Author: Dann Corbit
Date: 16:49:51 07/31/01
Go up one level in this thread
On July 31, 2001 at 19:40:26, stuart taylor wrote: >On July 31, 2001 at 16:02:56, Dann Corbit wrote:> >>3000 ELO is completely meaningless unless you specify precisely the competition. >>Against all the kindergarten children in the US, the weakest chess engines have >>an ELO of 3000. > >Dann, you always seem to be quite ready to spoil alot of fun, don't you? >Or is it perhaps that your honest mind puts alot of fun into question for you, >and you get frustrated that others speak blindly without content, whilst you >would like to see answers from people which DO put content back into it, and >therefore you probe for people to answer to those needs for some real content? These questions are just as fun for me as a carefully directed question. Perhaps more so. I think it would be valuable to have people understand what they are really asking. If the question must be rephrased to have an accurate answer, then it should be rephrased. If you don't care about accuracy, then why ask the question? This is the sort of thing that cannot be accurately answered by a poll. It requires data. The data is being accumulated. Eventually, it will be complete. For sure, at some point, the best machines will be better than the best humans. I doubt if that day has already arrived, but the arrival is surely inevitable. >Well, I would simply say that we are speaking aproximately, at todays ultimate >human competition level e.g. FIDE or PCA (or whatever It's called). > >Perhaps you can find ANSWERS on how chess level can be more fixed numerically or >otherwise? These answers (e.g. equating FIDE levels with SSDF for example) could be found by experimentation. In other words, play thousands of games using programs against people. Gradually, this is happening. Eventually, we may have some sort of strongly reinforced [sure] answer. We have an approximate answer already which is, "Computers are very, very strong. They may be equal to GM level on good hardware." Kind of fuzzy, and not really satisfying, but that's what we have right now. There is also expert opinion that confirms this assessment. We already know that the programs are good and strong. Probably what is really being asked is "When will the best program be better than the best person?" I don't think (for instance) that the upcoming Kramnik match will answer that question one way or another. But it were repeated over and over with the same result, that would answer the question.
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.