Author: Uri Blass
Date: 11:40:58 06/04/01
Go up one level in this thread
On June 04, 2001 at 13:41:45, Uri Blass wrote: >On June 04, 2001 at 13:09:42, José Carlos wrote: > >>On June 04, 2001 at 10:12:36, Christophe Theron wrote: >> >>>On June 04, 2001 at 05:29:47, José Carlos wrote: >> >>[snip] >> >>>> I think you misunderstood him. He said 'material only', which means that the >>>>program choses the first move it generates and sticks to it unless it sees it's >>>>losing material. 'h3' is just a move that doesn't lose material in 30 plies. >>>>There are a lot of others, but the program doesn't mind, since he thinks they >>>>all are worth the same. >>>> >>>> José C. >>> >>> >>> >>>The fact that h3 was not losing material after 30 plies was not obvious, at >>>least to me. >> >> I didn't say the opposite. I simply didn't catch his joke, and tried to >>explain why the program gives a move like h3 when doing material only. >> >>>As someone else said, a tool like this one could be useful to improve opening >>>theory by computing exactly which opening lines could lead to forced loss of >>>material. >>> >>> >>> >>> Christophe >> >> I agree, but Rudolf said it only works in the starting position, not in >>midgame positions, so I think when you start moving pieces, the tree will get >>bigger and it will be impossible to perform such a deep search. >> Anyway, the test is interesting. No doubt. The question is 'how better (if >>any) is material only at d30 compared to a good eval at d15?'. Probably some >>people (Vincent?) would prefer good eval at d15. > >I think that everyone is going to prefer good evaluation at depth 15 >and not only material evaluation at depth 30. > >I will say more than it. >I believe that Sos with normal evaluation at depth X >is going to beat Sos with only material evaluation at depth 2X >for every X except maybe for X=1,2,3. > >1.h3 moves cannot work. > >Rodulf can try it for small X. > >Uri Before somebody corrects me I want to make clear that when I say for every X I mean for X that you can practically get. Of course when X is big enough the only material evaluation is going to solve chess but these X's are nbot relevant for practical purposes. 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.