Author: Jeremiah Penery
Date: 17:43:22 01/11/03
Go up one level in this thread
On January 11, 2003 at 19:23:20, Omid David Tabibi wrote: >On January 11, 2003 at 08:23:43, Frank Phillips wrote: > >>"The next question is, and many people are asking it, do we know how Deep Junior >>compares in strength with Deep Blue? The really interesting thing, from the AI >>point of view in general and for computer chess researchers in particular, is >>that Deep Junior examines something like one percent of the number of positions >>per second of Deep Blue. But despite this Deep Junior may well play better chess >>because its "understanding" of the game is better. It appears to have more chess >>knowledge and understanding in its evaluation function than Deep Blue did, and >>this compensates for the difference in positions-per-second.." Extract from >>Levy on Chessbase.com site >> >> >>From what I read in Behind Deep Blue I find this surprising. But then again, I >>no nothing about Junior other than it is an awesome program. >> >>If only Hsu had produced his chip so we could have answered this question rather >>than use it to fires. >> > >If say Deep Blue was 100x faster than Deep Junior, then I suggest that you >conduct the following test: > >Take an engine which has a very simple evaluation function, and its speed (NPS) How do we know this would compare in ANY WAY with Deep Blue's evaluation function? Even if both are very simple, they can be very different. >is about Junior's speed, turn off all its selectivity (e.g. null-move pruning, Deep Blue obviously had _some_ selectivity. And to make things even close to being a good proxy for DB, you'd have to implement all sorts of extensions that they did. >futility pruning, etc), and let it play against Deep Junior. The time control >should be 100x in favor of the brute force engine, e.g. 500 min/game for the >brute force one, and 5 min/game for Deep Junior. > >I will gladly bet on a convincing win for Deep Junior. In the experiment _you_ have set up, I'd agree with you, given some minimum sufficient time for DJ - i.e., one minute/game for DJ would cause it to lose because of insufficient tactical sight, IMO. In the theoretical experiment against Deep Blue, we have no clue, for several reasons. Maybe the biggest reason is that we really don't know what exactly DJ OR DB do in their search/evaluation. We can guess, and maybe we can come close, but I don't think that's enough.
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.