Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Game 2 - did IBM cheat?

Author: Vasik Rajlich

Date: 00:36:50 06/07/05

Go up one level in this thread


On June 06, 2005 at 20:36:55, Vincent Diepeveen wrote:

>On June 06, 2005 at 17:25:51, Jack Lad wrote:
>
>>[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
>>
>>http://chess.eusa.ed.ac.uk/Chess/DeepBlue97/game2.html
>>
>>Which was the human like move that DB should not have played according to GK?
>
>37.Be4
>

The controversial moment was one move earlier:

[D] r1r1q1k1/6p1/p2b1p1p/1p1PpP2/PPp5/2P4P/R1B2QP1/R5K1 w - -

What engines will resist 36. Qb6 in favor of 36. axb5?

Shredder 9 says:

 11/23	 0:01 	+1.50 	36.Qb6 Rd8 37.axb5 Rab8 38.Qxa6 e4 39.Qa7 Qe5 40.Qe3 Re8
41.Re1 Qxd5 42.Bxe4 (531.020) 401
 12/32	 0:03 	+1.75++	36.Qb6 Rd8 (1.228.510) 395
 12/32	 0:03 	+1.75 	36.Qb6 Rd8 37.Be4 (1.353.950) 397
 13/32	 0:05 	+1.74 	36.Qb6 Rd8 37.Be4 Rab8 38.Qxa6 Qh5 39.axb5 Qh4 40.Re2 Qg3
41.Bf3 Qf4 (2.112.230) 399
 14/31	 0:09 	+1.78 	36.Qb6 Rd8 37.Be4 Rac8 38.Qxa6 bxa4 39.Qxa4 Qh5 40.Qa7 Qg5
41.Re1 Bc7 (3.993.922) 400
 15/38	 0:20 	+1.83 	36.Qb6 Qf8 37.axb5 axb5 38.Qxb5 Rxa2 39.Rxa2 Qe7 40.Kf1 h5
41.Qb6 Rd8 (8.402.625) 412
 16/39	 0:34 	+1.76 	36.Qb6 Rd8 37.Be4 Rac8 38.Qxa6 bxa4 39.Qxa4 Qh5 40.Qa7 Qg5
41.Re1 Re8 42.Qe3 Qxe3+ 43.Rxe3 Ra8 44.Rxa8 (14.434.341) 412
 17/41	 1:05 	+1.88 	36.Qb6 Qf8 37.axb5 axb5 38.Qxb5 Rxa2 39.Rxa2 Qe7 40.Be4 Qc7
41.Kf1 Kf8 (26.555.750) 403
 18/45	 2:02 	+1.89 	36.Qb6 Rd8 37.Be4 Rac8 38.Qxa6 bxa4 39.Qxa4 Qh5 40.Qa7 Qg5
41.Qf2 Rd7 42.Ra8 Rxa8 43.Rxa8+ Bf8 44.Kh2 Qc1 45.Qg3 (49.459.465) 404
 19/49	 3:04 	+1.91 	36.Qb6 Rd8 37.Be4 Rac8 38.Qxa6 bxa4 39.Qxa4 Qh5 40.Qa7 Qg5
41.Qf2 Rd7 42.g4 Re8 43.b5 h5 (74.205.652) 401

In multi-move:

 17	 1:42 	+1.91 	36.Qb6 Rd8 37.Be4 Rac8 (38.961.144) 380
 17	 2:12 	+0.37 	36.axb5 axb5 (51.240.717) 385


It was certainly an interesting decision for an engine to make.

