Author: Roy Brunjes
Date: 07:02:00 02/08/03
Go up one level in this thread
On February 08, 2003 at 02:45:19, Uri Blass wrote: >On February 07, 2003 at 19:54:41, Roy Brunjes wrote: > >>On February 07, 2003 at 19:51:09, John Merlino wrote: >> >>>On February 07, 2003 at 19:01:46, Will Singleton wrote: >>> >>>>It's all a sham. Sorry to take the minority opinion, but it is clear to me that >>>>GK is building up DJ, for the sole purpose of a rematch. >>>> >>>>I just heard him say that DJ is the "best program that has ever existed." >>> >>>Additionally, if you saw the pre-recorded interview with (I think it was) Amir >>>Ban, he said something like "Deep Junior was, THREE YEARS AGO, superior to Deep >>>Blue" (capitalization added by yours truly for emphasis). That seems a bit >>>far-fetched to me. >>> >>>What say you all? >>> >>>jm >> >>Of course this is a slippery one to debate. I personally believe that DJ is so >>much slower than Deeper Blue that it would be no contest. Chess programming is >>one art form where speed kills (to an extent). And the extent to which Deeper >>Blue is (was) faster than Deep Junior 8 is roughly 100. Sorry, I just cannot >>see DJ holding up under the strain of all those tactics that the 100x >>performance difference inevitably brings. >> >>Roy > >I do not believe in your theory. > >1)I believe that in 120/40 speed is not that important(it may be interesting to >try commercial programs of the time of deeper blue(beginning of 1997) against >Junior8 when the programs of 1997 get 10 hours per move against 6 minute per >move of Junior8. > >I am not so sure if the programs of 1997 will win the match. > >2)I also believe that not having bugs is important and the deep blue team simply >not had the time to test. > >3)I also do not know the speed of deeper blue. >It is possible that they did only 20M nodes per second and we have no proof >except what IBM said. > >I do not see a reason to believe IBM after Deeper blue stopped to compete. > >It is natural if people suspect that you have something to hide if you stop to >play. > >Uri Uri, Your comments are very good ones! I would also like to see a match as you state in point 1). I am not sure someone would want to dedicate that much time (and the necessary many hundreds of games to get statistical relevance) though. It would be interesting, no doubt! Any takers?? I suppose this would mean something like Fritz 5 (16 bit?) My memory of the software available in 1997 is not so good! Roy
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.