Komodo5.1 v Houdini3 Match - Long Time Controls
Moderators: hgm, Rebel, chrisw
-
- Posts: 1280
- Joined: Tue Aug 18, 2009 3:06 am
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
I just noticed this match you are running. Great job Sean! I look foward to playing over the games.
-
- Posts: 1187
- Joined: Wed Jan 06, 2010 3:11 pm
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
Well Don I am sorry to come back on this, but I just find it funny that you probably forgot what the probably ''not just someone'' Larry wrote earlier on 19 july: http://www.talkchess.com/forum/viewtopi ... 67&t=48566Don wrote:Here is my take:kranium wrote:Sean-
to recap...
this tournament has:
no book move limit
(lines sometimes running past move 30 or more)
no position reversal of color
(a cardinal sin in testing)
and of course your very powerful "Large Neutral Opening Books"
(which have been shown to end out of book with one side getting a huge (almost full pawn) advantage?)
what's really laughable is this tournament's credibility
Some of these issues are not ideal but they are not definitely no "cardinal sin" either. There is nothing improper about this test and each tester has their own way.Code: Select all
1. long book - short is preferred but but my own testing shows it's not that as big a deal as I used to think. It does slightly favor the weaker player. 2. long time control - this favors Komodo (relative to short time control) 3. no book limit - bigger deal that point 1 if it takes you out of the most interesting part of the game. Favors the weaker program. 4. no position reversal - no issue here except for statistical noise as this requires more games to be equally fair. In a short match this favors the weaker player and asymptotically makes no difference as the match gets longer. 5. Note that choice of openings is "unfair" in the sense that some programs do better with some openings - this is mostly solved by more games as is point 1 and 4 assuming more games means more different kinds of openings. 6. Openings that end with huge advantage (or drawish positions.) Like points 4 and 5 this wastes time and requires more games to get past the extra statistical noise. It also minimizes the chances of a victory for the stronger program given a match of the same size. And the difference in match score will be reduced.
The primary thing about these conditions is not unfairness but just the fact that it would require a lot more games to resolve who is the better program - that doesn't mean it's "unfair" it just means there is more noise in the test. For example Houdini could have gotten all the winning positions - over enough games it will average out. The only thing that could be construed as a "true" fairness issue is time control, but ALL tests have that issue - depending on the test and hardware and time control some programs will be favored over others and there is nothing you can do about that. Why do you think we like the long time control tests better?
If you take an opening that ends with a big advantage, you will often find that one program will win it and another will only manage to draw so I don't have any issue with ending book with at least a few positions where one side has a fairly large advantage as long as the advantage is not so much as to make the win almost a sure thing for both programs.
My only real complaint here is the reversal of color issue - unless you are going to play at least many hundreds of games that does introduce a lot of statistical noise, especially when there are a number of unbalanced positions and you are playing a short match. Someone said earlier that this make the results "random" which is a very gross exaggeration and is simply not the case. Every match is "random" in a sense that no matter the difference in ELO the stronger play might lose any given game but that is not uniformly random and neither is having some unbalanced positions.
Summary - most of the condition of this match favor Komodo for statistical reasons only - but that advantage would asymptotically be reduced to nothing if Sean plays enough games. In other words Houdini will still win this match if Sean plays enough games assuming that Houdini is the superior program at this hardware and time control which I am not disputing.
So everyone just shut up and enjoy the match!!
.."I would like to know why this match or any other match is run without reversing colors after each game? No matter if the book is 2 moves or 30, it is the only fair way to test and greatly reduces the luck factor. I can't think of any argument against doing this."
