engines & positional play

[Terry McCracken]

Computers are good but still they don't know chess or at least they don't play top quality positional chess like humans do. Computers are short-range "thinkers" while humans are excellent at long-range thinking. Humans are excellent pattern recognizers while computers are not. Humans are much better at positional (or strategical) chess while computers are much better at tactical chess.

Wasn't it said that after Rybka, this is no longer true?

Positionally Rybka is very good, but long range plans are beyond computers.

Terry

[smirobth]

A little over ten years ago, another great chess player got a surprise:
[d]r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - bm axb5; am Qb6;

The computer played axb5. Back in those days, none of the PC programs could find it. I bet that lots of them can find it now.

Hi Dann,
This famous position has been discussed a lot, and I think the consensus these days is that Qb6 is stronger than axb5. axb5 doesn't win against best defense (Kasparov resigned in a drawn position!), while Qb6 very likely might win by force.

I wonder if someone with one of those monster boxes and SMP Rybka cold give it a pounding for a full weekend. I would be keenly interested to see the outcome.
I am giving a bunch of strong programs a one hour each look tonight, but there are machines that can probably duplicate the quality of that analysis in 15 minutes or less.

Hi Dann,
I ran the position on my quad-core Opteron for a couple days. Rybka does switch from Qb6 to axb5 after 17 hours/29 plies, but the score change is not compelling:

r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - 0 1

Analysis by Rybka 2.3.2a mp :

1.Qb6 Bc7 2.Qe6+ Qxe6
+/- (1.12) Depth: 5 00:00:00
1.Qb6 Bc7 2.Qe6+ Qxe6
+/- (1.12) Depth: 7 00:00:00 9kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8
+/- (1.29) Depth: 7 00:00:00 11kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8 4.axb5 axb5 5.Be4
+/- (1.23) Depth: 8 00:00:00 17kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8 4.axb5 axb5 5.Kf2 Kf8
+/- (1.19) Depth: 9 00:00:00 26kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8 4.axb5 axb5 5.Be4 Kf8 6.Kh2
+/- (1.15) Depth: 10 00:00:00 50kN
1.Qb6 Rd8 2.Be4 Be7 3.axb5 Rd6 4.Qa5 Bd8
+/- (1.33) Depth: 11 00:00:00 140kN
1.Qb6 Rd8 2.Be4 Be7 3.axb5 Rd6 4.Qa5 Bd8 5.Qa4
+/- (1.33) Depth: 12 00:00:00 161kN
1.Qb6 Rd8 2.Be4 Rac8 3.Qxa6 bxa4 4.Qxa4 Qh5 5.Qd1 Qg5 6.Qd2
+/- (1.38) Depth: 13 00:00:01 250kN
1.Qb6 Rd8 2.Be4 Rac8 3.Qxa6 bxa4 4.Qxa4 Qh5 5.Qa7 Qg5 6.Qf2 Rc7
+- (1.46) Depth: 14 00:00:02 453kN
1.Qb6 Rd8 2.Be4 Rac8 3.Qxa6 bxa4 4.Qxa4 Qh5 5.Qa7 Qg5 6.Qf2 Rc7 7.Ra6 Kf7
+/- (1.28) Depth: 15 00:00:06 1406kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qd8 6.Bxe4 Bb6+ 7.Rxb6 Rxb6
+/- (1.27) Depth: 16 00:00:10 2250kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qd8 6.Bxe4 Bb6+ 7.Rxb6 Rxb6
+/- (1.19) Depth: 17 00:00:13 2920kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qxf5 7.Ba4 Red8
+/- (1.12) Depth: 18 00:00:21 4787kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qxf5 7.Ba4 Red8
+/- (1.12) Depth: 19 00:00:33 7759kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qxf5 7.Ba4 Red8
+/- (1.06) Depth: 20 00:00:50 12410kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qe5 6.b6 Qh2+ 7.Kf1 Bg3
+/- (0.98) Depth: 21 00:01:40 24820kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qe5 6.b6 Bd6 7.b7 Re8
+/- (0.94) Depth: 22 00:03:05 49299kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qe5 6.b6 Bd6 7.b7 Re8
+/- (0.97) Depth: 23 00:05:12 83726kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.84) Depth: 24 00:10:54 183mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.77) Depth: 25 00:22:58 385mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.78) Depth: 26 00:42:55 716mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.84) Depth: 27 01:20:19 1307mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.76) Depth: 28 02:30:19 2427mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/= (0.69) Depth: 29 09:54:50 9623mN
1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Bb6 7.Kg4 Qd7
+/= (0.70) Depth: 29 17:15:40 15019mN
1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Kf8 7.Qa6 Rb6
+/= (0.69) Depth: 30 21:37:58 18965mN
1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Qd7 7.Qa6 Rb6
+/= (0.69) Depth: 31 33:06:48 29390mN

[Dann Corbit]

A little over ten years ago, another great chess player got a surprise:
[d]r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - bm axb5; am Qb6;

The computer played axb5. Back in those days, none of the PC programs could find it. I bet that lots of them can find it now.

