Author: Lar Mader
Date: 19:39:47 03/03/06
Ok, I know this topic has been discussed a lot (beaten to death?!), but I am
still curious about this move:
Game 2, IBM Kasparov vs. Deep Blue Rematch
Deep Blue played
36.axb5
I know that the move 37.Be4 is a move that is found by some current chess
engines.
But this 36.axb5, to my knowledge, is never selected by current chess engines.
I can't duplicate this move on Fritz 9 or Deep Shredder 9. Below I provide the
analysis of Deep Shredder 9 running for 2 days, and Fritz 9 running for 1 day.
Has anyone been able to duplicate this move with a current chess program? I'm
especially curious about Rybka, since it seems to find some interesting moves.
Here is the position, full game is below.
[D]r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - 0 36
Here is the link for anyone who wants to read the Deep Blue logs.
http://www.research.ibm.com/deepblue/watch/html/c.html
These logs, by the way, came out roughly 3 years after Deep Blue was dismantled.
Hmmmm... am I the only one that thinks that that is perhaps a little odd. So,
let's see... the project is over, Deep Blue was dismantled immediately, and then
they just let these logs gather dust for 3 years, while Kasparov ranted and
raved about them cheating? Then they cough 'em up out of the "blue". Why and
how would this happen? Conspiracy theorists want to know :)
According to these logs (if I understand them), Deep Blue spent almost 7 minutes
on this move. They claimed that Deep Blue calculated a peak of 200 million
nodes/sec, but that realistically it was more like 130 million nodes per second
on average, or some such.
In 7 minutes (using the 130 mN/s figure), this would be roughly 55 billion nodes
considered, perhaps.
In the analysis below:
After 21 hours, Fritz 9 had considered around 74 billion nodes.
After 47 hours, Deep Shredder 9 had considered 65 billion nodes.
So in principle, both of these engines did at least as much work as Deep Blue,
on this move. Yes, the trees searched certainly looked much different, but the
amount of work was comparable, i.e. the speed of deep blue has been taken out of
the equation. Now it's just a question of search and eval quality.
I have a hard time believing that the position evaluation of Deep Blue was
better than modern eval functions, and modern search algorithms have improved
tremendously.
Thus, my point is: I would think that Fritz and Deep Shredder would consider a
more optimal tree of nodes than Deep Blue, given algorithmic advances, and with
a better eval function. i.e., for the same amount of work, I would expect a
better result from Fritz or Shredder than Deep Blue could produce.
But they don't choose axb5. This really seems odd. It's especially odd given
that Deep Blue's choice of axb5 dovetails nicely with move 37.Be4, which isn't
typically found easily at this earlier point in the search (from move 36).
By the way, it should be noted that Shredder considered axb5 the 2nd best move,
and Fritz considered it the 4th best move.
Here is the whole game, and the engine analysis follows below:
[Event "IBM Kasparov vs. Deep Blue Rematch"]
[Site "New York, NY USA"]
[Date "1997.05.04"]
[Round "2"]
[White "Deep Blue"]
[Black "Kasparov, Garry"]
[Opening "Ruy Lopez: closed, Smyslov defense"]
[ECO "C93"]
[Result "1-0"]
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 19.a4 Nh4 20.Nxh4 Qxh4
21.Qe2 Qd8 22.b4 Qc7 23.Rec1 c4 24.Ra3 Rec8 25.Rca1 Qd8 26.f4 Nf6
27.fxe5 dxe5 28.Qf1 Ne8 29.Qf2 Nd6 30.Bb6 Qe8 31.R3a2 Be7 32.Bc5 Bf8
33.Nf5 Bxf5 34.exf5 f6 35.Bxd6 Bxd6 36.axb5 axb5 37.Be4 Rxa2
38.Qxa2 Qd7 39.Qa7 Rc7 40.Qb6 Rb7 41.Ra8+ Kf7 42.Qa6 Qc7 43.Qc6 Qb6+
44.Kf1 Rb8 45.Ra6 1-0
Deep Blue - Kasparov,G, IBM Kasparov vs. Deep Blue Rematch 1997
r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - 0 1
Analysis by Fritz 9:
36.Qb6 Qd7 37.axb5 Rcb8 38.Qe3 Qxb5 39.Be4 Re8
+- (1.46) Depth: 7/18 00:00:00 53kN
36.Qb6 Qd7 37.axb5 Rcb8 38.Qe3 Bc7 39.Kh2 Bb6 40.Qg3 Qxb5
± (1.36) Depth: 8/20 00:00:00 105kN
36.Qb6!
