} 8. O-O-O {7} Bb7 {4}
9. Rhg1 {6} Nbd7 {4} 10. Kb1 {5} g6 {2} 11. a3 {8} c5 {1} 12. Bf3 {1} c4 {11}
13. h4 {9} a6 {2} 14. h5 {6} b5 {9} 15. Rg5 {6} Nb6 {22} 16. Rdg1 {8} Qd7 {13}
17. Qe3 {6} Qd6 {14} 18. f5 {3} exf5 {6} 19. Rxf5 {2} Ne4 {14} 20. Nxe4 {7}
dxe4 {1} 21. Bxe4 {3} Bxe4 {1} 22. Qxe4 {4} Rae8 {4} 23. Re5 {2} Rxe5 {1} 24.
dxe5 {1} Qd5 {2} 25. Qe3 {3} Na4 {8} 26. f4 {0} Rd8 {5} 27. hxg6 {7} hxg6 {2}
28. Qh3 {damn- what to do against the perpetual after Rxg6? 2} Qd4 {the other t
hreat was maybe Rh1 - so I secured the draw and felt like an idiot again. 40}
29. Rxg6+ {1} fxg6 {1} 30. Qe6+ {0} Kg7 {1} 31. Qf6+ {0} Kh7 {2} 32. Qe7+ {1}
Kh6 {2} 33. Qh4+ {2} Kg7 {1} 34. Qe7+ {1} Kg8 {2} 35. Qe6+ {1} Kg7 {1} 36. Qf6+
{0} Kh7 {1} 37. Qe7+ {1} Kh6 {2} 38. Qh4+ {1} Kg7 {0} 39. Qe7+ {2} Kh6 {0} 40.
Qh4+ {2} 1/2-1/2
[Event "Blitz:5'"]
[Site "?"]
[Date "2007.02.25"]
[Round "?"]
[White "lem"]
[Black "Rybka 2.2 32-bit"]
[Result "1/2-1/2"]
[Annotator ",lem"]
[SetUp "1"]
[FEN "rnbqkb1r/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "94"]
[TimeControl "300"]
1. Nf3 {0} d5 {10} 2. d4 {7} Nc6 {8} 3. c4 {2} e6 {12} 4. Nc3 {3} Bb4 {12} 5.
Bg5 {4} f6 {3} 6. Bf4 {3} g5 {28} 7. Be3 {2} dxc4 {7} 8. g3 {7} Bd7 {11} 9. Bg2
{12} Ne7 {7} 10. Nd2 {30} Bc6 {0} 11. Bxc6+ {6} Nxc6 {1} 12. O-O {4} Bxc3 {2}
13. bxc3 {1} Qd5 {9} 14. f3 {15} Rd8 {7} 15. Bf2 {5} O-O {4} 16. e4 {2} Qb5 {12
} 17. Rb1 {4} Qa5 {4} 18. Qc2 {8} b5 {5} 19. a4 {15} Qxa4 {6} 20. Qxa4 {16}
bxa4 {0} 21. Nxc4 {2} Rb8 {5} 22. Na3 {15} Rb3 {5} 23. Rxb3 {12} axb3 {2} 24.
Rb1 {4} Rb8 {4} 25. Kf1 {4} a5 {7} 26. Ke2 {3} a4 {9} 27. Kd3 {2} Kf7 {3} 28.
f4 {1} gxf4 {2} 29. gxf4 {0} Ne7 {5} 30. c4 {6} f5 {3} 31. e5 {1} Ng6 {3} 32.
Be3 {3} b2 {1} 33. Nb5 {17} Rxb5 {noooooooooooo 0} 34. cxb5 {2} a3 {0} 35. Kc2
{7} a2 {0} 36. Kxb2 {2} axb1=Q+ {0} 37. Kxb1 {1} Ne7 {2} 38. Kc2 {5} Ke8 {0}
39. Kd3 {6} Kd7 {0} 40. Kc4 {2} c6 {3} 41. bxc6+ {9} Kxc6 {2} 42. Bd2 {11} Nd5
{0} 43. Kd3 {2} Nb6 {2} 44. Kc3 {2} Kd5 {1} 45. Kd3 {1} Kc6 {1} 46. Kc3 {1} Kd5
{1} 47. Kd3 {0} Kc6 {if it plays on it'll win on time 2} 1/2-1/2
![[lu]](/mwf/v12/flags/lu.png "Luxembourg")
Hi,
1) It seems that handicaps can be given to much stronger players than I would have ever thought possible. (Two months ago, I would have told you that I can beat God at tournament time-control with an extra knight.)
do you think that games of a strong Rybka-assisted human giving a knight to an IM or weak GM would be interesting today or could be interesting in the near future? I would think that such a team would be a lot more effective at handicap play than a computer can be today.