Hi Dann,
This famous position has been discussed a lot, and I think the consensus these days is that Qb6 is stronger than axb5. axb5 doesn't win against best defense (Kasparov resigned in a drawn position!), while Qb6 very likely might win by force.

I wonder if someone with one of those monster boxes and SMP Rybka cold give it a pounding for a full weekend. I would be keenly interested to see the outcome.
I am giving a bunch of strong programs a one hour each look tonight, but there are machines that can probably duplicate the quality of that analysis in 15 minutes or less.

Hi Dann,
I ran the position on my quad-core Opteron for a couple days. Rybka does switch from Qb6 to axb5 after 17 hours/29 plies, but the score change is not compelling:

r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - 0 1

Analysis by Rybka 2.3.2a mp :

1.Qb6 Bc7 2.Qe6+ Qxe6
+/- (1.12) Depth: 5 00:00:00
1.Qb6 Bc7 2.Qe6+ Qxe6
+/- (1.12) Depth: 7 00:00:00 9kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8
+/- (1.29) Depth: 7 00:00:00 11kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8 4.axb5 axb5 5.Be4
+/- (1.23) Depth: 8 00:00:00 17kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8 4.axb5 axb5 5.Kf2 Kf8
+/- (1.19) Depth: 9 00:00:00 26kN
1.Qb6 Bc7 2.Qe6+ Qxe6 3.fxe6 Rab8 4.axb5 axb5 5.Be4 Kf8 6.Kh2
+/- (1.15) Depth: 10 00:00:00 50kN
1.Qb6 Rd8 2.Be4 Be7 3.axb5 Rd6 4.Qa5 Bd8
+/- (1.33) Depth: 11 00:00:00 140kN
1.Qb6 Rd8 2.Be4 Be7 3.axb5 Rd6 4.Qa5 Bd8 5.Qa4
+/- (1.33) Depth: 12 00:00:00 161kN
1.Qb6 Rd8 2.Be4 Rac8 3.Qxa6 bxa4 4.Qxa4 Qh5 5.Qd1 Qg5 6.Qd2
+/- (1.38) Depth: 13 00:00:01 250kN
1.Qb6 Rd8 2.Be4 Rac8 3.Qxa6 bxa4 4.Qxa4 Qh5 5.Qa7 Qg5 6.Qf2 Rc7
+- (1.46) Depth: 14 00:00:02 453kN
1.Qb6 Rd8 2.Be4 Rac8 3.Qxa6 bxa4 4.Qxa4 Qh5 5.Qa7 Qg5 6.Qf2 Rc7 7.Ra6 Kf7
+/- (1.28) Depth: 15 00:00:06 1406kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qd8 6.Bxe4 Bb6+ 7.Rxb6 Rxb6
+/- (1.27) Depth: 16 00:00:10 2250kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qd8 6.Bxe4 Bb6+ 7.Rxb6 Rxb6
+/- (1.19) Depth: 17 00:00:13 2920kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qxf5 7.Ba4 Red8
+/- (1.12) Depth: 18 00:00:21 4787kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qxf5 7.Ba4 Red8
+/- (1.12) Depth: 19 00:00:33 7759kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qxf5 7.Ba4 Red8
+/- (1.06) Depth: 20 00:00:50 12410kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qe5 6.b6 Qh2+ 7.Kf1 Bg3
+/- (0.98) Depth: 21 00:01:40 24820kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qe5 6.b6 Bd6 7.b7 Re8
+/- (0.94) Depth: 22 00:03:05 49299kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Bc7 5.Ra6 Qe5 6.b6 Bd6 7.b7 Re8
+/- (0.97) Depth: 23 00:05:12 83726kN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.84) Depth: 24 00:10:54 183mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.77) Depth: 25 00:22:58 385mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.78) Depth: 26 00:42:55 716mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.84) Depth: 27 01:20:19 1307mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/- (0.76) Depth: 28 02:30:19 2427mN
1.Qb6 Qe7 2.axb5 Rab8 3.Qxa6 e4 4.Qa7 Qe5 5.Qe3 Re8 6.b6 Qh2+ 7.Kf1 Bf4
+/= (0.69) Depth: 29 09:54:50 9623mN
1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Bb6 7.Kg4 Qd7
+/= (0.70) Depth: 29 17:15:40 15019mN
1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Kf8 7.Qa6 Rb6
+/= (0.69) Depth: 30 21:37:58 18965mN
1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Qd7 7.Qa6 Rb6
+/= (0.69) Depth: 31 33:06:48 29390mN

Thanks for this analysis. It confirmed what I guessed. If Kasparov says something and 100 experts say something else, I guess Kasparov is pretty close to right anyway.

