Author: Uri Blass
Date: 16:09:14 07/10/05
Go up one level in this thread
On July 10, 2005 at 18:29:53, Djordje Vidanovic wrote: >On July 10, 2005 at 16:32:28, Uri Blass wrote: > >>On July 10, 2005 at 12:36:49, Djordje Vidanovic wrote: >> >>>A couple of months ago I raved about the new Shredder 9. While I still think >>>that Shredder 9 is amazing, the new Fruit 2.1 seems to have opened my eyes with >>>regard to Shredder 9. While I was not quite sure how to account for Shredder >>>9's playing strength at the time I posted here, now things appear clearer to me: >>> Shredder 9 is the best *searcher* of all programs and its search was much >>>improved at the expense of "knowledge". My original idea was that Shredder's >>>author had coded more knowledge about space, tempi, some new extensions in the >>>Q-search etc., but Fruit 2.1 and its evident success against Shredder 9 in their >>>head-to-head battles on playchess.com have shown me that the secret of Shredder >>>is definitely in its search. How else would a simple, efficient deep-searching >>>program such as Fruit 2.1 (knowledge-wise simple:-)) manage to be in a dead-heat >>>race with Shredder 9 for spot #1? >>> >>>Any ideas? >>> >>> >>>Djordje >> >>I do not think that Fruit2.1 is simple. >> >>I did not learn most of it's evaluation and I hope that learning it's evaluation >>will not only help Movei but also help me to be a better player(hopefully to the >>level of 2200 fide rating). >> >>About myself: >>I think that I am in the level of 2000-2050 fide rating at 2 hours per game or >>90+30 fisher time control inspite of the fact that my last Israeli rating is >>2103 because I have to admit that I had a lot of luck in my last tournament >>games including winning on time against a player with 2100 from a lost position >>with 30 seconds incresement only because he forgot to look at the clock). >> >>I also got amazing result of 9.5/10 against players with rating 1800-1950 when >>in half of these games I came out of lost positions or inferior positions. >> >>Note that I expect to save lost positions against weaker player and sometimes to >>win but what happened in the last israeli update is clearly above my >>expectations and normally I do not expect to get more than a draw in endgames >>when I am a pawn down and it happened twice. >> >>Uri > > >Interesting personal memoir, I read it with pleasure. However, I fail to see a >any connection between the topic I addressed and your description of your chess >skill. > >I still think that Fruit lacks a lot in terms of knowledge and I am sure that >its search is simple, elegant, and optimal. All of which explains its amazing >depth and precision. I am sure fruit(like other programs) lacks a lot in terms of knowledge but it may know things that I do not know or I underestimate or overestimate so I think that knowing it's evaluation may help humans. > >To prove what I just said, please run the "Marathon" tactics test from the >Chessbase suite and see how Fruit fares there. You may compare it to Fritz, >which is, oh, what a paradox, now a program that comes with one of the largest >chunks of chess knowledge. What do you learn from comparing between fruit and Fritz in tactical test? The program that does better may do it thanks to better evaluation or thanks to deeper search. personally I found that improving the evaluation helped movei to do better in tactical test suites. 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.