Greetings,
Neuromancer
1) It seems that handicaps can be given to much stronger players than I would have ever thought possible. (Two months ago, I would have told you that I can beat God at tournament time-control with an extra knight.)
do you think that games of a strong Rybka-assisted human giving a knight to an IM or weak GM would be interesting today or could be interesting in the near future? I would think that such a team would be a lot more effective at handicap play than a computer can be today.
Greetings,
Neuromancer
Somehow, that doesn't sound as cool. It's not just a scientific experiment, it's a contest. For that, it's hard to beat man vs machine.
Vas
Vas
CCC is not the World.
> On CCC people already laugh about this "event". Mostly people wanna see real chess with both sides having a complete set of pieces on the board. So it seems that only the hardcore Rybka worshippers will watch.
Yes, that's what I'm saying... GM takes win all in all.
thanks whoever for organizing this. As a personal feeling, I would have preferred to see a "time odd" game.
It is hard to guess the outcome, since IMHO it strongly depends on the size and quality of LK opening book and JE preparation.
It is hard to guess the outcome, since IMHO it strongly depends on the size and quality of LK opening book and JE preparation.
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
Vas, I presume you have asked Chessbase/Playchess wether they would provide the Internet relay ?
I'd be surprised if Rybka manages to win this match, pawn odds against a GM is tough.
In reply to all the inquiries so far, here is some more information. I (Larry Kaufman) am the organizer of the match, it will be played at my house in Maryland using my quad computer. The announcement did not make it clear that each of the eight White pawns will be given in turn as the handicap, so no worries about repeat games. The opening book will be minimal, just enough to minimize the chance of prepared games.
The choice of the handicap was carefully made, to make an interesting match. Regarding alternate suggestions, time odds would not make for a close match unless the odds were huge; even with one hour for Ehlvest and five minutes for Rybka I'd bet on Rybka. I would love to have Rybka offer the traditional "pawn and move" (f7) handicap, but in my opinion Rybka is not yet quite strong enough to offer this handicap (the biggest possible pawn handicap) to a player like Ehlvest in a serious match and expect to win the match. Ehlvest was once (around 1990) in the World Top Five, and at age 44 he is hardly an old man. I do believe that within a couple years this handicap might be suitable for such a worthy opponent. As for knight odds, my best estimate now is that it is a suitable handicap for an average I.M. in blitz against Rybka, but in serious chess a strong G.M. would probably win nearly every game.
As for forecasts, mine is that Rybka will win by 2-1 with five draws. In my private testing Rybka has been able to defeat other computers at this handicap with considerably higher ratings than Ehlvest, but of course they are just playing normal chess, whereas Ehlvest can adapt his play to the handicap. So it should be quite close.
The choice of the handicap was carefully made, to make an interesting match. Regarding alternate suggestions, time odds would not make for a close match unless the odds were huge; even with one hour for Ehlvest and five minutes for Rybka I'd bet on Rybka. I would love to have Rybka offer the traditional "pawn and move" (f7) handicap, but in my opinion Rybka is not yet quite strong enough to offer this handicap (the biggest possible pawn handicap) to a player like Ehlvest in a serious match and expect to win the match. Ehlvest was once (around 1990) in the World Top Five, and at age 44 he is hardly an old man. I do believe that within a couple years this handicap might be suitable for such a worthy opponent. As for knight odds, my best estimate now is that it is a suitable handicap for an average I.M. in blitz against Rybka, but in serious chess a strong G.M. would probably win nearly every game.
As for forecasts, mine is that Rybka will win by 2-1 with five draws. In my private testing Rybka has been able to defeat other computers at this handicap with considerably higher ratings than Ehlvest, but of course they are just playing normal chess, whereas Ehlvest can adapt his play to the handicap. So it should be quite close.
![[kw]](/mwf/v12/flags/kw.png "Kuwait")
Great !!! Should be really interesting.