+- (1.64) Depth: 9/22 00:00:00 220kN
36.Qb6 Bc7 37.Qe6+ Qxe6 38.fxe6 Rcb8 39.Be4 Bd6 40.axb5 Rxb5 41.Kf2 Kf8
+- (1.80) Depth: 9/24 00:00:00 369kN
36.Qb6 Bc7 37.Qe6+ Qxe6 38.fxe6 Rcb8 39.Be4 Bd6 40.axb5 Rxb5 41.Kf2 Kf8
+- (1.80) Depth: 10/20 00:00:00 410kN
36.Qb6 Bc7 37.Qe6+ Qxe6 38.fxe6 Rab8 39.axb5 axb5 40.Ra6 Rd8 41.Be4 Bb6+ 42.Kf1
f5 43.e7
+- (1.84) Depth: 11/26 00:00:01 745kN
36.Qb6 Bc7 37.Qe6+ Qxe6 38.fxe6 Rab8 39.axb5 axb5 40.Kf1 Bd6 41.Ke2 Kf8 42.Be4
Ke7 43.Ra7+
+- (1.68) Depth: 12/33 00:00:03 2289kN
36.Qb6 Bc7 37.Qe6+ Qxe6 38.fxe6 Rab8 39.Kf1 Kf8 40.axb5 axb5 41.Ra7 Rb6 42.Be4
Rd6 43.Ke2
+- (1.73) Depth: 13/31 00:00:06 5781kN
36.Qb6 Bc7 37.Qe6+ Qxe6 38.fxe6 Rab8 39.Kf1 Kf8 40.axb5 axb5 41.Be4 Bd6 42.Ke2
Ke7 43.g3 Rc7 44.Ke3
+- (1.71) Depth: 14/33 00:00:16 14624kN
36.Qb6!
+- (2.00) Depth: 15/40 00:00:37 35526kN
36.Qb6 Rd8 37.Be4 Kh7 38.Qc6 Qxc6 39.dxc6 Rab8 40.axb5 axb5 41.Ra7 Rdc8 42.Kf2
Bc7 43.Kf3 Kg8
+- (1.81) Depth: 16/39 00:01:15 70739kN
36.Qb6--
+- (1.53) Depth: 17/41 00:02:48 159264kN
36.Qb6 Rd8 37.Be4
+- (1.53) Depth: 17/41 00:03:01 172179kN
36.Qb6--
± (1.25) Depth: 18/55 00:11:33 662396kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Bc7 40.Ra6 Bb6+ 41.Rxb6
± (1.24) Depth: 18/58 00:13:25 772277kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8 41.b6 Qh2+ 42.Kf1 Bf4
43.Qg1 Qg3 44.Qf2 Qh2
± (1.26) Depth: 19/46 00:24:46 1428950kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qf2 axb5 39.Be4 Qd8 40.Kh2 Ra8
± (1.03) Depth: 20/52 00:41:03 2373942kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qf2 axb5 39.Be4 Qd8 40.Kh2 Ra8 41.Ra7 Rxa7 42.Qxa7
Rc7 43.Qe3
± (1.04) Depth: 21/52 01:57:43 6836329kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qf2 axb5 39.Be4 Qd8 40.Kh2 Ra8 41.Ra7 Rxa7 42.Qxa7
Rc7 43.Qe3
± (1.04) Depth: 22/53 03:45:33 13112880kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qf2 axb5 39.Be4 Qd8 40.Kh2 Ra8 41.Ra7 Rxa7 42.Qxa7
Rb8 43.Ra6 Bf8
± (1.04) Depth: 23/83 07:40:14 26775814kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qf2 axb5 39.Be4 Qd8 40.Kh2 Ra8
± (1.04) Depth: 24/80 21:12:21 74891917kN
(Mader, 01.03.2006)
Deep Blue - Kasparov,G, IBM Kasparov vs. Deep Blue Rematch 1997