-
- Posts: 10282
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
The reason is very simple.beram wrote:Well Don I am sorry to come back on this, but I just find it funny that you probably forgot what the probably ''not just someone'' Larry wrote earlier on 19 july: http://www.talkchess.com/forum/viewtopi ... 67&t=48566Don wrote:Here is my take:kranium wrote:Sean-
to recap...
this tournament has:
no book move limit
(lines sometimes running past move 30 or more)
no position reversal of color
(a cardinal sin in testing)
and of course your very powerful "Large Neutral Opening Books"
(which have been shown to end out of book with one side getting a huge (almost full pawn) advantage?)
what's really laughable is this tournament's credibility
Some of these issues are not ideal but they are not definitely no "cardinal sin" either. There is nothing improper about this test and each tester has their own way.Code: Select all
1. long book - short is preferred but but my own testing shows it's not that as big a deal as I used to think. It does slightly favor the weaker player. 2. long time control - this favors Komodo (relative to short time control) 3. no book limit - bigger deal that point 1 if it takes you out of the most interesting part of the game. Favors the weaker program. 4. no position reversal - no issue here except for statistical noise as this requires more games to be equally fair. In a short match this favors the weaker player and asymptotically makes no difference as the match gets longer. 5. Note that choice of openings is "unfair" in the sense that some programs do better with some openings - this is mostly solved by more games as is point 1 and 4 assuming more games means more different kinds of openings. 6. Openings that end with huge advantage (or drawish positions.) Like points 4 and 5 this wastes time and requires more games to get past the extra statistical noise. It also minimizes the chances of a victory for the stronger program given a match of the same size. And the difference in match score will be reduced.
The primary thing about these conditions is not unfairness but just the fact that it would require a lot more games to resolve who is the better program - that doesn't mean it's "unfair" it just means there is more noise in the test. For example Houdini could have gotten all the winning positions - over enough games it will average out. The only thing that could be construed as a "true" fairness issue is time control, but ALL tests have that issue - depending on the test and hardware and time control some programs will be favored over others and there is nothing you can do about that. Why do you think we like the long time control tests better?
If you take an opening that ends with a big advantage, you will often find that one program will win it and another will only manage to draw so I don't have any issue with ending book with at least a few positions where one side has a fairly large advantage as long as the advantage is not so much as to make the win almost a sure thing for both programs.
My only real complaint here is the reversal of color issue - unless you are going to play at least many hundreds of games that does introduce a lot of statistical noise, especially when there are a number of unbalanced positions and you are playing a short match. Someone said earlier that this make the results "random" which is a very gross exaggeration and is simply not the case. Every match is "random" in a sense that no matter the difference in ELO the stronger play might lose any given game but that is not uniformly random and neither is having some unbalanced positions.
Summary - most of the condition of this match favor Komodo for statistical reasons only - but that advantage would asymptotically be reduced to nothing if Sean plays enough games. In other words Houdini will still win this match if Sean plays enough games assuming that Houdini is the superior program at this hardware and time control which I am not disputing.
So everyone just shut up and enjoy the match!!
.."I would like to know why this match or any other match is run without reversing colors after each game? No matter if the book is 2 moves or 30, it is the only fair way to test and greatly reduces the luck factor. I can't think of any argument against doing this."
Sean Evans is not interested in testing in a fair way.
Sean prefer komodo and not houdini and increasing the luck factor
has good chances to improve the result of the weaker program that he prefers.
Edit:Note that I noticed that sean started this thread only after komodo won the first game.
I suspect that we were going to read nothing about this match in case that houdini3 was the leader in the first games.
-
- Posts: 1777
- Joined: Thu Jun 05, 2008 10:58 pm
- Location: Canada
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
Code: Select all
(120m+10s)/40+(120m+10s)/40+(120m+10s) 0
1 Komodo5.1 +16/=48/-16 50.00% 40.0/80 1600.00
2 Houdini3 +16/=48/-16 50.00% 40.0/80 1600.00
[pgn]
[Event "(120m+10s)/40+(120m+10s)/40+(120m+10s)"]
[Site "?"]