Thanks for the info and for organizing the event, and also for providing a strong anti-human version of rybka. It is remarkable that many people are enthusiastically engaged with rybka. As for time odds, I generally play against rybka with time and move odd: rybka 1minute (sometimes 1' + 1'') and ponder ON, me 30 minutes + 30''.
Anyway, I am pretty sure that in this case going pawn odd is the best choice. Losing when your adversary is largely time handicapped is really a bad feeling, worst than losing at pawn odd. But still, it would be interesting to see some time odd match in the future.
The formula with eight different starting positions makes the game hotter (and is certainly better for rybka than the classical f7-pawn and move odd). Could you tell us if Ehlvest received the opening book in advance?
Thx again.
Anyway, I am pretty sure that in this case going pawn odd is the best choice. Losing when your adversary is largely time handicapped is really a bad feeling, worst than losing at pawn odd. But still, it would be interesting to see some time odd match in the future.
The formula with eight different starting positions makes the game hotter (and is certainly better for rybka than the classical f7-pawn and move odd). Could you tell us if Ehlvest received the opening book in advance?
Thx again.
I think that it is obvious that Ehlvest did not receive the opening book in advance
because the target of the opening book is to prevent preperation.
quote:
"The opening book will be minimal, just enough to minimize the chance of prepared games."
Uri
because the target of the opening book is to prevent preperation.
quote:
"The opening book will be minimal, just enough to minimize the chance of prepared games."
Uri
Thanks for this providing this interesting experiment, Larry!
Personally I predict Ehlvest will not win any game. Probably some draws, lets say 5 wins + 3 draws for the machine.
As someone mentioned, why not relay the match on ICC/Playchess?
Next we want to see Larry stand for his word and betting any average IM 1000$ for beating Rybka-LK knight-odds :-)
With the computers getting too strong, proper odds games may be the future, to keep the challenge and to explore new fresh aspects of the game.
I would like to suggest another kind of odds game:
If you are allowed to do takebacks at any time (with the clock still running) how many do you have to use to draw or win against the program?
Regards,
John
Personally I predict Ehlvest will not win any game. Probably some draws, lets say 5 wins + 3 draws for the machine.
As someone mentioned, why not relay the match on ICC/Playchess?
Next we want to see Larry stand for his word and betting any average IM 1000$ for beating Rybka-LK knight-odds :-)
With the computers getting too strong, proper odds games may be the future, to keep the challenge and to explore new fresh aspects of the game.
I would like to suggest another kind of odds game:
If you are allowed to do takebacks at any time (with the clock still running) how many do you have to use to draw or win against the program?
Regards,
John
I will not be surprised.
I am not sure if pawn advantage is always significant and I suspect that sometimes the side without the pawn has significant compensation(for example without pawn e2 it is more easy for white to develop the pieces)
Uri
I am not sure if pawn advantage is always significant and I suspect that sometimes the side without the pawn has significant compensation(for example without pawn e2 it is more easy for white to develop the pieces)
Uri
Thanks Larry for this info.I hope it will turn out to be a good match.OH my earlier post I was a bit excited talking about such large knight odds but would be interesting for IM maybe as you say.This will be insightful and interesting.
Your chess friend in Maryland(Now snowy)
John
Your chess friend in Maryland(Now snowy)
John
It is certainly true that White will have some compensation for the missing pawn in all cases except the "f" pawn. However, in the case of "b" "c" "e" and "g" pawns, the compensation is at best one tempo, since simply moving that pawn would open all the same diagonals and also control some good squares. As the usual rule (which I think is reasonably accurate) states that three tempi equal a pawn, it is clear that these are all significant handicaps. The "d" pawn's removal gives slighty more comp, because the queen is on a half-open file, but this should hardly be worth more than the squares that would be controlled by a pawn on d4, so even in this case the compensation is clearly insufficient. Actually removing the "d" pawn is similar to playing a Smith-Morra Gambit a tempo down, except that the "c" pawns are off the board. The Smith-Morra is considered dubious at GM level normally; a tempo down it should be quite unsound.
This leaves the two edge pawn handicaps. Only in these cases can it be seriously argued that the compensation is anywhere near to being enough. Edge pawns are worth less to start, and the semi-open file for the rook is quite nice. In my opinion Black will still have an advantage even in these two cases, but his advantage may be only slightly more than the advantage White normally has in an even game.
