Author: Robert Hyatt
Date: 18:53:38 09/18/01
Go up one level in this thread
On September 18, 2001 at 21:08:03, K. Burcham wrote: >deep blue vs kasparov >game 1 > >here is the conversation with kasparov and his team trying to figure out why >deep blue played the "blunder" Rd1. here they find out it was not a blunder. >deep blue first had to solve Rf5. then it had to solve Rg8. >after deep blue solved both of these moves as a mate, then it chose Rd1. >todays programs will indicate this as a blunder, because they cannot see that > Rg8 or Rf5 or Kh5 or e3 fail lower than 44...Rd1. >yes the game was over for deep blue in this position. >the point here is deep blue did choose 44...Rd1. >kasparov was convinced that Rd1 was a blunder until his study with his > computer that night in the hotel. > >also based on this info in the deep blue log, > --> 44. ... Rd1 <-- 16/72:55, >does this mean depth is 16/72? and deep blue only took 55 seconds to move > in this position? > No... that is "hand typed" I believe. The DB search depths are given as 11(6) and the like... the 72:55 seems to suggest that this took 72 minutes to search, which seems excessive. If you look at the previous line, you will see the 11(6) which is the software(hardware) depth of that search. Apparently Rf5 failed low on an alpha value of -260, it timed out after almost 1300 seconds. It continued to go on due to the big score drop, and found Rg8 but that was still -183. It continued to Rd1 failed high on -180 and it eventually took that move as the one to play. I can't quite reconcile the 72:55 however... aha... wait a minute... that is total time left on the clock until the next time control... Should have noticed this quicker... > >after 44...Rd1, kasparov plays 45.g7 then deep blue resigned. > > >That night in the hotel we crowded around the Fritz screen and went through the >game. Garry entered and checked a number of variations. In the position before >the final move by Deep Blue he had a problem. > >The game continued 44...Rd1 45.g7 and Black resigned. Sitting in front of the >computer Garry started trying out other lines, beginning with 44...Rf5+. Then he >turned to me (his "computer expert") and said: "How could Deep Blue play >44...Rd1 and lose immediately? 44...Rf5+ also loses, but it puts up much more >resistance. How can a computer commit suicide like that?" I had no immediate >answer, it did indeed look very strange. Garry continued to potter around with >Fritz, and suddenly he found the solution himself. After 44...Rf5+ White has >45.Ke3! with beautiful forced wins: 47...Rd8 48.g7 Rxc2+ 49.Ke1 Rc1+ 50.Kf2 Rc2+ >51.Kg3 Rc3+ (somewhere around here Fritz started to announce mates) 52.Kh4 Rc1 >(after 52...Rd1 we have 53.g8N+! and mate) 53.g8Q Rh1+ 54.Kg3 Rg1+ 55.Kf4 Rf1+ >56.Ke5 Rd5+ 57.Ke6, and if you really want to go on, it's 57...Rf6+ 58.Kxf6 Rd6+ >59.Ke7 Rd7+ 60.Kxd7 c5 61.Qg6 mate. Another line we checked with Fritz was >50...e3+ 51.Kg2 e2 (51...Rc2+ 52.Kh3 is mate in 8) 52.g8Q Rxg8 53.fxg8Q Rg1+ >54.Kf3 Rxg4 55.Qh8+ Kg6 56.Qe8+ Kf5 57.Qf7+ Ke5 58.Kxg4, which had the whole >team laughing. > > > > [D] 4r3/8/2p2PPk/1p1r4/pP2p1R1/P1B5/2P2K2/8 b - - > >kburcham
This page took 0.01 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.