Author: Don Dailey
Date: 11:32:47 06/30/98
Go up one level in this thread
On June 30, 1998 at 04:14:33, Ed Schröder wrote: >>Posted by Bruce Moreland on June 30, 1998 at 02:24:24: > >>In Reply to: Re: Rebel10's anti-GM revised... posted by Ed Schröder on June >>29, 1998 at 18:56:48: > >>On June 29, 1998 at 18:56:48, Ed Schröder wrote: > >>>>Posted by Bruce Moreland on June 29, 1998 at 18:07:31: > >>>>I think it is about marketing, personally. It's kind of a "Rebel 10 -- >>>>now with more brighteners!" kind of thing. Who knows what a brightener >>>>is? > >>>>I am not trying to get your secrets out of you, I think it is dubious >>>>to start a discussion of a feature without giving a little more detail >>>>about the feature, to do otherwise isn't to invite discussion of a >>>>computer chess topic, it's more to invite discussion of Rebel. > >>>Ed is on a marketing trip, Ed is dubious, Ed makes noise.... > >>>That's not a good starting point if you want an answer. > >>I am not a syncophant come seeking knowledge. > >>The original post was a pointer to your home page. At that location is some >>text that says that you have changed Rebel's style and that you think it is >>stronger. > >>There is a vague reference to speculative play, and an example of a position >>where the key move is found sooner. > >>There are no details about what you did to achieve this, although you gave the >>feature a name. > >>How can anyone discuss this feature, which is ostensibly the point of this >>thread, since the name of the feature is in the title of the thread, without >>knowing what the feature is? > >>I'd rather discuss the feature than listen to people say how this could be the >>next great thing without knowing what it is. > >I have been quite informative about what anti-GM is and how the idea was >born. I will add some more (positional based) examples to my home page >soon. > >Bottom line, strong chess players try to find out the weak parts of a >chess program. It has a name, it's called "anti-COMP" play and if it >is done properly your (my) (all) program(s) are in deep trouble. > >"Anti-GM" is a first try to shoot back. It's my opinion every chess >program needs such a piece of software at least if you want to compete >in the GM area. > >If I can convince you that the latter is true then I have given you >already 50% of the solution. > >- Ed - > > >>bruce Hi Ed, I can appreciate Bruces point of view on this. Of course you have no obligation whatsover to tell us any of the details, but you are claiming that you are! Did I miss the posts that explains your strategy? Everything I know about this from all your posts can be summed up in one sentence, "Play in such a way as to beat Grandmasters." It sounds intriguing and now everyone is hailing it as the next breakthrough. If you are reluctant to give detail we will respect that. But just like Bruce says, we have nothing yet to talk about. And the idea is very very old (in computer chess terms that is.) And now please forgive me for a little sarcasm. I hope you take it in the good natured way it is intended: Humans are able to play Grandmaster level chess by only looking at a few nodes and yet it requires many millions of nodes for computers to even approach this level. I am convinced that this is due to the fact that our programs are not selective enough. Therefore I have concieved of a new strategy called human-selective which models a search after a human style search. And just wait till you see it in action, no more 15-20 ply limits on search depth. No more looking at millions of ridiculous and silly moves. Finally a chess program that does the obvious thing and only looks at plausible moves in the tree search just like humans do! I think everyone will agree this is a breakthrough idea. Isn't it a great one! - Don
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.