Author: Vincent Diepeveen
Date: 22:25:51 01/28/00
Go up one level in this thread
in all those cool graphs dudes take not take into account a couple of things. First of all we had hashtables, then nullmove then now EGTB, way better books recently, some work done at eval everywhere. Some like me are even busy making a huge eval. Much of the 'lineair curve' so far that some want to show as evidence is for largest part because of quality advances in the software, *not* the hardware. If you put an old program at nowadays hardware, then those don't play as well. I'm always surprised that such improvements in software get neglected by graph makers. Vincent On January 28, 2000 at 17:26:47, Robert Hyatt wrote: >On January 28, 2000 at 00:41:49, Len Eisner wrote: > >>On January 27, 2000 at 21:50:08, Christophe Theron wrote: >> >>>On January 27, 2000 at 18:01:39, Peter Kappler wrote: >>> >>>>On January 27, 2000 at 13:52:38, Christophe Theron wrote: >>>> >>>>>On January 26, 2000 at 20:06:04, Peter Kappler wrote: >>>>> >>>>>>On January 26, 2000 at 18:23:50, Bruce Moreland wrote: >>>>>> >>>>>>>On January 26, 2000 at 18:10:10, Dann Corbit wrote: >>>>>>> >>>>>>>>IOW, more horsepower is a tough way to make chess programs play better. There >>>>>>>>is also evidence (according to some) that the increase in speed has >>>>>>>>*diminishing* returns. Hence, it may take a terahertz to get there. Don't know >>>>>>>>of any material that could do that, not even a Josephson Junction. >>>>>>> >>>>>>>I think it's a great way. You just take a vacation, preferably a long one, and >>>>>>>when you come back you make one call to Gateway and poof, free Elo points. >>>>>>> >>>>>>>Got an article that shows that the Elo curve flattens out with increased depth? >>>>>>> >>>>>>>bruce >>>>>> >>>>>> >>>>>>Ironically, Dann and I recently debated this very topic. If you use the 7-day >>>>>>filter, you can probably find the discussion under "DBx1000 = how strong?" or >>>>>>"tactical sufficiency threshold". >>>>>> >>>>>>It seems obvious to me that the curve must flatten out with increased depth, but >>>>>>I'd be fascinated to see evidence to the contrary. >>>>>> >>>>>>--Peter >>>>> >>>>> >>>>>I'd be fascinated to see evidence of what you say. >>>>> >>>>> >>>>> Christophe >>>> >>>>Hi Christophe, >>>> >>>>I must admit that seeing you and Bruce on the opposite side of the fence has me >>>>beginning to doubt myself. :-) >>>> >>>>I find this topic particuarly interesting, as I think it says a lot about the >>>>basic nature of chess. >>> >>> >>>You are absolutely right. >>> >>>Chess programmers do not, I think, understand so far the "basic nature of >>>chess". At least I claim that I do not understand it. >>> >>>Every time I manage to make a new algorithm work, I learn a little bit more >>>about it. >>> >>>I think that chess programming brings us, year after year, closer to a better >>>understanding of the game. >>> >>>Every assumption about the game must be very carefully checked. Most of the time >>>the assumptions I made were WRONG, and proved wrong by experimenting. >>> >>>As I just consider myself as a normal guy (not overly smart but no too stupid >>>either), I would advice anyone to be extremely careful with common sense >>>assumptions in computer chess. >>> >>>That's why I am very skeptical when somebody comes here, explain in length a >>>smart looking theory, but is unable after that to prove it works by producing a >>>very strong program. A theory that pleases the human mind is something I always >>>find suspicious. To name a few: >>>* Dimishing returns (or Tactical barrier) >>>* Slow=positional, Fast=tactical >>>and so on... >>> >>> >>> >>>> My intuition tells me that in most chess positions, a >>>>certain search depth is sufficient to "understand" that position, and avoid any >>>>serious problems. Searching beyond that depth would give diminishing returns. >>> >>> >>>For a human player maybe. Not for a computer program. >>> >>>The human brain loses itself in the variations after a while. A computer program >>>does not. >>> >>> >>> >>>>It seems like a simple experiment to conduct. Just play lots of self-play games >>>>between an n-ply and n+1 ply search, and after 500 games, increase n. Keep >>>>increasing 'n' Surely these experiments have been conducted. >>>> >>>>If you can point me to any relevant articles, I'd be grateful. I have read the >>>>"Crafty goes deep" article in ICCA, but the results didn't seem conclusive to >>>>me. >>>> >>>>--Peter >>> >>> >>>Ernst Heinz has conducted the same kind of experiment and published the results >>>as well in the ICCAJ. >>> >>>He also posted here that he has treated the subject in his new book, and also >>>said that so far no experiment has been able to prove the "dimishing returns" >>>phenomena. >>> >>> >>> >>> Christophe >> >>Let?s do a little thought experiment. Imagine a program that could search deep >>enough to solve the game of chess. I know it?s not possible today, but just >>imagine it. Now, if the deepest possible search solves the game, how can there >>be diminishing returns for increased search depth? >> >>Len > > >That is the trivial case. Because you do not search deeper when you find a >forced mate. But on the iteration before the last one, going one ply deeper >certainly does make a difference. :)
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.