<SICILIAN:

Games of 40 moves or less:
0.35 / 0.36 / 0.29

Games of more than 40 moves:
0.36 / 0.31 / 0.33

CARO-KANN:

Games of 40 moves or less:
0.34 / 0.42 / 0.23

Games of more than 40 moves:
0.36 / 0.33 / 0.31>

QUEEN'S INDIAN DEFENCE:

Games of 40 moves or less:
0.28 / 0.55 / 0.18

Games of more than 40 moves:
0.36 / 0.38 / 0.26

Interestingly, the huge proportion of draws in shorter games also now eats into white's scoring rate, but the advantage over black is significant. Black fails to equalize in longer games.

NIMZO INDIAN DEFENCE:

Games of 40 moves or less:
0.29 / 0.46 / 0.25

Games of more than 40 moves:
0.35 / 0.36 / 0.29

Similar statistics to the QID, though with greater success for black early in the game. Black does not quite equalize in longer games, and again draws are very common in shorter games.

DUTCH DEFENCE:

Games of 40 moves or less:
0.42 / 0.32 / 0.26

Games of more than 40 moves:
0.42 / 0.31 / 0.27

Not good for black at all!

MODERN BENONI (incl. Benko):

Games of 40 moves or less:
0.42 / 0.28 / 0.30

Games of more than 40 moves:
0.41 / 0.29 / 0.30

Percentage of games that are 40 moves or less:

1.e4
Sicilian: 56%
Caro-Kann: 55%

Games of more than 40 moves
From 1995: 0.47 / 0.37 / 0.16
From 1980: 0.46 / 0.36 / 0.18

Black suffers badly in a longer game.

ORTHODOX
Games of 40 moves or less:
From 1995: 0.35 / 0.51 / 0.15
From 1980: 0.29 / 0.59 / 0.12

Games of more than 40 moves
From 1995: 0.44 / 0.33 / 0.23
From 1980: 0.41 / 0.39 / 0.20

Black's chances early in the game are very low, and he doesn't come close to equalizing in a longer game. Interestingly the drawing percentage has come down in recent times.

SLAV
Games of 40 moves or less:
From 1995: 0.33 / 0.49 / 0.18
From 1980: 0.31 / 0.52 / 0.17

Games of more than 40 moves
From 1995: 0.39 / 0.36 / 0.25
From 1980: 0.39 / 0.36 / 0.25

SEMI-SLAV
Games of 40 moves or less:
From 1995: 0.35 / 0.44 / 0.21
From 1980: 0.34 / 0.45 / 0.21

Games of more than 40 moves
From 1995: 0.37 / 0.36 / 0.27
From 1980: 0.37 / 0.37 / 0.27

Under 40 moves:

32%/50%/18%

Over 40 moves:

39%/35%/26%

The "profile" of the Ruy is similar to that of a 1. d4 opening: relatively drawish in shorter games, but with a large advantage for White at any length. The <Spanish Torture> earns its name.

Under 40 moves:

35%/46%/20%

Over 40 moves:

39%/42%/19%

Compared to the regular Ruy, Black does better in short games, and worse in long ones. Makes sense, given the Open Ruy's characteristics: better piece play, looser pawns. In any event, my personal score against Dr. Palmer's O/R is a solid 0.

Under 40 moves:

31%/55%/14%

Over 40 moves:

43%/36%/21%

To my surprise, terrible for Black! Slightly more drawish than the Ruy in shorter games, but the draws come at Black's expense, not White's; and thoroughly bad for Black in long games.

Games 40 moves or less:

40%/34%/26% (same as before)

Games over 40 moves:

37%/30%/33%

59% of all Alekhine's' don't make it to move 41.

Revised French numbers:

40 moves or under:

36%/39%/24% (same)

Over 40 moves:

39%/31%/30%

Black's disadvantage in long games is relatively large for a KP opening, small for a QP opening.

57% of games are 40 moves or less.

Revised Pirc/Robatsch numbers:

40 moves or under:

40%/31%/29% (same)

Over 40 moves:

38%/30%/33%

56% of all Pirc/Robatsch defenses are 40 moves or less.

