Author: Dann Corbit
Date: 01:23:25 07/11/00
Go up one level in this thread
On July 07, 2000 at 06:26:46, Adrien Regimbald wrote: >Hello, > >As with any other tool, if misused, CAP data might be harmful to your program's >success. Many people have used the CAP data to great success within their >program, so I think that the problem is not with the CAP data, but in how you >are using it. > >I would also like to say that I think your posts on this subject are verging on >hostile towards the CAP project and Dann. Both Vincent and myself act like whinging sacks of cat vomit from time to time. I take no offense at Vincent and (for that matter) I am just as fiery in return. In fact, I emailed an apology to Vincent for my tone. At any rate, some people are very "matter of fact" such as Dr. Hyatt and Vincent Diepeveen. I don't take any offense at his remarks. In fact, I value his criticism. There are two reasons it has value. First, if we accept any research without question, we are setting ourselves up for a fall. And second, some people might not be aware of some of the issues involved in using this data. Hence this discussion had a dual valuable purpose. >Dann has done a lot of terrific work >for computer chess, and many people find the results of his CAP project >extremely useful. [Feigns false modesty with a blush] >This information is made freely available, with loads of >volunteer effort going into it. It seems to me that complaining about it is in >very poor taste. > >As far as hand generated books go, I think that most authors are willing to >accept the position that will occur maybe once every 100 games where the opening >book might be wrong, rather than have the opening book be 100% hand verified >(which doesn't necessarily even mean it's any better), and leaving book on move >3 or 4. There is a compromise here - would you rather accept a small mistake >once every hundred games, or get generally smacked about in the opening in every >game? I personally would prefer the former, and apparently so do most authors. I think a hybrid approach is best. A hand tuned book by top experts is going to be better than an automatic book. However, it will cost millions of dollars to hand evaluate many millions of positions. In short, it's not gonna happen any time soon. So you are left with a small (perhaps nonexistant -- depending upon the size of your wallet) book that is written by hand, and a much larger automatic book. I think that even the hand-tuned book might possibly benefit by using the right mathematical approach. As it stands, CAP is raw information in a giant "landfill-sized" heap of data. What to do with it is anybody's question and nobody has all the answers to that question yet. However, I feel that it may prove to be incredibly valuable. We certainly don't know its import yet at this early juncture. If nothing else, it shows that basically impossible tasks can be handled by international cooperation and made freely available to the public. Perhaps that will be its greatest value. Universities, corporations and individuals have all made major contributions. The first project, taking several months, would now be done in one day and much better than that first stab at it. If you look at the CPU hours involved, it is simply mind-boggling. I already have enough data to spend a lifetime of enjoyment in research and rediscovery. I have no idea where it will end and even if it will end. In any case, the experiment is an unqualified success in my view (at least for me, which [from my selfish standpoint] is good enough). Will it make the world's best opening book? Probably not at this juncture, but it beats a sharp stick in the eye.
This page took 0.01 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.