I think that the reason Kasparov was stunned is that he probably prepared this sequence very carefully. I guess he played against chess computers and they all chose Qb6 and they all lost to him. Of course, Deep Blue was not at all like other chess computers of the time.

I honestly believe that the dismantling of Deep Blue was one of the greatest tragedies of science. Sure, it could have been programmed to take greater risks and it might have won more easily. But I guess that the 200M NPS would still produce stunning results from time to time today against Hydra or Rybka but we'll never know.

Really, positions like this are why I love chess. The thing that looks obviously better is not always best and sometimes you have to work really hard to understand why. But when you really understand why, it's a wonderful feeling.

I envy your chess ability and I suspect that real chess greatness is to some degree hardwired into the brain at birth. And so I thank you for your analysis and insight. If I were to have to hire a consultant like you to look at chess positions for me, I am quite sure I could not afford it.

[smirobth]

Thanks for this analysis. It confirmed what I guessed. If Kasparov says something and 100 experts say something else, I guess Kasparov is pretty close to right anyway.

I think that the reason Kasparov was stunned is that he probably prepared this sequence very carefully. I guess he played against chess computers and they all chose Qb6 and they all lost to him. Of course, Deep Blue was not at all like other chess computers of the time.

I honestly believe that the dismantling of Deep Blue was one of the greatest tragedies of science. Sure, it could have been programmed to take greater risks and it might have won more easily. But I guess that the 200M NPS would still produce stunning results from time to time today against Hydra or Rybka but we'll never know.

Really, positions like this are why I love chess. The thing that looks obviously better is not always best and sometimes you have to work really hard to understand why. But when you really understand why, it's a wonderful feeling.

I envy your chess ability and I suspect that real chess greatness is to some degree hardwired into the brain at birth. And so I thank you for your analysis and insight. If I were to have to hire a consultant like you to look at chess positions for me, I am quite sure I could not afford it.

Thanks. However I didn't really use any analytical skills or insight on this position... just plugged it into my computer. Speaking of which it has now passed 53 hours/depth 32:

1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Qd7 7.Qa6 Rb6
+/= (0.69) Depth: 32 53:00:16 46767mN

Still no change in the eval.

By the way I very seriously doubt that Kasparov had seen the move 36.Qxb6 during his pregame preparations. The game left known theory way back at move 18, and if Kasparov had foreseen how strong Deep Blue's subsequent play was going to be after his move 18 choice I doubt he would have selected this line. I think much more likely is that he just found the move sequence axb5 followed by Be4, eliminating all possible Black counter-play, to be much more human like than he was used to seeing from computers up to that point. Kasparov hates passive positions where he does not have counter-play.

[Dann Corbit]

Thanks for this analysis. It confirmed what I guessed. If Kasparov says something and 100 experts say something else, I guess Kasparov is pretty close to right anyway.

I think that the reason Kasparov was stunned is that he probably prepared this sequence very carefully. I guess he played against chess computers and they all chose Qb6 and they all lost to him. Of course, Deep Blue was not at all like other chess computers of the time.

I honestly believe that the dismantling of Deep Blue was one of the greatest tragedies of science. Sure, it could have been programmed to take greater risks and it might have won more easily. But I guess that the 200M NPS would still produce stunning results from time to time today against Hydra or Rybka but we'll never know.

Really, positions like this are why I love chess. The thing that looks obviously better is not always best and sometimes you have to work really hard to understand why. But when you really understand why, it's a wonderful feeling.

I envy your chess ability and I suspect that real chess greatness is to some degree hardwired into the brain at birth. And so I thank you for your analysis and insight. If I were to have to hire a consultant like you to look at chess positions for me, I am quite sure I could not afford it.

Thanks. However I didn't really use any analytical skills or insight on this position... just plugged it into my computer.

I was speaking in general terms. I recognize you as a supreme chess player who not only has a great deal of insight, but are also willing to share it. You also communicate in terms that I can understand.