As for some other posts, anyone who plays Rybka at handicap on a 32 bit single processor must bear in mind that the 64 bit MP on a quad is roughly a class stronger. As for playing blitz with an average IM at knight odds, bear in mind that "average IM" does not mean "average IM who plays blitz all day on the internet"; clearly among players of any level there are blitz specialists and those who only play blitz occasionally. As for takebacks as a handicap, this is indeed a valid idea, but I'm afraid that if the player is a mere 2600 Elo GM he would need so many takebacks to have equal chances that it would be laughable. Maybe for Kramnik or Anand or Topalov it might be reasonable.
This leaves the two edge pawn handicaps. Only in these cases can it be seriously argued that the compensation is anywhere near to being enough. Edge pawns are worth less to start, and the semi-open file for the rook is quite nice. In my opinion Black will still have an advantage even in these two cases, but his advantage may be only slightly more than the advantage White normally has in an even game.
As for some other posts, anyone who plays Rybka at handicap on a 32 bit single processor must bear in mind that the 64 bit MP on a quad is roughly a class stronger. As for playing blitz with an average IM at knight odds, bear in mind that "average IM" does not mean "average IM who plays blitz all day on the internet"; clearly among players of any level there are blitz specialists and those who only play blitz occasionally. As for takebacks as a handicap, this is indeed a valid idea, but I'm afraid that if the player is a mere 2600 Elo GM he would need so many takebacks to have equal chances that it would be laughable. Maybe for Kramnik or Anand or Topalov it might be reasonable.
Yep, when I play rybka time odds, I do that on a 32-bit old laptop. And still, I lose most of the games, and basically I never win. I paste here two games of Ehlvest against Zappa.
[Event "Ehlvest-Zappa(C)"]
[Site "New York, USA"]
[Date "2005.07.08"]
[Round "?"]
[White "Zappa(C)"]
[Black "GM_Ehlvest"]
[Result "1-0"]
[Opening "Robatsch (modern) defense"]
[ECO "B06"]
[NIC "KF.06"]
[Time "10:32:11"]
[TimeControl "7200+0"]
1. e4 g6 2. d4 d6 3. Nc3 c6 4. Be3 Nf6 5. Be2 Nbd7 6. f4 b5 7. a3 Nb6 8. d5 Bb7 9. Nf3 a6 10. dxc6 Bxc6 11. Qd4 Na4? (of course Rb8 is the only move) 12. Nxa4 bxa4 13. Ng5 e6 14. Qc3?! (wrong order of moves, f5 is better) Bb7 15. f5 exf5 16. Bc4 d5 17. exd5 Bg7 18. Qe5+?! (poor move since Bc4! was smashing through) Qe7 19. Qxe7+ Kxe7 20. O-O-O Kd7 21. Nxf7 Rhe8 22. Rhe1 Ng4? 23. Bc5 Ne5 24. Nd6 Nxc4 25. Nxc4 Rad8 26. Re6 Bh6+ 27. Kb1 White wins 1-0
[Event "Ehlvest-Zappa(C)"]
[Site "New York, USA"]
[Date "2005.07.09"]
[Round "?"]
[White "GM_Ehlvest"]
[Black "Zappa(C)"]
[Result "1/2-1/2"]
[Opening "Ruy Lopez: closed, Bogolyubov variation"]
[ECO "C91"]
[NIC "RL.20"]
[Time "10:51:14"]
[TimeControl "7200+0"]
1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O 8. d4 d6 9. c3 Bg4 10. Be3 Bh5 11. h3 Na5 12. Bc2 Nc4 13. Bc1 Nd7 14. b3 Ncb6 15. Nbd2 c5 16. g4 Bg6 17. Nf1 cxd4 18. cxd4 Rc8 19. Re2 exd4 20. Nxd4 d5 21. e5 Bxc2 22. Nxc2 Re8 23. Nd4 Bc5 24. e6 fxe6 25. Nxe6 Qe7 26. Bb2 Bxf2+ 27. Rxf2 Qxe6 28. Qd4 Qe5 29. Rd1 Qxd4 30. Bxd4 Rf8 31. Rxf8+ Kxf8 32. Ne3 g6 33. Re1 Rc6 34. Kf2 Kf7 35. Kf3 Nc8 36. Ke2 Ne7 37. Kd3 Nc5+? 38. Bxc5 Rxc5 39. Kd4 Rc6
Game drawn 1/2-1/2
[Event "Ehlvest-Zappa(C)"]
[Site "New York, USA"]
[Date "2005.07.08"]
[Round "?"]
