Author: Tom Likens
Date: 08:54:06 04/27/00
Go up one level in this thread
On April 27, 2000 at 10:12:13, blass uri wrote: >> >>A large reason for the dramatic increase in computer chess has been the >>fantastic increase in home computer processing power. Most of the best >>programs are tactical monsters. But, it is highly doubtful if any program >>has an evaluation anywhere near as sophisticated as the top Grandmasters. > >Impossible to compare because grandmasters do not think in the same way as >computers. > >I think no GM thinks something like this: >e4 gives me after analysis 0.27 pawns advantage and after analyzing d4 I found >that it gives me 0.28 pawns advantage so I will play d4. > True, but ultimately the evaluation function is created by a human being (barring genetic algorithmic tuning, which has had only limited success). This person imparts their understanding of chess to the program (limited of course by what can easily be represented and programmed). And grandmasters will always, by definition, have a better understanding of chess than us mere mortal programmers (ignoring endgame tablebases which are a bit of an abberration since they represent perfect knowledge). On the other hand it is often difficult to quantify the knowledge grandmasters possess because they have no concept of what can and cannot be easily programmed. I shudder to think of how strong a machine would be that was as fast as Deep Blue but applied the complete knowledge of Kasparov at each of the 200 million nodes it searched each second. Much of that knowledge would be wasted of course on ridiculous positions, but it would explore many branches that on the surface would seem absurd but would ultimately prove very very clever. >> >>And this is where IM/GM can make contributions for a long time to computer >>chess. I don't believe it is any coincidence that after Joel Benjamin >>joined the Deep Blue team it made an incredible leap in strength. Even >>though Joel is nowhere near Kasparov's strength. > >I do not know if it made an incredible leap in strength thanks to better >software or thanks to better hardware. Granted, it is very hard to quantify the Deep Blue result. More than likely, a combination of both (like every chess program). > >Uri --Tom
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.