Author: Jeremiah Penery
Date: 19:44:16 10/17/02
Go up one level in this thread
On October 17, 2002 at 08:48:41, Uri Blass wrote: >On October 17, 2002 at 08:13:38, emerson tan wrote: > >>On October 16, 2002 at 20:27:57, martin fierz wrote: >> >>>On October 16, 2002 at 17:57:10, Gian-Carlo Pascutto wrote: >>> >>>>On October 16, 2002 at 17:53:25, Jeremiah Penery wrote: >>>> >>>>>On October 16, 2002 at 16:43:53, Gian-Carlo Pascutto wrote: >>>>> >>>>>>I find it tasteless that Hsu claims that Deep Blue was in a 'different >>>>>>class' from Fritz and that it was not match for them, when he has no >>>>>>ability to back it up and the only public things we have about the two >>>>>>show differently. >>>>> >>>>>How do they show differently? >>>> >>>>'very old' Fritz beat 'very old' Deep Blue >>> >>>...and in a shveshnikov book line at that which was in the fritz book and not in >>>the DB book AFAIK. using one game alone as a proof is not a good idea. and if >>>that game was decided by the better opening book, then it gets downright >>>ridiculous. >>> >>>aloha >>> martin >>> >>> >>Kasparov wasalso beaten by deep Blues opening book in game six and kasparov loss >>by a point. Thats one game up as proof also. Deep Blue team started claiming >>victory. downright ridiculous also? Kasparov played a bad opening, deliberately, probably because he could beat Fritz at the time by using that opening. >I think that both games were lost not because of bad opening line. Maybe the bad opening was not the immediate cause of the loss, it definitely was a big contribution. >In the case of the game of deep blue prototype against Fritz3 the reason was >clearly a tactical mistake. Correct. 'DB prototype' in 1995 playing a couple tactical mistakes means absolutely nothing about the 1997 machine, however. They had gigantic bugs in 1995, and nobody ever said their evaluation then (or in 1996) was all that much better than anyone else's. >They have excuses for that tactical mistake but I do not care about it. >I do not know how much time did deep blue use for c4 but time management is part >of the program and they should expect the circumstances that happened. > >I do not know if it lost connection after failing low for some objective problem >but even in that case it could remember the fail low and use more time. They lost communication to the machine, during which time it had already found a better move. When they reconnected, they had to completely restart the search. They had some major bugs, which caused them not to be able to find the better move again. Even though it had failed low, it didn't produce a better move by the timeout, so it played the fail low move and ended up losing.
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.