Author: Christopher R. Dorr
Date: 15:04:30 01/05/99
Go up one level in this thread
On January 05, 1999 at 17:50:48, KarinsDad wrote: >On January 05, 1999 at 16:53:34, Christopher R. Dorr wrote: >> >> > >[section snipped] > >>This raises the larger question of whether or not we are nearing the top >>strength of current algorithms. Are we near that limit? Hardware will of course >>make things better, but really, how much has the absolute state of the art >>increased over the last couple years. A couple dozen rating points? If that? >> >>If there were significant advances to be found in the fundamental algorithms >>that underly virtually all modern chess programs, I think that somone would have >>found them. There have ben advances in the implementation of those that >>currently exist, but when is the last time somebody had a fundamental change in >>the way their program plays chess that brought them many points? > >Well, using the example that nobody has yet had a major paradigm shift to >improve their algorithms in a major way to prove that it cannot be done doesn't >quite fly either. Right now, I am working on a chess program that has at least >two features in it which have the potential to significantly improve the >selection process that I have never heard of anywhere in the newsgroups, on the >internet, or in the journals. Does that mean that Fritz or Junior doesn't >implement one or both of these features, at least in some manner? I have no way >of knowing. The commercial developers are (and properly so) closed mouthed about >the internals of their programs. Will my advancements result in the next leap in >computer chess technology? Probably not. At best, I think I can eventually be >competitive with the leaders. However, the point is that if I am successful in >creating a competitive program, how much better would an experienced team with a >lot of resources behind them be? There were a lot of people who felt that >Kasparov would never lose a match at tournament speeds to Deep Blue either. > >There seems to be one constant in the universe. Never say never. No matter what >concept seems farfetched today, tomorrow will often prove it to not be so >(although you are perfectly justified to look at the track record for the last >10 years and say, "Hey, we haven't had a major leap yet and nothing appears to >be looming on the horizon."). I guess the answer to this question is beyond the >human "event horizon". > But in a way, this emphasizes my point. You are making some interesting leaps forward *without* 50 engineers and a zillion bucks. Perhaps the implementation would be easier if you had them, but how would their presence have made the *idea* come faster or clearer? Individuals make these kinds of intuitive leaps, not groups. Groups combine to make advances and discoveries, yes, but they don't have an 'intuition'; they don't have a way to make a connection with their past experience, because they are composed of the individuals. The individuals make the discoveries. Bob Hyatt has essentially worked alone on Crafty; Yet it is nearly as good as anything made by and company or group. He made the advances himself. They were a product of his knowledge and history; his understanding, education, and intuition. Having 50 assistants might have made some of the coding easier (or not, as anyone who ever did 'team programming' will tell you), but the ideas that made the advance came to *him*, not to some group. I'll never say never (well, I guess I just did:) , but there are some things that money and manpower smply cant hurry or advance quickly; I feel that chess programming is one of those things. Cheers, Chris Dorr >:) > >KarinsDad > >> >>In short, Microsoft could buy every chess program on this planet, and combine >>the best of these to make a better program than exists now. No argument. But >>could they significantly advance the state of the art (as the original poster >>stated that a former MicroSoft manager told him)? I really don't think so. > >PS. Without a single improvement in the basic engine and just with improvements >in the opening books, learning routines, tablebases, and GUIs, you do advance >the state of the art. > >> >>Chris Dorr >> >>
This page took 0.02 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.