Speaking of which it has now passed 53 hours/depth 32:

1.axb5 axb5 2.Be4 Qd8 3.Kh2 Rxa2 4.Qxa2 Rb8 5.h4 Bc7 6.Kh3 Qd7 7.Qa6 Rb6
+/= (0.69) Depth: 32 53:00:16 46767mN

Just for fun, here is the game from Chess Assistant (so that it has the annotations in it):

[Event "Match"]
[Site "New York (USA)"]
[Date "1997.??.??"]
[Round "2"]
[White "Computer "Deep Blue""]
[Black "Kasparov Garry (RUS)"]
[Result "1-0"]
[ECO "C93"]
[WhiteElo "0"]
[BlackElo "2820"]
[Annotator "Kodinets Konstantin L (RUS)"]
[Source ""]
[Remark "V"]

1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.O-O Be7 6.Re1 b5 7.Bb3
d6 8.c3 O-O 9.h3 h6 10.d4 Re8 11.Nbd2 Bf8 12.Nf1 Bd7 13.Ng3 Na5
14.Bc2 c5 15.b3 Nc6 16.d5 Ne7 17.Be3 Ng6 18.Qd2 Nh7
( 18...a5 {!?} ) 19.a4 {!} Nh4 {a novelty 19... Be7 - Lobron - Short, Amsterdam 1983}
( 19...Be7 20.Nf5 Bxf5 21.exf5 Nh4 22.Nxh4 Bxh4 23.f6 {!?} Bxf6
24.axb5 axb5 25.Rxa8 Qxa8 26.Bxh6 gxh6 27.Qxh6 e4 28.Bxe4 Ng5
29.Qxf6 Nxe4 30.Rxe4 Rxe4 31.Qg5+ Kf8 32.Qh6+ Ke7 33.Qg5+ Kf8
{1/2 : 1/2} ) 20.Nxh4 Qxh4 21.Qe2 Qd8 22.b4 Qc7 23.Rec1 {!} {%09}
c4 24.Ra3 {!} Rec8 25.Rca1 Qd8 26.f4 {!?} ( 26.axb5 axb5 27.Ba7
{!} {'with the idea' Bd1, Qa2$16 Seirawan} ) Nf6 ( 26...exf4
{} 27.Bxf4 Nf6 28.Be3 {} {'with the idea' Bd4$16} ) 27.fxe5 dxe5
28.Qf1 {?} ( 28.Qf2 $16 ) Ne8 29.Qf2 Nd6 30.Bb6 Qe8 31.R3a2 Be7
{?!} ( 31...f6 ) 32.Bc5 Bf8 {?!} ( 32...Qd8 {%03'better is'}
{} ) 33.Nf5 {!} Bxf5 34.exf5 f6 {34... e4!? Seirawan}
( 34...Qd8 35.f6 {!} gxf6 36.Bb6 $16 ) 35.Bxd6 {!} Bxd6 36.axb5
( 36.Qb6 Rd8 37.axb5 Rab8 38.Qxa6 e4 $131 {Seirawan} ) axb5 37.Be4
{!} Rxa2 38.Qxa2 Qd7 39.Qa7 Rc7 40.Qb6 Rb7 41.Ra8+ Kf7
( 41...Kh7 {%03'better is'} ) 42.Qa6 Qc7 43.Qc6 {!} Qb6+ 44.Kf1
{?} ( 44.Kh1 {!} Rb8 45.Ra6 Qe3 46.Qxd6 Re8 47.Ra1 Qxe4 48.Ra7+
Kg8 49.Qd7 Qe1+ 50.Kh2 {} ) Rb8 45.Ra6 {?} {%09} ( 45.Qd7+ {!}
Kg8 46.Ra7 Bf8 47.Qe6+ {!} Kh8 48.Qxb6 Rxb6 49.Ke2 $16 ) Z0 {[ K. KODINETS ]}
( 45...Qe3 {!} 46.Qxd6 ( 46.Qd7+ Kg8 47.Qxd6 Rf8 {=} ) Re8 {!}
47.Bf3 ( 47.h4 {!?} Qxe4 {} 48.Ra7+ Kg8 49.Qd7 Qf4+ 50.Kg1 Qe3+
51.Kh2 Qf4+ 52.Kh3 Re7 {!!} 53.Ra8+ ( 53.Qc8+ Kh7 54.Rxe7 h5
{!!} 55.Qc5 Qg4+ 56.Kh2 Qxh4+ 57.Kg1 Qe1+ {=} ) Kh7 54.Qc8 Re8
{!!} 55.Qxe8 Qxf5+ 56.Kh2 Qf4+ {=} ) Qc1+ 48.Kf2 Qd2+ 49.Be2
Qf4+ 50.Ke1 Qc1+ 51.Bd1 Qxc3+ 52.Kf1 Qc1 {!} {=} ) 1-0

Still no change in the eval.

By the way I very seriously doubt that Kasparov had seen the move 36.Qxb6 during his pregame preparations.

Since there is only one game in my entire database with that move in it, I suspect that you are right about that. I wish that Kasparov would expound on it more.

The game left known theory way back at move 18, and if Kasparov had foreseen how strong Deep Blue's subsequent play was going to be after his move 18 choice I doubt he would have selected this line. I think much more likely is that he just found the move sequence axb5 followed by Be4, eliminating all possible Black counter-play, to be much more human like than he was used to seeing from computers up to that point. Kasparov hates passive positions where he does not have counter-play.