Author: Matthew Hull
Date: 08:22:50 01/27/03
Go up one level in this thread
On January 27, 2003 at 02:15:29, Bruce Moreland wrote: >On January 26, 2003 at 22:02:11, Thomas McBurney wrote: > >>On January 26, 2003 at 21:37:27, Derek Paquette wrote: >> >>>AM I right to say this? >>>Have any GM's BESIDES kram and kasp (who don't count, they were in the match) >>>said any things towards this subject? >> >>Deep Thought II could search around 10 million nodes per second, and this is >>using hardware designed in 1985. Deep Blue, I think was around 100 times faster >>than Deep Thought. Gary Kasparov easily defeated Deep Thought. So given that >>DJ probably isn't even anywhere near as fast as Deep Thought II then I don't >>feel that it can compete with the likes of Gary. > >The problem with this is that you don't know, you have to guess. This is not >your problem, the problem is that the machine is unavailable and does not >compete. > >As someone who has some interest in saying true things and in not saying false >things, I find it very difficult to say anything about DB, or for that matter >DT. > >Absolutely no offense intended, but listening to anyone try to figure out DB is >like reading medieval scientific writings. Good conclusions are hampered by >lack of evidence and the necessity to avoid contradiction with religious dogma >and previous scientific work that is deemed to be undeniably true. > >The fact is that unless the machine is produced and examined, we cannot conclude >anything about it. > >IBM broke the machine down because they had achieved their goal -- the general >population thought that they -- IBM -- had solved chess for all intents and >purposes. I oppose this as completely as I can, for two reasons: > >1) It did immense damage to the field of computer chess. The general population >thinks that chess is solved. I would not say solved, but that IBM's machine had surpassed human capability. I don't think the public understands what solving chess means. >2) The entire thing is not how science is supposed to work. Science does not >take a result that may very well be due to chance, and refuse to attempt to >repeat it. As you say below, it was not a scientific excercise, at least from upper management's POV. Hsu & Co. where the real scientists, and the few GMs that participated will have a better impression of it's real strength. But the crowd at this forum will believe nothing said by the DB team, unless they can lay hands on the beast themselves. Some would not believe even then. So there it stands. > >I absolutely refuse to allow anyone to make an unchallenged assertion that DB >was strong at all. The fact is, we do not know if it was strong or not. I think the GM consultants who played the thing in the lab know. > >DB was a PR project. It was not a scientific project. The stated goal of the >DB project was to produce as many "impressions" as a Super Bowl commercial, for >less money. Absolutely true -- no joke. > >I give IBM's result as much credence as I give to that UFO cult's assertions >that they have produced a human clone, Well, we have seen more of DB2 than of any purported clone, so your statement is not quite on target, IMO. > and I think that it is sick that the >computer chess community stands for this. Not sure what this statement is saying. But it is a misfortune that IBM management were so hopelessly insensitive to the historic value of their achievment that they tossed the fruits of their labor, for all intents and purposes, into the trash can. However, it not uncommon for corporate mangement to be lacking in imagination and basic intellectual curiosity. That's life. Regards, Matt > >>After watching the game today, I get the impression that DJ is broken. The >>moves e5 and Re8 look like stupid moves and I cannot replicate these moves with >>any of my chess programs even after two hours of thinking. Does anybody know if >>the DJ team has been fiddling with the program in preparation for the match? > >I don't believe that they are allowed to fiddle with it. > >It is possible that it is broken, but it is also possible that: > >1) It has been tuned to play very actively, and the machine decided to pitch an >exchange, in a difficult position, for some random reason involving perceived >compensation. My own program certainly does this from time to time, even when >the position is not necessarily difficult. > >2) The strong hardware combined with DJ's well-known great strength created a >situation wherein DJ knew it was toasted, and it thrashed out a move when it saw >that its original choice was going down the drain. > >Either of these is very possible. > >bruce
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.