r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - - 0 1
Analysis by Deep Shredder 9:
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00 18kN
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00 18kN
36.Qb6 Qe7
± (1.14) Depth: 9/14 00:00:00 18kN
36.Qb6 Qe7
± (1.14) Depth: 10/14 00:00:00 52kN
36.Qb6 Qe7
± (1.14) Depth: 11/17 00:00:00 108kN
36.Qb6 Qe7
± (1.14) Depth: 12/12 00:00:00 225kN
36.Qb6 Qe7
± (1.14) Depth: 13/13 00:00:01 453kN
36.Qb6 Qe7
± (1.14) Depth: 14/14 00:00:01 847kN
36.Qb6 Qe7
± (1.14) Depth: 15/16 00:00:03 1552kN
36.Qb6 Qe7
± (1.14) Depth: 16/17 00:00:07 3106kN
36.Qb6 Qe7
± (1.14) Depth: 17/27 00:00:10 4952kN
36.Qb6 Qe7
± (1.14) Depth: 18/18 00:00:19 9111kN
36.Qb6 Qe7
± (1.14) Depth: 19/24 00:00:32 15072kN
36.Qb6 Qe7
± (1.14) Depth: 20/20 00:01:03 29392kN
36.Qb6 Qe7
± (1.14) Depth: 21/28 00:01:49 51096kN
36.Qb6 Qe7
± (1.39) Depth: 22/52 00:10:15 269665kN
36.Qb6 Qe7
± (1.39) Depth: 22/52 00:16:24 418189kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Bxe4 Re8 40.Kf1 Qe5 41.Bf3 Rbd8 42.Ra3
Qxf5 43.Qa7 Qf4
± (1.35) Depth: 23/59 00:29:57 753619kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8 41.b6 Qxf5 42.Ba4 Re7
43.Bc6 Qe5 44.Ra7 Qh2+ 45.Kf1 f5 46.Rxe7
± (1.17) Depth: 24/57 01:07:17 1661960kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8 41.b6 Qxf5 42.Rd1 Qe5
43.Ba4 Qh2+ 44.Kf1 Re7 45.Bb5 f5 46.Bxc4
± (0.95) Depth: 25/59 02:20:12 3428674kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8 41.b6 Qxf5 42.Rd1 Qe5
43.Ba4 Qh2+ 44.Kf1 Re7 45.Bb5 f5 46.Bxc4
± (1.14) Depth: 26/61 04:16:25 6241570kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8 41.b6 Qxf5 42.Ba4 Re7
43.Bc6 Qe5 44.Ra8 f5 45.Rxb8+ Bxb8 46.Ra8 Rf7
± (1.00) Depth: 27/61 07:07:21 10324389kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8 41.b6 Qh2+ 42.Kf1 Bf4
43.Qg1 Qg3 44.Qf2 Qh2 45.Ke2 e3 46.Qf3 Be5 47.Ra3 Rxb6
± (1.04) Depth: 28/64 12:29:26 17947658kN
36.Qb6 Qe7
± (0.79) Depth: 29/67 21:34:29 30468028kN
36.Qb6 Qe7 37.axb5 Rab8 38.Qe3 axb5 39.Ra6 Qd8
² (0.52) Depth: 29/70 47:05:07 65855136kN
(Mader, 03.03.2006)
This page took 0.02 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.