[White "Zappa(C)"]
[Black "GM_Ehlvest"]
[Result "1-0"]
[Opening "Robatsch (modern) defense"]
[ECO "B06"]
[NIC "KF.06"]
[Time "10:32:11"]
[TimeControl "7200+0"]
1. e4 g6 2. d4 d6 3. Nc3 c6 4. Be3 Nf6 5. Be2 Nbd7 6. f4 b5 7. a3 Nb6 8. d5 Bb7 9. Nf3 a6 10. dxc6 Bxc6 11. Qd4 Na4? (of course Rb8 is the only move) 12. Nxa4 bxa4 13. Ng5 e6 14. Qc3?! (wrong order of moves, f5 is better) Bb7 15. f5 exf5 16. Bc4 d5 17. exd5 Bg7 18. Qe5+?! (poor move since Bc4! was smashing through) Qe7 19. Qxe7+ Kxe7 20. O-O-O Kd7 21. Nxf7 Rhe8 22. Rhe1 Ng4? 23. Bc5 Ne5 24. Nd6 Nxc4 25. Nxc4 Rad8 26. Re6 Bh6+ 27. Kb1 White wins 1-0
[Event "Ehlvest-Zappa(C)"]
[Site "New York, USA"]
[Date "2005.07.09"]
[Round "?"]
[White "GM_Ehlvest"]
[Black "Zappa(C)"]
[Result "1/2-1/2"]
[Opening "Ruy Lopez: closed, Bogolyubov variation"]
[ECO "C91"]
[NIC "RL.20"]
[Time "10:51:14"]
[TimeControl "7200+0"]
1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O 8. d4 d6 9. c3 Bg4 10. Be3 Bh5 11. h3 Na5 12. Bc2 Nc4 13. Bc1 Nd7 14. b3 Ncb6 15. Nbd2 c5 16. g4 Bg6 17. Nf1 cxd4 18. cxd4 Rc8 19. Re2 exd4 20. Nxd4 d5 21. e5 Bxc2 22. Nxc2 Re8 23. Nd4 Bc5 24. e6 fxe6 25. Nxe6 Qe7 26. Bb2 Bxf2+ 27. Rxf2 Qxe6 28. Qd4 Qe5 29. Rd1 Qxd4 30. Bxd4 Rf8 31. Rxf8+ Kxf8 32. Ne3 g6 33. Re1 Rc6 34. Kf2 Kf7 35. Kf3 Nc8 36. Ke2 Ne7 37. Kd3 Nc5+? 38. Bxc5 Rxc5 39. Kd4 Rc6
Game drawn 1/2-1/2
![[in]](/mwf/v12/flags/in.png "India")
Hi,
Is it because no GM is ready to take on Rybka at its full strength. It would ne nice to see Anand or Kramnik taking on Rybka at its fullest strength.
JP Singh
Is it because no GM is ready to take on Rybka at its full strength. It would ne nice to see Anand or Kramnik taking on Rybka at its fullest strength.
JP Singh
![[ph]](/mwf/v12/flags/ph.png "Philippines")
Anand can't even win against an old program called Rebel?! What more with Rybka
anand played against rebel blitz matches , i think anand can play good chess against rybka
Mr. Marcus
1) Anand vs Rebel was long time back when Anand lost Blitz and won Standard time control for your kind information
2) Dont underestimate Rebel(Prodeo) it was and is one of the best antihuman chess program around
3) If Kamsky and many GMs can get draws against Rybka , i think Anand can fare better
Regards
Vikrant
1) Anand vs Rebel was long time back when Anand lost Blitz and won Standard time control for your kind information
2) Dont underestimate Rebel(Prodeo) it was and is one of the best antihuman chess program around
3) If Kamsky and many GMs can get draws against Rybka , i think Anand can fare better
Regards
Vikrant
I see no reason to give kramnik to play against computers because kramnik simply does not want to win against them.
player who resign in drawn position against Fritz and does stupid mistakes against computers that even I expect myself not to do in serious games(including missing mate in 1) is not a player who want to win.
