Author: blass uri
Date: 09:55:09 06/13/00
Go up one level in this thread
On June 13, 2000 at 12:35:55, Dann Corbit wrote: >On June 13, 2000 at 06:34:38, blass uri wrote: > >>On June 13, 2000 at 05:45:35, Dann Corbit wrote: >> >>>On June 13, 2000 at 05:01:25, blass uri wrote: >>>[snip] >>>>The problem is that a lot of humans do not want to play against programs in >>>>their best. >>>> >>>>The result is that computers are not allowed to play in most human tournaments. >>>> >>>>If you want to see more comp-human games in tournaments the only possibility is >>>>to have limitations for the programs in part of the tournaments. >>>> >>>>I think that limitation about the hardware is the only possible idea to get more >>>>comp-human games. >>> >>>If this is true, then the GM's are cowards. IMO-YMMV. If they would expend >>>effort learning how to defeat computers they would not be so unnecessarily >>>fearful of them. Even at that, though, someday the computers will be better. >> >>I think that most of the GM's do not want to play against computers. >> >>We can have some computer-humans games when computers have no limitation but if >>we want to have more games we need to do limitation of the hardware in part of >>the tournaments. >> >>I am interested to see the games between computers with no limitation about the >>hardware and humans but it is also interesting to know if programmers can do >>better programs when there is a limitation about the hardware that cause >>limitation about the size of the program. >> >>Doing the best program when the engine cannot use more than 100Kbytes when the >>size of the engine is included in the 100Kbytes may be also an interesting >>competition. >> >>I prefer this idea about what happened in the Israeli league when the rule said >>that the software must be the same in all the games(except opening books) >>because in the israeli league if a program has a bug you can do nothing to fix >>it and in this case the programmers can change things between rounds. > >Make way for "gag" programming if that stance becomes a reality. DarkThought >already implements much of the endgame information as algorithms and has >compacted the endgame database so much that it fits into ram. > >Besides which, the endgame database files only add 50 ELO or so. > >If you look at the Fritz 4 opening book, you can easily see that it could be >encoded in perhaps 4K. From each of the possible 20 starting root moves, each >subsequent choice can be encoded in a single byte. > >Tiny hash tables can be overcome by ludicrous CPU speed or (better yet) by >multiple CPU's. Gigahertz CPU's and multiple CPU systems are already possible. > >I think "Make the computer fight with one arm tied behind its back" will only >make the GM's look a lot worse, not better. In fact, that is a no-win >situation. > >"I beat the computer!" > >Yes, but it had one hand tied behind its back. > >And if you lose -- even worse. I think that today against GM's it is better to play with the best hardware because the best hardware is not good enough to be almost sure that the programs win and the idea to give program to play with one hand tied behinds its back against GM's is only relevant for the future(not the near future) but if we want programs to play also against weaker players of 2000-2100(programs stopped to play in tournaments against these players because they won too much) we can do it by limitation about the programs. I agree that database is not so important but if a program cannot use more than 100 kbytes then I am not sure if all the search rules of today programs can be used because the size of the engine may be more than 100 kbytes so programs will be even weaker. If we use 386 for the programs it also can help humans to have better chances and suddenly humans will have chances against programs and we may see programs again in tournaments against humans. Uri
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.