[Date "2013.08.02"]
[Round "80"]
[White "Komodo5.1"]
[Black "Houdini3"]
[Result "1/2-1/2"]
[ECO "B33"]
[Annotator "16... 0.31"]
[PlyCount "140"]
[TimeControl "40/7200+10:40/7200+10:7200+10"]
{2412MB, Combinedbooks.ctg, Laptop1} 1. e4 {0} c5 {B 0} 2. Nf3 {0} Nc6 {B 0} 3.
d4 {0} cxd4 {B 0} 4. Nxd4 {0} Nf6 {B 0} 5. Nc3 {0} e5 {B 0} 6. Ndb5 {0} d6 {B 0
} 7. Bg5 {0} a6 {B 0} 8. Na3 {0} b5 {B 0} 9. Nd5 {0} Be7 {B 0} 10. Bxf6 {0}
Bxf6 {B 0} 11. c3 {0} Ne7 {B 0} 12. Nxf6+ {0} gxf6 {B 0} 13. Bd3 {0} Bb7 {B 0}
14. Qe2 {0} d5 {B 0} 15. O-O-O {0} Qc7 {B 0 last book move} 16. exd5 {0} Bxd5 {
0.31/26 305} 17. Be4 {353} O-O-O {0.28/27 64} 18. Bxd5 {819} Nxd5 {0.28/29 0}
19. Qe4 {813 (g3)} Qc6 {0.25/26 419} 20. Nc2 {16} Kc7 {0.32/26 327} 21. g3 {
131 (Rhf1)} h5 {0.25/25 548} 22. h4 {641 (Rhe1)} Rd6 {0.25/28 350} 23. Rhf1 {
518 (Rd3)} Rhd8 {0.23/26 294} 24. Qe2 {51 (Rfe1)} Kb8 {0.16/26 557} 25. Kb1 {
389 (Rfe1)} Qd7 {0.25/27 225} 26. Rc1 {249} Qf5 {0.17/28 234} 27. Ka1 {131}
R8d7 {0.25/27 215} 28. f4 {470 (c4)} Qg4 {0.16/25 307} 29. Qxg4 {142} hxg4 {0.
17/27 0} 30. f5 {910 (fxe5)} Kc7 {0.22/29 629} 31. Rfe1 {101} Kd8 {0.17/30 237}
32. h5 {313} Ke7 {0.21/31 175} 33. Re4 {76} Kf8 {0.15/30 304} 34. Rxg4 {32} Ne7
{0.13/32 419} 35. h6 {161} Rd2 {0.12/33 415} 36. Ne3 {89} Rh2 {0.12/33 306} 37.
Rh4 {93} Rxh4 {0.12/33 540} 38. gxh4 {0} Kg8 {0.10/32 3} 39. Ng4 {422 (Rg1+)}
Rd6 {0.13/33 536} 40. Kb1 {78} Kh7 {0.19/31 111} 41. Rf1 {206} Rc6 {0.12/31 0}
42. Kc2 {355} b4 {0.13/31 0} 43. Rf3 {296} a5 {0.20/30 282} 44. Kd2 {23} Ng8 {
0.11/28 208} 45. b3 {451} Nxh6 {0.21/31 0} 46. Nxh6 {132} Kxh6 {0.33/32 79} 47.
c4 {76} Kh5 {0.39/30 122} 48. Rf1 {124 (Rh3)} Kxh4 {0.20/33 195} 49. Ke3 {405}
a4 {0.26/35 0} 50. Rc1 {643 (Rh1+)} Kg5 {0.26/32 227} 51. Ke4 {66} Rc5 {0.91/
33 1647} 52. bxa4 {0} Ra5 {0.88/28 163} 53. Rc2 {201} Rxa4 {0.81/30 0} 54. Rg2+
{66} Kh4 {0.94/29 222} 55. Kd5 {35} Ra3 {1.03/29 136} 56. c5 {40} b3 {0.86/30
116} 57. axb3 {42} Rxb3 {0.89/30 0} 58. Rc2 {745} Rd3+ {0.44/37 375} 59. Ke4 {
175} Rd4+ {0.02/38 0} 60. Kf3 {308 (Ke3)} Rd3+ {0.02/42 142} 61. Ke4 {308 (Ke2)
} Rd4+ {0.02/43 43} 62. Ke3 {62 (Kf3)} Rd7 {0.02/36 192} 63. c6 {97 (Rc4+)} Rd8
{0.02/39 155} 64. c7 {83} Rc8 {0.02/38 28} 65. Ke4 {92 (Kd3)} Kg5 {0.02/37 197}
66. Rc1 {347 (Rc4)} Kg4 {0.02/39 551} 67. Rc6 {125 (Rc2)} Kg5 {0.02/36 157} 68.
Rc1 {142 (Rc5)} Kg4 {0.02/62 143} 69. Rg1+ {560 (Rc2)} Kh4 {0.02/40 276} 70.