Vas

>however this is big BS from kasparov. first of all crafty and diep in those days
>used to play Be4 too at 10-12 ply or so.
>
>The point is that deep blue just got 11 ply and we search deeper, so our engines
>play Qb6 in the end, deep blue with its shallow 1997 searches was happy to hit
>11 ply and didn't get to 13 ply or so where Diep by then plays Qb6.
>
>Please note that 37.Qb6 is a forced win for white, as also Seirawan indicated.
>I've been writing out here in CCC a few years ago the winning line for about 20
>moves or so until real huge scores for Qb6.
>
>At small 11 ply search depths however in those days engines of course just saw
>opposite bishops and didn't want to play it.
>
>The whole deep blue match versus kasparov is of course 1 big publicity match. We
>would have seen a 3d match kasparov-deep blue where kasparov would have drawn
>deep blue, and then a 4th match with kasparov winning and a 5th match with
>kasparov losing again and a 6th match with kasparov winning again.
>
>Just a show each year.
>
>Kasparov was very naive to believe he would get another match from a big company
>like IBM when losing to the computer.
>
>My only fear is that the same happens with Hydra.
>
>It wins 2 or 3 matches against GM's and then they shredder the machine and claim
>superiority and cash in their huge bonus from sheik and leave computerchess.
>
>The impact of the marketing department of IBM is what did the damage, additional
>to the believe this machine was unbeatable. This where it is nothing more than a
>10-12 ply searching gnuchess.
>
>Which for 1997 was of course a huge search depth, but by 2005 standards it's
>what we get within a few seconds.
>
>Especially quotes deduced from it, like: "chess has been solved", have done the
>real damage to sponsors.
>
>Kasparov has been praising the machine bigtime until 1 hour after the match.
>
>The first press conference kasparov gave right after the match, he praised the
>programmers and the machine. I saw that press conference live at CNN.
>
>1 hour later he had a second press conference where he was shouting, screaming,
>looking pathetic and accusing people of all kind of idiocy.
>
>However if we look in the logfiles as they have been published, we CAN notice a
>few weird things.
>
>First of all we know that the machine has crashed several times in the first few
>games. Where are those crashes in the logfiles?
>
>If my thing crashes, then i have to resetup the position. So you'll find the
>same position analyzed 2 times, or stuff given in 2 times. At least you will
>find some extra stuff in the logfile which i didn't find in the deep blue II
>logfiles. But perhaps i didn't look too well. Someone with more time to do that
>can do perhaps, Uri?
>
>Please note in that last game, it's obvious from the logfile that deep blue
>plays Nxe6 from book. Losing 9 minutes in total for just 8 moves or so. So the
>deep blue II machine is real slow reacting onto its openings book. Perhaps some
>verification search or just timing it out is real slow. Probably that last.
>
>Yet in that time that it takes to search we can see that it has no clue about
>the position at all. It's score for itself after Nxe6 when guessing fxe6 is 0.00
>
>If i let diep search *before* Nxe6 i already get white up way over 1 pawn.
>00:00      7 7 0 (2) 1 (0,0) -0.863 Ng5xe6 f7xe6
>00:00     12 12 0 (2) 1 (0,0) 0.512 Ng5-e4 Nf6xe4 Bd3xe4
>00:00    136 136 0 (2) 2 (0,1) 0.005 Ng5-e4 Bf8-e7
>00:00  34150 683 0 (2) 3 (0,3) 0.512 Ng5-e4 Nf6xe4 Bd3xe4
>00:00  54275 2171 0 (2) 4 (0,6) -0.083 Ng5-e4 Nf6xe4 Bd3xe4 Bf8-d6
>00:00  96840 9684 0 (2) 5 (1,11) 0.218 Ng5-e4 Nf6xe4 Bd3xe4 Nd7-f6 Be4-d3
>++ g5-e6 procnr=0 terug=235 org=[218;219] newwindow=[218;520000]
>00:00 114706 18353 0 (2) 5 (2,15) 0.484 Ng5xe6 f7xe6 Bd3-g6 Ke8-e7 Bc1-f4 Nd7-b6
>
>00:00 131920 31661 0 (2) 6 (2,35) 0.415 Ng5xe6 Qd8-e7 O-O f7xe6 Bd3-g6 Ke8-d8
>00:00 144763 63696 0 (2) 7 (3,72) 0.821 Ng5xe6 f7xe6 Bd3-g6 Ke8-e7 Bc1-f4 Qd8-b6
> O-O Nf6-d5 Qd1-c1 Nd5xf4 Qc1xf4 Qb6xb2
>00:00 147896 116838 0 (2) 8 (4,106) 0.821 Ng5xe6 f7xe6 Bd3-g6 Ke8-e7 Bc1-f4 Qd8-
>b6 O-O Nf6-d5 Qd1-c1 Nd5xf4 Qc1xf4 Qb6xb2
>00:02 157423 349481 0 (2) 9 (16,218) 0.901 Ng5xe6 f7xe6 Bd3-g6 Ke8-e7 O-O Nd7-b6
> Bc1-f4 Nb6-d5 Bf4-g3 Bc8-d7
>00:12 153002 1872746 0 (2) 10 (42,412) 1.068 Ng5xe6 Qd8-e7 O-O f7xe6 Bd3-g6 Ke8-
>d8 Nf3-e5 Nd7xe5 d4xe5 Nf6-d7 Qd1-d3 Qe7-b4 Bc1-d2 Qb4xb2 f2-f4
>00:18 158128 2987041 0 (2) 11 (54,621) 1.068 Ng5xe6 Qd8-e7 O-O f7xe6 Bd3-g6 Ke8-
>d8 Nf3-e5 Nd7xe5 d4xe5 Nf6-d7 Qd1-d3 Qe7-b4 Bc1-d2 Qb4xb2 f2-f4
>00:51 155712 8031675 0 (2) 12 (93,984) 1.068 Ng5xe6 Qd8-e7 O-O f7xe6 Bd3-g6 Ke8-
>d8 Nf3-e5 Nd7xe5 d4xe5 Nf6-d7 Qd1-d3 Qe7-b4 Bc1-d2 Qb4xb2 f2-f4
>02:18 156565 21748568 0 (2) 13 (139,1668) 1.226 Ng5xe6 Qd8-e7 O-O f7xe6 Bd3-g6 K
>e8-d8 Bc1-f4 Qe7-b4 Qd1-e2 Bf8-e7 c2-c3 Qb4-a5 b2-b4 Qa5-a3 Nf3-e5 Nd7xe5 Qe2xe5
>
>This on a dual K7 2.1Ghz. Please note i already had in the year 2001 a dual k7
>(1.2Ghz), so by now this hardware is just so so outdated.
>
>There is really no cheating in deep blue II. Let's just wonder how much money
>kasparov has earned on those matches in total. It really is way more than the
>average human being guesses. He has been paid well to lose.
>
>Only after he lost he realized the marketing problem he had created by losing to
>a marketing giant like IBM. That might explain his reaction a few hours *after*
>the match, which was an opposite reaction to his reaction directly after.
>
>Vincent



This page took 0 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.