Uri
player who resign in drawn position against Fritz and does stupid mistakes against computers that even I expect myself not to do in serious games(including missing mate in 1) is not a player who want to win.
Uri
If the Rybka loses a game or even this match there will always be an excuse to why - a missing pawn! But if the GM cannot tame Rybka with a pawn deficit this will be absolutely mean ... This is a lose/lose situation no matter how it is carried out unless there is no handicap whatsoever, and both sides are equaled. I am sure the GM knows what pawn is going to be missing and have formulated some strategy against it, and adversely this 'private' version of Rybka LK is probably specifically programmed for handicaps. However, this will be interesting.
There are 8 games and in everyone of them different pawn is going to be missing so
"I am sure the GM knows what pawn is going to be missing" is not relevant here.
I think that it may be interesting to know what is the result of rybka against herself in these conditions in order to evaluate if missing pawn is a significant handicap(I am not sure if it is a significant handicap).
Uri
"I am sure the GM knows what pawn is going to be missing" is not relevant here.
I think that it may be interesting to know what is the result of rybka against herself in these conditions in order to evaluate if missing pawn is a significant handicap(I am not sure if it is a significant handicap).
Uri
I had Rybka play a thousand games with herself (using the randomizer) at 8 ply at each of the pawn handicaps. These are very fast games, faster even than 1' (bullet) chess, but Rybka is still playing at around the level of a master at 40/2. The two edge pawns proved to be just moderate handicaps, on average 67 Elo, roughly double the normal White advantage in chess. The other six pawns were all major handicaps, averaging 171 Elo, nearly a class (f2 was the worst to remove). It was also clear from the trend (comparing to even shorter searches) that at longer time controls, the handicap would be greater, since the strength of the receiving player counts for more than the strength of the receiving player. I also did some tests against strong s.p. programs like Fritz 10 and Shredder 10 at time controls ranging from blitz to 25' +10", which proved that this handicap (all 8 white pawns one by one) is roughly fair between Rybka 64 bit MP quad and these SP programs (2.2n2 lost narrowly, 2.3 LK won narrowly). Since the rating gap between 2.3 LK and these oher programs is almost exactly a class on the last CEGT 40/20 list, and allowing another 40 point or so for my quad, it is clear that the handicap is somewhat more than one class between computers.
However with a human involved, it should be a greater handicap, since he can modify his "programming" for the handicap. Therefore it should be quite a close match, I feel, though we are optimistic!
However with a human involved, it should be a greater handicap, since he can modify his "programming" for the handicap. Therefore it should be quite a close match, I feel, though we are optimistic!
Hi,
I had Rybka play a thousand games with herself (using the randomizer) at 8 ply at each of the pawn handicaps. These are very fast games, faster even than 1' (bullet) chess, but Rybka is still playing at around the level of a master at 40/2.
I wonder if you or Vasik have ever thought about trying to use UCT search in Rybka for special purposes like very long time controls or automatic opening discovery or analysis of positions where fortress-type motives are likely to appear? UCT is an algorithm based on statistics-driven exploration of a partial gametree built from randomized playouts of entire games, which has recently seen very successful application in Go. In go, it has been successful primarily because of the ability of sampling-based methods to see things which happen far behind the horizon of an ordinary search. If, as you suggest, Rybka is capable of reaching master-level strength with a relatively shallow search, then I could imagine that using such shallow searches for move ordering in unexplored nodes of the UCT tree might provide a very effective sampling policy in chess. While I intuitively tend to think that doing this type of search will in chess not in general beat just going deeper with the standard alpha-beta search for the same amount of time (because just going deeper with the standard search is going to discover tactics faster), I could imagine that it could provide interesting information in situations where it is particularly important to discover strategically deep features of a position which are not detected by the evaluation function.
Any thoughts on this? :)
Greetings,
Neuromancer
I had Rybka play a thousand games with herself (using the randomizer) at 8 ply at each of the pawn handicaps. These are very fast games, faster even than 1' (bullet) chess, but Rybka is still playing at around the level of a master at 40/2.