Rc1 {111 (Rh1+)} Kg4 {0.02/55 146} 1/2-1/2
[/pgn]
-
- Posts: 1777
- Joined: Thu Jun 05, 2008 10:58 pm
- Location: Canada
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
Code: Select all
(120m+10s)/40+(120m+10s)/40+(120m+10s) 0
1 Komodo5.1 +16/=49/-16 50.00% 40.5/81 1640.25
2 Houdini3 +16/=49/-16 50.00% 40.5/81 1640.25
[pgn]
[Event "(120m+10s)/40+(120m+10s)/40+(120m+10s)"]
[Site "?"]
[Date "2013.08.02"]
[Round "81"]
[White "Houdini3"]
[Black "Komodo5.1"]
[Result "1/2-1/2"]
[ECO "B97"]
[Annotator "27. -0.30"]
[PlyCount "98"]
[TimeControl "40/7200+10:40/7200+10:7200+10"]
{2412MB, Combinedbooks.ctg, Laptop1} 1. e4 {B 0} c5 {0} 2. Nf3 {B 0} d6 {0} 3.
d4 {B 0} cxd4 {0} 4. Nxd4 {B 0} Nf6 {0} 5. Nc3 {B 0} a6 {0} 6. Bg5 {B 0} e6 {0}
7. f4 {B 0} Qb6 {0} 8. Qd2 {B 0} Qxb2 {0} 9. Rb1 {B 0} Qa3 {0} 10. e5 {B 0}
dxe5 {0} 11. fxe5 {B 0} h6 {0} 12. Bh4 {B 0} g5 {0} 13. exf6 {B 0} gxh4 {0} 14.
Be2 {B 0} Qa5 {0} 15. O-O {B 0} Nd7 {0} 16. Kh1 {B 0} Qg5 {0} 17. Rf4 {B 0} e5
{0} 18. Nd5 {B 0} exd4 {0} 19. Qxd4 {B 0} Kd8 {0} 20. Rd1 {B 0} h3 {0} 21. g3 {
B 0} Bd6 {0} 22. Nb6 {B 0} Bc5 {0} 23. Qb2 {B 0} Kc7 {0} 24. Nxd7 {B 0} Rd8 {0}
25. Nxc5 {B 0} Rxd1+ {0} 26. Bxd1 {B 0 last book move} Qxc5 {0} 27. Qd4 {-0.30/
31 1323} Qxd4 {606} 28. Rxd4 {-0.23/33 0} Be6 {193} 29. a3 {-0.19/34 1816} Rg8
{856} 30. Kg1 {-0.19/33 0} Rg6 {500} 31. Kf2 {-0.20/31 391} Rxf6+ {167} 32. Ke1
{-0.20/31 232} Rg6 {1243} 33. Rh4 {-0.16/31 0} Rg5 {1537} 34. g4 {-0.15/31 0}
Kd6 {652 (Re5+)} 35. Rxh6 {-0.14/29 738} Rg6 {159} 36. Rxh3 {-0.10/30 582} Bxg4
{164} 37. Rd3+ {-0.11/31 1284} Ke6 {0} 38. Re3+ {-0.12/29 389} Kd7 {50 (Kf6)}
39. h4 {-0.06/30 372} f5 {359 (Bxd1)} 40. h5 {-0.06/32 368} Rg5 {311} 41. h6 {
-0.05/32 0} Rh5 {451} 42. Bxg4 {-0.04/35 0} fxg4 {160} 43. Rg3 {-0.02/34 45}
Rxh6 {233 (Rh4)} 44. Rxg4 {-0.02/37 295} Rd6 {505 (Rh2)} 45. Ke2 {-0.02/39 246}
b5 {341} 46. c4 {-0.02/45 0} Kc6 {167} 47. cxb5+ {-0.02/49 73} axb5 {80} 48. a4
{-0.01/54 483} bxa4 {98 (Re6+)} 49. Rxa4 {0.00/0 0} Kd5 {0} 1/2-1/2
[/pgn]
-
- Posts: 1777
- Joined: Thu Jun 05, 2008 10:58 pm
- Location: Canada
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
Code: Select all
(120m+10s)/40+(120m+10s)/40+(120m+10s) 0
1 Komodo5.1 +4 +17/=49/-16 50.61% 41.5/82
2 Houdini3 -4 +16/=49/-17 49.39% 40.5/82
[pgn]
[Event "(120m+10s)/40+(120m+10s)/40+(120m+10s)"]
[Site "?"]
[Date "2013.08.02"]
[Round "82"]
[White "Komodo5.1"]
[Black "Houdini3"]
[Result "1-0"]
[ECO "B92"]
[Annotator "11... 0.02"]
[PlyCount "123"]
[TimeControl "40/7200+10:40/7200+10:7200+10"]
{2412MB, Combinedbooks.ctg, Laptop1} 1. e4 {0} c5 {B 0} 2. Nf3 {0} d6 {B 0} 3.
d4 {0} cxd4 {B 0} 4. Nxd4 {0} Nf6 {B 0} 5. Nc3 {0} a6 {B 0} 6. Be2 {0} e5 {B 0}
7. Nb3 {0} Be7 {B 0} 8. Bg5 {0} Nc6 {B 0} 9. O-O {0} O-O {B 0} 10. Qd3 {0} a5 {
B 0 last book move} 11. a3 {574} a4 {0.02/29 313} 12. Nd2 {150} Nd4 {0.09/28
116} 13. Nf3 {361} Nxe2+ {0.08/30 0} 14. Qxe2 {140} Be6 {0.13/28 98} 15. Rfd1 {
606} Ra5 {0.07/30 0} 16. Nd2 {490} Bg4 {0.09/30 0} 17. f3 {250} Be6 {0.13/29
413} 18. Nc4 {160 (Kh1)} Ra6 {0.16/29 568} 19. Bxf6 {698} Bxf6 {0.12/30 171}
20. Kh1 {193} Qd7 {0.12/29 111} 21. g3 {19 (Ne3)} Rc8 {0.01/25 426} 22. Ne3 {
311} Bd8 {0.14/27 952} 23. Ncd5 {245} Kh8 {0.12/26 41} 24. Rac1 {246} Bc7 {0.
17/26 575} 25. Nb4 {215} Ra5 {0.14/26 12} 26. Rd2 {137 (Nbd5)} Bb8 {0.14/26 227
} 27. Rb1 {213 (Rcd1)} g6 {0.15/25 510} 28. Ned5 {49 (Nbd5)} Qd8 {0.10/25 441}
29. Rbd1 {666} Kg7 {0.06/26 0} 30. Qg2 {391 (Ne3)} Kh8 {0.11/24 406} 31. f4 {
126 (Qe2)} Bg4 {0.28/24 297} 32. Rf1 {71} exf4 {0.34/27 138} 33. Qf2 {33} Rc4 {
0.41/26 174} 34. Qxf4 {36} Bh3 {0.44/27 289} 35. Re1 {567 (Rfd1)} Be6 {0.30/25
179} 36. Ne3 {129 (c3)} Rc8 {0.46/25 268} 37. Nbd5 {0} Rac5 {0.52/26 363} 38.
c3 {222 (Nf6)} Qf8 {0.52/25 161} 39. Red1 {0} h6 {0.62/23 198} 40. Ng4 {122
(Kg2)} f5 {0.88/22 150} 41. exf5 {192 (Nge3)} Rc4 {1.47/26 270} 42. fxe6 {52}
Rxf4 {1.44/26 0} 43. Nxf4 {153} Kh7 {1.23/26 30} 44. Kg2 {450} h5 {1.32/29 267}
45. Nf2 {24 (Ne3)} Qe7 {1.67/23 351} 46. Re1 {95 (Ne4)} Kg7 {4.63/24 149} 47.
Nd5 {37} Qg5 {4.98/26 203} 48. Ne4 {497} Qe5 {5.00/29 0} 49. Ree2 {223} Qxe6 {
5.56/28 1067} 50. Nef6 {290} Qxf6 {5.60/27 101} 51. Nxf6 {247} Kxf6 {5.04/28 0}
52. Re4 {374 (Rd4)} Kf5 {5.12/26 808} 53. Re7 {491 (Rxa4)} Ba7 {5.08/27 174}
54. Rxd6 {141} Bc5 {5.21/28 19} 55. Rf7+ {27} Ke4 {5.45/27 184} 56. Rxg6 {996
(Rf4+)} b6 {5.42/25 126} 57. Rd7 {63} Rf8 {5.87/26 140} 58. Re6+ {88} Kf5 {5.
89/27 0} 59. Re2 {46 (Re1)} Rg8 {6.51/26 936} 60. Rd5+ {57} Kg6 {6.93/28 299}
61. Ree5 {423 (h4)} Rb8 {7.23/26 257} 62. h4 {85 Resigns} 1-0
[/pgn]
This is one of the few games where it seems Houdini just did not understand the position. There were at least three bad moves (?) that I noticed on my analysis engine's eval.
Cordially,
Sean
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
That is someone else who said that, not me. I don't agree that it's the only fair way to test and I think he was merely exaggerating.beram wrote:Well Don I am sorry to come back on this, but I just find it funny that you probably forgot what the probably ''not just someone'' Larry wrote earlier on 19 july: http://www.talkchess.com/forum/viewtopi ... 67&t=48566Don wrote:Here is my take:kranium wrote:Sean-
to recap...
this tournament has:
no book move limit
(lines sometimes running past move 30 or more)
no position reversal of color
(a cardinal sin in testing)
and of course your very powerful "Large Neutral Opening Books"
(which have been shown to end out of book with one side getting a huge (almost full pawn) advantage?)
what's really laughable is this tournament's credibility
Some of these issues are not ideal but they are not definitely no "cardinal sin" either. There is nothing improper about this test and each tester has their own way.Code: Select all
1. long book - short is preferred but but my own testing shows it's not that as big a deal as I used to think. It does slightly favor the weaker player. 2. long time control - this favors Komodo (relative to short time control) 3. no book limit - bigger deal that point 1 if it takes you out of the most interesting part of the game. Favors the weaker program. 4. no position reversal - no issue here except for statistical noise as this requires more games to be equally fair. In a short match this favors the weaker player and asymptotically makes no difference as the match gets longer. 5. Note that choice of openings is "unfair" in the sense that some programs do better with some openings - this is mostly solved by more games as is point 1 and 4 assuming more games means more different kinds of openings. 6. Openings that end with huge advantage (or drawish positions.) Like points 4 and 5 this wastes time and requires more games to get past the extra statistical noise. It also minimizes the chances of a victory for the stronger program given a match of the same size. And the difference in match score will be reduced.
The primary thing about these conditions is not unfairness but just the fact that it would require a lot more games to resolve who is the better program - that doesn't mean it's "unfair" it just means there is more noise in the test. For example Houdini could have gotten all the winning positions - over enough games it will average out. The only thing that could be construed as a "true" fairness issue is time control, but ALL tests have that issue - depending on the test and hardware and time control some programs will be favored over others and there is nothing you can do about that. Why do you think we like the long time control tests better?
If you take an opening that ends with a big advantage, you will often find that one program will win it and another will only manage to draw so I don't have any issue with ending book with at least a few positions where one side has a fairly large advantage as long as the advantage is not so much as to make the win almost a sure thing for both programs.
My only real complaint here is the reversal of color issue - unless you are going to play at least many hundreds of games that does introduce a lot of statistical noise, especially when there are a number of unbalanced positions and you are playing a short match. Someone said earlier that this make the results "random" which is a very gross exaggeration and is simply not the case. Every match is "random" in a sense that no matter the difference in ELO the stronger play might lose any given game but that is not uniformly random and neither is having some unbalanced positions.
Summary - most of the condition of this match favor Komodo for statistical reasons only - but that advantage would asymptotically be reduced to nothing if Sean plays enough games. In other words Houdini will still win this match if Sean plays enough games assuming that Houdini is the superior program at this hardware and time control which I am not disputing.
So everyone just shut up and enjoy the match!!
.."I would like to know why this match or any other match is run without reversing colors after each game? No matter if the book is 2 moves or 30, it is the only fair way to test and greatly reduces the luck factor. I can't think of any argument against doing this."
But even so it is compatible with my statement where I said that every game has some chance (luck) involved and this does in fact increase that just as Larry said. It increase that factor for a single game and if you understand statistics the more games you play the less of a factor that becomes.
Still, I identified that as the only real concern I had with the conditions of this match if it is going to be a short match.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
-
- Posts: 1777
- Joined: Thu Jun 05, 2008 10:58 pm
- Location: Canada
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
Code: Select all
(120m+10s)/40+(120m+10s)/40+(120m+10s) 0
1 Komodo5.1 +4 +17/=50/-16 50.60% 42.0/83
2 Houdini3 -4 +16/=50/-17 49.40% 41.0/83
[pgn]
[Event "(120m+10s)/40+(120m+10s)/40+(120m+10s)"]
[Site "?"]
[Date "2013.08.03"]
[Round "83"]
[White "Houdini3"]
[Black "Komodo5.1"]
[Result "1/2-1/2"]
[ECO "D45"]
[Annotator "25. -0.03"]
[PlyCount "134"]
[TimeControl "40/7200+10:40/7200+10:7200+10"]
{2412MB, Combinedbooks.ctg, Laptop1} 1. d4 {B 0} Nf6 {0} 2. Nf3 {B 0} e6 {0} 3.
c4 {B 0} d5 {0} 4. Nc3 {B 0} c6 {0} 5. e3 {B 0} Nbd7 {0} 6. Qc2 {B 0} Bd6 {0}
7. Be2 {B 0} O-O {0} 8. b3 {B 0} b6 {0} 9. O-O {B 0} Bb7 {0} 10. Bb2 {B 0} Qe7
{0} 11. Rad1 {B 0} Rfe8 {0} 12. e4 {B 0} Nxe4 {0} 13. Nxe4 {B 0} dxe4 {0} 14.
Qxe4 {B 0} Rad8 {0} 15. Bd3 {B 0} f5 {0} 16. Qe3 {B 0} c5 {0} 17. d5 {B 0} exd5
{0} 18. cxd5 {B 0} Qxe3 {0} 19. fxe3 {B 0} Rxe3 {0} 20. Bxf5 {B 0} Nf8 {0} 21.
Bc1 {B 0} Re2 {0} 22. Rf2 {B 0} Rxf2 {0} 23. Kxf2 {B 0} h6 {149} 24. Bb2 {B 0
last book move} b5 {167} 25. a4 {-0.03/24 741} bxa4 {477 (a6)} 26. bxa4 {0.07/
26 634} c4 {113 (Ba8)} 27. Bd4 {0.05/27 443} Bxd5 {194} 28. Bxa7 {0.07/27 259}
Bc7 {190} 29. g3 {0.05/27 319} Kf7 {576} 30. Bc2 {0.07/28 48} Ng6 {672} 31. Be3
{0.07/28 444} Ne7 {808} 32. Nd4 {0.06/29 0} Ba5 {438} 33. Rb1 {0.07/28 1163}
Rd6 {300} 34. Ke2 {0.03/27 395} Rf6 {193 (Rd8)} 35. Rf1 {0.00/30 661} Rxf1 {514
} 36. Kxf1 {0.01/31 0} Nc6 {675} 37. Nb5 {0.02/31 862} Ne5 {87} 38. Ke2 {0.03/
30 924} Bf3+ {1009} 39. Kf2 {0.03/31 0} Bd5 {478 (Bc6)} 40. h3 {0.04/29 544}
Nc6 {0 (Bb4)} 41. Ke2 {0.02/27 160} c3 {354 (Ke6)} 42. Kd1 {0.06/25 304} Kf6 {
496 (Nb4)} 43. Kc1 {0.12/28 225} Bg2 {741 (Ke5)} 44. h4 {0.14/29 183} Ke5 {307}
45. Bc5 {0.13/29 0} Bf1 {345} 46. Bd6+ {0.10/28 34} Ke6 {201 (Kd5)} 47. Bb3+ {
0.08/28 222} Kf5 {82 (Kf6)} 48. Bf8 {0.11/29 198} Kf6 {87} 49. Bc5 {0.10/28 91}
Kf5 {149 (Ke5)} 50. Bd1 {0.10/27 188} Ke5 {250} 51. Bf8 {0.09/28 115} Kf6 {240}
52. Bc2 {0.08/29 0} Ne5 {256 (Be2)} 53. Bc5 {0.05/28 203} Nd3+ {33} 54. Bxd3 {
0.05/28 0} Bxd3 {66} 55. Bd4+ {0.02/30 199} Ke6 {148 (Kf7)} 56. Bxg7 {-0.01/30
838} h5 {107 (Bxb5)} 57. Bxc3 {0.16/31 169} Bxb5 {89} 58. Bxa5 {0.17/31 0} Bc6
{338 (Bxa4)} 59. Bc7 {0.17/34 204} Bxa4 {138} 60. Kd2 {0.15/36 88} Kd7 {150
(Kf5)} 61. Bf4 {0.15/35 669} Bb5 {391 (Ke6)} 62. Kc3 {0.16/35 187} Bf1 {380
(Ke6)} 63. Kd4 {0.15/33 161} Be2 {261 (Ke6)} 64. Kd5 {0.18/34 150} Bg4 {209
(Bf3+)} 65. Bd6 {0.19/35 157} Ke8 {243 (Bf3+)} 66. Bc5 {0.20/34 168} Bd1 {143
(Kd7)} 67. Kd6 {0.19/35 198} Bg4 {191 (Kf7)} 1/2-1/2
[/pgn]
-
- Posts: 212
- Joined: Tue Jun 25, 2013 8:19 pm
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
Well,
This match, and all the controversy about proper testing surrounding it, will certainly dampen whatever the final result will be.
This match, and all the controversy about proper testing surrounding it, will certainly dampen whatever the final result will be.
-
- Posts: 2129
- Joined: Thu May 29, 2008 10:43 am
Re: Komodo5.1 v Houdini3 Match - Long Time Controls
+1PaulieD wrote:Well,
This match, and all the controversy about proper testing surrounding it, will certainly dampen whatever the final result will be.
the bigger controversy IMO, is the amount of 'spin' being disbursed by the author,
who seems perfectly willing to directly contradict himself as well as his esteemed partner
in an determined attempt to deliver some 'validity' to this seriously flawed match...
of course the motive is clear, Komodo is a commercial endeavor
and anyone swallowing it all 'hook line and sinker' may mean another copy sold
both authors have been selling the notion (for some time now) that Komodo is approaching Houdini in strength,
the caveat being: but only at long time controls
(which, of course, is a extremely difficult thing to prove or disprove unless you have enormous amount of time, patience, and CPUs, almost impossible for the average user)
well, both authors have admitted here that this match favors Komodo, and i believe heavily...
for ex: there's at least 1 game where Komodo emerges from the opening with an almost full 1 pawn advantage!
i.e.: concerning objective engine testing and measurement of strength: it's a complete waste of time
ok now, back to the Dailey show!
catchphrase of the week: 'statistical noise' !