Author: David Dory
Date: 20:59:20 04/23/03
Go up one level in this thread
On April 23, 2003 at 21:54:18, stuart taylor wrote: >On April 23, 2003 at 17:54:12, stuart taylor wrote: > >>About what? >>Well, if I don't say about what, it could only mean one thing, if you think >>logically. >>I'll finish my question a bit later. >>Big clue....I'm posting this on CCC. >> So from that you should not only understand the subject, but even exact >>question IMHO. >>S.Taylor > >Honestly, It seems that no one understands. >I mean that since this forum is about computer chess, so I'm asking if we can >really complain about the present level of computer chess, i.e. playing strength >of the top programs. i.e. even though there may be much more yet to improve >upon, can we really say as consumers that we simply don't yet have the product >we have been hoping for? > >In other words, Is there something we simply cannot do yet, since the playing >strength is not strong enough? > >Presumably, now that we have Fritz 8008, and Shredder 7.04, and about 3-4 which >are very close. > >Perhaps a year or two ago, we DID have what to complain about, or certainly 3-4 >years ago. But DID we then? If so, do we now? > >Of course, it depends what you want playing strength for. If all you want is >that a program should give you a hard time, I'm not such a fool as to ask be >asking if we have reached that stage yet! > >(I'm also thinking at present, that there's a limit to how great playing >"strength" can be, without great wisdom being seen, such as planning etc. So >when we speak of greater strength, we are thinking of wisdom, as we have almost >come to limits of what can be done without great chess wisdom, havn't we?) > >I have my own thoughts on this question, but It's still not clear to me what >people generally feel about it, and the specifics. > >Could I please hear your opinions and evaluations? >Thanks >S.Taylor I thought the first post in this thread was a troll. :) Yes, I think we have something to complain about, still. There are many positions where the program has no clue what to do, simply because the answer is over the search horizon. Despite all the hardware improvements, the search still looks at many thousands (oftentimes 100's of thousands) of positions to find the best move. Meanwhile, the human might examine 10 to 20 possible moves only, to find the same best move. Anyone looking for "wisdom" in a chess program has long since doubled over in nausea at the dismal CC results returned by the AI hotshots of the last two decades. All their high expectations and proclamations turned out to be nothing but a brain fart. The triumph of CC has come only from those working "down and dirty" with broad searches, and many incremental improvements. The best news is the lower branching factor several programs have created. It's great to see the programs getting better - but that's not to say most of them are much "smarter". Our EGTB's are moving the "stupid" bar away from the end game, while more extensive openings are moving said bar further away from the opening. It makes the programs stronger, but NOT smarter, anywhere they have to move "on their own", in between. We still have to make excuses for the programs "oh well, this is a zugzwang problem that isn't detected while in null-move, (for faster answer, please choose your excuse by number next time)". After all these years, we shouldn't have to make excuses for the program's chosen move. <My favorite personal gripe with man/program competition:> ========================================================== Although EGTB's are fascinating info, I believe they mock the spirit of man/program competition, and shouldn't be allowed for same. What are EGTB's except a HUGE cheat sheet, far surpassing the memory of any human? Fine for program vs. program, chess study, etc., but they should be banned for man/program competition. <end gripe> ========================================================== As a consumer, I'm happy as a pig in mud. As a fan of CC, I'm unsatisfied that we haven't found some very new and insightful way to code up chess "wisdom" very successfully. Maybe the 64 bit CPU's will help make this possible. Dave
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.