I wonder if you or Vasik have ever thought about trying to use UCT search in Rybka for special purposes like very long time controls or automatic opening discovery or analysis of positions where fortress-type motives are likely to appear? UCT is an algorithm based on statistics-driven exploration of a partial gametree built from randomized playouts of entire games, which has recently seen very successful application in Go. In go, it has been successful primarily because of the ability of sampling-based methods to see things which happen far behind the horizon of an ordinary search. If, as you suggest, Rybka is capable of reaching master-level strength with a relatively shallow search, then I could imagine that using such shallow searches for move ordering in unexplored nodes of the UCT tree might provide a very effective sampling policy in chess. While I intuitively tend to think that doing this type of search will in chess not in general beat just going deeper with the standard alpha-beta search for the same amount of time (because just going deeper with the standard search is going to discover tactics faster), I could imagine that it could provide interesting information in situations where it is particularly important to discover strategically deep features of a position which are not detected by the evaluation function.
Any thoughts on this? :)
Greetings,
Neuromancer
Hi Neuromancer,
I haven't had time to read up on UCT yet, but from your short description it sounds quite a bit like a fairly new learning method some have dubbed Sherwin learning. The idea is roughly to do this:
1) Play out games from a position.
2) Give moves in the resulting tree evaluation bonuses or penalties based on the results from these play-outs.
3) Have these bonuses and penalties affect the move selection in the subsequent play-outs (as in, a move has scored 35% in the playouts, while search says 45% - then maybe the combined score of the move will be 40%).
4) Possibly, a small random move bonus/penalty could also be added to the move selection when a position has been reached more than once. This in order to catch moves close to the root that could be good alternatives even when the first chosen move by search gets a bonus from the playouts (as opposed to a penalty).
I think this learning method is promising both as an opening discovery/exploration tool, and as a tool to reach a more precise evaluation of a position due to the discovery of strategically deep features. The new Rybka Randomizer is another more straight-forward step in this direction.
I haven't had time to read up on UCT yet, but from your short description it sounds quite a bit like a fairly new learning method some have dubbed Sherwin learning. The idea is roughly to do this:
1) Play out games from a position.
2) Give moves in the resulting tree evaluation bonuses or penalties based on the results from these play-outs.
3) Have these bonuses and penalties affect the move selection in the subsequent play-outs (as in, a move has scored 35% in the playouts, while search says 45% - then maybe the combined score of the move will be 40%).
4) Possibly, a small random move bonus/penalty could also be added to the move selection when a position has been reached more than once. This in order to catch moves close to the root that could be good alternatives even when the first chosen move by search gets a bonus from the playouts (as opposed to a penalty).
I think this learning method is promising both as an opening discovery/exploration tool, and as a tool to reach a more precise evaluation of a position due to the discovery of strategically deep features. The new Rybka Randomizer is another more straight-forward step in this direction.
This type of technique does seem promising. At the moment, it needs human direction and a very long time control. I don't know how far away we are from actually being able to use it automatically during game play.
Vas
Vas
I recall your mentioning of a tool like Rybka randomizer possibly being used mid-search in order to, for example, figure out if an opposite colored bishop ending being a draw. This reminded me of stuff I have read about Back Gammon engines. Essentially, some guy managed to make a very, very strong Back Gammon engine by use of some neural network methods. Then, in order to figure out just how strong it was, and where it had its weaknesses, it was mentioned that one could take a position and do a play-out by any sufficiently strong program. The general agreement was that this play-out would then give a correct evaluation of the position. Since then, I have been wondering at what point in hardware advancement it would be possible to make a "perfect" Back Gammon program that just relied on this method.
As to the question about how far we are from being able to use it automatically during game play (in chess), I think that multi-core systems certainly lead us in that direction. Say you have 128 cores. Those last 64 cores may give us a 1.7 speedup. But they could also be used to, one by one, play out some clearly critical position and estimate its evaluation. It's still highly speculative, but anyway, I think the trend is in that direction.
Whatever it is, I think Sherwin learning is a fairly-easy-to-implement feature that should not be underestimated :-)
As to the question about how far we are from being able to use it automatically during game play (in chess), I think that multi-core systems certainly lead us in that direction. Say you have 128 cores. Those last 64 cores may give us a 1.7 speedup. But they could also be used to, one by one, play out some clearly critical position and estimate its evaluation. It's still highly speculative, but anyway, I think the trend is in that direction.
Whatever it is, I think Sherwin learning is a fairly-easy-to-implement feature that should not be underestimated :-)
1
2
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill