Coincidence Statistics of the moves of GM Kramn...

Originally posted by Wulebgr
80-90%

Is that excluding the opening?

Originally posted by Los Angeles
Coincidence Statistics of the moves of GM Kramnik
with recommendations of the chess program Fritz 9
Elista, October 4, 2006

After very detailed analyze of all games of the match we would like to present to your attention coincidence statistics of the moves of GM Kramnik with recommendations of chess program Fritz 9.

First game:
From 75 moves: A ...[text shortened]... z9.

Sincerely
Silvio Danailov




http://forumforme.com/showthread.php?p=36#post36[/b]

No duh he'sone of the best chess players in the world!

Originally posted by Wulebgr
80-90%

Utter rubbish. For one or two games out of a hundred, maybe. But consistently over many games, once out of theory, GM's moves match the top engine moves 50% to 60%.

They match the top 2 engine moves 70% to 80%.

They match the top 3 engine moves 75% to 85%.

Tactical games may produce long strings of matches. Positional games may produce long strings of differences. But these things average out to approach the levels I give above.

Whenever one uses statistics to "prove" an hypothesis in order to determine the significance of a match-up. you have to have extensive control data. And your methodology has to be intelligent enough to differentiate between tactical and positional chess.

In isolation, the statistics quoted by team Topalov appear very high. On the other hand, they have given no criteria for the match-up, ie set engine times, or set engine levels. They have give no comparative data. What were Topalov's match-ups in the same games? A whole lot of subjectivity can creep into an interpretation, especially if you have a vested interest.

During several minutes of "thinking" Fritz could come up with a half a dozen "top" moves as its search depth increases. Did Team Topalov pick and choose the "matches" to suit their case? We don't know.

But I would treat their evidence as suspect.

Originally posted by Superman
XanthosNZ, My idea was that this thinking lines were not published, and that is why Kasparov was claiming the human help, or am I wrong?

Well your idea was wrong. Understandably as even Game Over the apparently definitive telling of the Kasparov - Deep Blue story said it was the case.

It's not.

http://www.chessninja.com/dailydirt/2005/04/log_jam_deep_blue_and_kasparov.htm

Just out of curiosity, did anyone ever checked out what's the statistics of Kramnik-fritz match-up in any of his previous tournament games? And then, did Kramnik also had the habit of frequent toilet visitation for other tournaments? Is there any difference between frequent/infrequent toilet visit to computer match-up? If the percentage of match-up is more or less the same, then the toilet visits shouldn't be an issue. Maybe it's just that Kramnik's mind is almost similar with the computer as far as chess is concerned?

Originally posted by XanthosNZ
Well your idea was wrong. Understandably as even Game Over the apparently definitive telling of the Kasparov - Deep Blue story said it was the case.

It's not.

http://www.chessninja.com/dailydirt/2005/04/log_jam_deep_blue_and_kasparov.htm

Hummmmmm.......... ok thanks.

Originally posted by XanthosNZ
Well your idea was wrong. Understandably as even Game Over the apparently definitive telling of the Kasparov - Deep Blue story said it was the case.

It's not.

http://www.chessninja.com/dailydirt/2005/04/log_jam_deep_blue_and_kasparov.htm

So Krammink "may have" used a computer to beat a Top GM,
and Kasparov claims that a "human" beat him with a Computer?
😠😳🙁🙄

kramnik is the best of the best, AND he is the best of the best in the AGE OF COMPUTERS.

his match ups might be a little different to normal gms or previous champion gms.

i would be surprised if they were not - he is most able to learn from those better than him - primarily computers, he will learn to see their tricks coming and respond to avoid, as they would also.

perhaps topa has learned to emulate the computer tricks and kram has learnedto defend against them

Originally posted by Bedlam
I swear you're stupid. I cant even be bothered to spend the time typing to crush this pointless prattle so I will just give you a link to Chess Ninja's which highlights the many flaws in Danailov's naff.

http://www.chessninja.com/dailydirt/2006/10/danailov_accuses_kramnik_of_cheating.htm

They sell Chess ninja thongs 😵

Here's my own analysis on the second game (which is claimed to have an 87% matchup).

[Event "WCh"]
[Site "Elista RUS"]
[Date "2006.09.24"]
[Round "2"]
[White "Topalov,V"]
[Black "Kramnik,V"]
[Result "0-1"]
[WhiteElo "2813"]
[BlackElo "2743"]
[EventDate "2006.09.23"]
[ECO "D19"]

1. d4 d5 2. c4 c6 3. Nc3 Nf6 4. Nf3 dxc4 5. a4 Bf5 6. e3 e6 7. Bxc4 Bb4 8. O-O Nbd7 9. Qe2 Bg6 10. e4 O-O 11. Bd3 Bh5 12. e5 Nd5 13. Nxd5 cxd5 14. Qe3 Bg6 15. Ng5 Re8 16. f4 Bxd3N This is the novelty according to my database. Fritz would play Rc8 here (+0.22) and gives the played move a score of 0.41 (third choice) at a depth of 15 ply.
17. Qxd3 f5 This doesn't even need much analysis. There are only three moves that doesn't lose a piece and of those this is the best by 0.25 of a pawn over g6 (other non-losing move is Nf8).
18. Be3 Nf8 Fritz says Qa5 (0.16), h6/Qb6/Rb8 (0.25). The text move gets +0.47 and is 10th equal in Fritz's choices.
19. Kh1 Rc8 Fritz agrees with this as the first choice at a depth of 16 ply and offers Re7 as equal to it and a6 as 0.03 pawns worse.
20. g4 Qd7 h6 is reckoned to be -0.50 by Fritz, almost half a pawn ahead of fxg4 and Qd7 (-0.06) at a depth of 16 ply.
21. Rg1 Be7 Early on Be7 doesn't appear at all and fxg4 is greatly favoured. By 15 ply Be7 is the second choice (-0.06) and fxg4 is still the top choice (-0.25).
22. Nf3 Rc4 Again fxg4 is favoured early but later loses out to Rc6 (-0.16). The played move scores -0.06 at a depth of 16 ply.
23. Rg2 fxg4 fxg4 is the strongest move by far here early on, 0.6 pawns ahead of every other move. This changes later and at 16 ply fxg4 scores -0.25 over Ng6 at -0.09.
24. Rxg4 Rxa4 7th Fritz choice. 0.19 behind the first choice.
25. Rag1 g6 First choice. Everthing else is 0.91 or more worse.
26. h4 Rb4 Bd8 and Rb4 score equally at 0.00. Ra5 scores 0.06 and a6 0.56.
27. h5 Qb5 First choice (0.00). Second choice is Rxb2 (0.19) at 16 ply.
28. Qc2 Rxb2 Equal with Rb3 at 0.00. Bd8 scores 0.44 and everything else is even further back.
29. hxg6 h5 Fritz feels comfortable of even play after Rxc2 although only a very confident human would even consider it. Text is 0.66 worse and second choice.
30. g7 hxg4 Only move.
31. gxf8=Q+ Bxf8 Only move.
32. Qg6+ Bg7 Only move.
33. f5 Re7 Scores worse than exf5 (0.12 vs. 0.00). Second choice of three candiate moves.
34. f6 Qe2 Only move.
35. Qxg4 Rf7 Only move.
36. Rc1 Rc2 Only move.
37. Rxc2 Qd1+ Second choice 0.5 behind Qxc2.
38. Kg2 Qxc2+ Only move.
39. Kg3 Qe4 Much worse than Qf5 (0.56) and way down the list of moves.
40. Bf4 Qf5 4th choice behind a5, b5 (scoring 0.00) and b6 (0.03). Text scores 0.06.
41. Qxf5 exf5 A true only move. Everything else scores over 13 in white's favour.
42. Bg5 a5 Only move that has a chance.
43. Kf4 a4 Only move that has a chance.
44. Kxf5 a3 Scores the same as Rc7. Both of those moves are 0.75 ahead of everything else.
45. Bc1 Bf8 Fritz would play a2 here which looks horrible but it seems it can find a way to an advantage like only an engine can (-1.09). The played move is 4th choice at 0.00
46. e6 Rc7 It's either this or Rh7 here and Rh7 is bad.
47. Bxa3 Bxa3 Obviously the only logical move.
48. Ke5 Rc1 Rc3 is much better (0.47 better). Text is third choice.
49. Ng5 Rf1 Fritz likes Rg1 over Rf1 by 0.28.
50. e7 Re1+ Obviously the only logical move.
51. Kxd5 Bxe7 Obviously the only logical move.
52. fxe7 Rxe7 Obviously the only logical move.
53. Kd6 Re1 Re1 and Re3 both score the same (-1.13) and Re2 scores -0.69, everything else is 0.00.
54. d5 Kf8 Only possible winning move. Scores -1.13 at a depth of 20 ply.
55. Ne6+ Ke8 Only possible winning move. Scores -1.13 compared to 0.00 for all other options.
56. Nc7+ Kd8 Only possible winning move. Scores -1.34 compared to 0.00 for all other options.
57. Ne6+ Kc8 Only possible winning move. Scores -1.34 compared to 0.00 for all other options.
58. Ke7 Rh1 Only possible winning move. Scores -1.34 compared to 0.00 for all other options.
59. Ng5 b5 Scores -4.53, second choice is Rd1 at -3.72
60. d6 Rd1 Only winning move. I did the following moves in reverse order (63. Re3 first) which gave greater depth faster on these moves (hashtables). It also may have reduced moves appearing as Fritz got to the depth it was analysing at.
61. Ne6 b4 Only winning move
62. Nc5 Re1+ Only winning move (by like 16 pawns).
63. Kf6 Re3 At a depth of 19 ply this scores -16.75 compared to Kd8 (-16.47) and Re2 (-15.94). All other moves give away the advantage.
0-1

So even with the only moves I get 72.9% and that's including many moves that not even a Class A player would miss.

Originally posted by Gatecrasher
Utter rubbish. For one or two games out of a hundred, maybe. But consistently over many games, once out of theory, GM's moves match the top engine moves 50% to 60%.

I disagree.

As I stated before, I've run engine analysis on hundreds of games, and do not find 65% or even 75% agreement surprising. Of course the engines match the GMs 100% in the openings, or the engines would play terribly.

In order to produce some data, while I slept I ran Fritz blundercheck on all the games from Linares 2005 (Kasparov's last event, and a good one for Topalov). It will take some time to compile the data (time I do not have today), but I've started.

Michael Adams played 224 post-book moves in the event. Fritz (set at 1/5 of a pawn threshhold and 10 seconds per move) found 45 moves it would prefer over these 224. Hence 80% agreement. Of Adams' 12 games in the event, only 2 fall into the 50-60% range.

Veselin Topalov had fewer variances from Fritz and more post-book moves than Adams: 40/273. His agreement was 85% overall with a low of 73%; he was at 94% in his last round win over Kasparov. His four 100% games included two short draws, which probably should be discounted.

Add such norming data (when compiled) to XanthosNZ's fine analysis of the game in question, and we see more reason to recognize every claim made by Danailov as bovine feces.

What a winning!

Topalov rashed Kramnik like he was playing with the T...c or the m....r.
Really nice game.



Topalov is picking up speed. More interesting games are to come.
I expect more great winnings from Topalov.

Originally posted by XanthosNZ
Didn't we go over this before powershaker? There is no evidence that Deep Blue was helped by humans. You can look at the logs that it outputted during the match yourself if you wish. They are available online. But hey, why let some facts stand in the way of a good conspiracy theory?

Well, it sure is humorous that now that Kram the Sham has been called out for cheating, now Topalov crushes him in a clear win! HAHAHAHA! See? Fritz9 no more baby!

Originally posted by Los Angeles
Topalov is picking up speed. More interesting games are to come.
I expect more great winnings from Topalov.

Yes! I indeed think Topalov will win this match now that Kramnik can't use Fritz9!??? LOL! The Russians can hand that title to Bulgaria now - if they wish! They might as well! Go Topy!

Originally posted by powershaker
Yes! I indeed think Topalov will win this match now that Kramnik can't use Fritz9!??? LOL! The Russians can hand that title to Bulgaria now - if they wish! They might as well! Go Topy!

Call me stupid but if Kramnik was using fritz9 before why couldnt he use it now?


Topalov is the laughing stock of the chess world.