King's Gambit (C30-39)

40 or fewer:

45%/19%/36%

Not bad!

40 or more:

39%/31%/31%

White retains an advantage even in long games. Long ago, Bronstein pointed out (and demonstrated) that the King's Gambit often led to good endgames for White.

70% of King's Gambits last 40 moves or less. Clearly an opening for manly men (and women, like distinguished practitioner Judit Polgar). There are 1,325 King's Gambits in the database from 1995 or later.

Giuco Piano

40 or fewer:

38%/33%/29%

More than 40:

41%/26%/33%

Not as good for White as the Ruy, but better than I would have expected.

57% of Giuco Pianos last 40 moves or less. There are 1,887 examples in the database from 1995 on.

Two Knights

40 moves or fewer:

39%/28%/33%

More than 40 moves:

39%/30%/32%

One of the better results against 1. e4, and noticeably better for Black than the Giuco Piano. This is interesting since the Two Knights has a bad theoretical reputation right now. But openings select players, as well as the other way around; those who prefer 3...Nf6 are probably slightly more ambitious, more confident, better booked, and perhaps stronger on average than those who choose 3...Bc5.

57% of Two Knights last 40 moves or fewer. There are 1,426 Two Knights in the database since 1995. The Two Knights and Giuco together are more than four times as popular as Alekhine's Defense.

Evans Gambit (which is a subset of the Giuco, of course)

40 moves or less:

36%/28%/36%

More than 40 moves:

29%/29%/42%

Not good for the Cinderella of openings! 57% of Evans Gambits last 40 moves or fewer. White's best hope seems to be that Black will skewer him in the middlegame, rather than slowly roasting him in the ending. In light of these results, it's not surprising that there are only 168 Evans since 1995 in the database.

Vienna Game

40 or fewer:

33%/35%/32%

More than 40:

36%/31%/33%

If White wants a quiet life with no pesky opening advantage, 2. Nc3 seems like an ideal choice. Surprisingly popular: There are 918 examples since 1995. Strange that White does slightly better in the Vienna in long games than in short ones.

Evans wrote something in <<Chess Life>> in response to this article, and it evolved into a debate between Evans and Santasiere in <<Chess Life>> in the early 1960's. I know that I don't have all of the articles or letters to the editor, but perhaps I can find a few.

I decided to take a look at the four players' performances with black as given in the database (subject to the usual caveats about the database).

Kramnik

+229/28.0%
-125/15.3%
=463/56.7%

Fischer

+211/47.5%
- 79/17.8%
=154/34.7%

Karpov

+436/29.7%
-219/14.9%
=815/55.4%

Kasparov

+385/37.7%
-128/12.5%
=509/49.8%

Kramnik's performance with Black is very similar to Karpov's (albeit slightly worse). He wins slightly more than a quarter of the time, loses less than 1/6th of the time, and draws the rest. Not too surprising.

Kasparov, impressively, wins more often than Karpov and Kramnik, while losing less often. Still, about half his games with Black end in draws. It follows that the +5=2 tournament performance alluded to earlier was an outlier.

Fischer loses more often than the others, but wins much, much more often. In fact, he is the only player of the four who wins more often than he draws. Of course, partisans of the three K's would observe that it's easier to rack up the wins against Arthur Bisguier than, say, Alexander Beliavsky or Vishy Anand.

Of players from the 1950s-70s, Fischer and Larsen stood out for their extremely low drawing percentages. But compared to players from around 1900, their percentages were just average.

I decided to look at Kramnik, Fischer, Karpov and Kasparov's performance as White.

Kramnik

+ 395/45.0%
- 60/6.8%
= 423/48.2%

Fischer

+ 345/58.5%
- 96/16.3%
= 149/25.3%

Karpov

+ 880/51.0%
- 132/7.7%
= 712/41.3%

Kasparov

+ 700/57.5%
- 63/5.2%
= 455/37.4%

These statistics have to be caveated very heavily, because all of these players (according to the database) played more games with White than with Black. With Fischer it's 590/444 (57.1%), Kramnik 878/817 (51.8%), Karpov 1,724/1,470 (54.0%), and Kasparov 1,218/1,022 (54.4%). So clearly there are a lot of exhibition and simultaneous games in these figures, which would presumably skew a player's results upwards. The skewing effect is most pronounced with Fischer and least pronounced with Kramnik.

That said, Kramnik's results are most similar to Karpov's, as they were with Black, but he draws noticeably more often than Karpov and wins and loses noticeably less often. With Black, Fischer was the only one who won more often than he drew; with White, Kramnik is the only one who draws more than he wins.

As before, Fischer loses more often and wins more often than any of the others. Kasparov's results, though, are stunning. His winning percentage is slightly behind Fischer's - if simul games were excluded, it would probably be a statistical tie, although I am sure Kasparov has faced much stronger opposition on average. But Kasparov loses far, far less often than Fischer -- less than 1/3 as often. And, Kasparov loses less often than Karpov, and even Kramnik. Put another way, Kasparov has three more losses with White than Kramnik, but 305 more wins!

Three things stand out in this survey for me:

1. Compared to Karpov, who is stylistically closest to him of the three, Kramnik has near identical results with Black, but is noticeably more drawish with White.

2. Fischer's results with Black are astounding.

3. Kasparov's results with White are mind-blowing.

Kramnik vs Navara, 2007

Kramnik has gone in for this opening several times in his career:

click for larger view

It reminded me of Petrosian's famously unambitious setup in game 5 of the match with Botvinnik:

click for larger view

Notably, Kramnik was twice as likely (16%) to draw his games in under 20 moves as Kasparov (7%) or even Karpov (8%). Not sure how much of that is related to illness in Kramnik's case.

So, what do all these numbers say about <shams>' original claim?

<Fischer, Karpov, Kasparov played nearly every game to win. Kramnik makes not even a show at this attitude. He plays not to lose.>

I agree with <code 13>'s evaluation: <This is wild generalisation.> :-) But shams' statement isn't completely off-base either. Fischer does seem to have played just about every game to win. Kasparov seems to have played to win all the time with White and quite often with Black -- more often than Kramnik and Karpov, at any rate. Neither Kramnik nor Karpov seems to have played to win with Black as a general matter. Of course they both played to win with White, but Karpov was rather more successful at it.

Kramnik plays a higher percentage of short draws than any of the other three. Fischer avoided short draws almost entirely, and he was probably the last WC of whom that can be said.

There is a lot more analysis to be done. In particular, it would be useful to measure Kramnik against his contemporaries, especially his great rivals Topalov and Anand. One thing that was clear from <tpstar>'s chart is that the 2000s have been very drawish so far, so measuring Kramnik's draw rates against men who peaked in the 1970s, 80s and even 90s is probably unfair to Kramnik.

That is a good point. I recalculated Karpov's performance as White, limiting it to games played in 1993 or earlier (I think Karpov in 1993 was roughly Kasparov's age at retirement).

+576/52.5%
- 72/6.7%
=449/40.9%

It makes less of a difference than I expected, to be honest. But he now loses very slightly less often than Kramnik, while continuing to win significantly more often. On the other hand, he still wins less often than Kasparov, and loses more.

I took a look at games of 100 moves or more, for no good reason. It's hard to separate out the rapid games from recent years, but, as near as I can tell, of the four Karpov played the most long games (+7-3=7). Fischer and Kasparov were successful (+2-0=1 Fischer, +3-0=3 Kasparov), Kramnik less so (+1-1=1). But the sample sizes are too small for that to mean anything.

Other players: Tal +0-1=5, Spassky +0-1=0, Petrosian +2-0=6, Botvinnik +1-0=5, Smyslov +3-1=5 (one of his wins came against Cramling in 1996!), Lasker +1-1=5, Korchnoi +5-2=7 (including a 115-move win against Ponomariev in 2001, and a 135-move draw against Sokolov in 2003!).