Author: Ricardo Gibert
Date: 11:03:12 08/15/03
Go up one level in this thread
On August 15, 2003 at 13:41:30, Dann Corbit wrote: >On August 15, 2003 at 00:18:49, Ricardo Gibert wrote: > >>On August 14, 2003 at 23:54:08, Dann Corbit wrote: >> >>>On August 14, 2003 at 22:40:04, Ricardo Gibert wrote: >>> >>>>If doubling NPS (all else being equal) produces a roughly 40 to 75 elo strength >>>>increase, then doubling knowledge produces how much elo strength increase? >>> >>>How do you measure doubling of knowledge? I doubt if anyone can measure it. >> >> >>Look up information theory on the web. > >What will that do for me? Probably nothing. > >> Also, you might find it helpful to >>consider EGTBs a model for all of chess. > >Every study I have seen shows that EGTB's do not make chess programs play any >stronger. By that measure, information has no value. So if hypothetically we were to actually have 32-man EGTBs, you would expect the program using them to play equally strong to a program not using any EGTBs? It helps to know the reasons why EGTBs do not significantly add strength to programs. Applying empirical results without understanding the reason for the result will inevitably lead you astray. > >>>>Or do increases in knowledge follow a completely different type of relationship >>>>to playing stength? >>>> >>>>What impact on NPS does a doubling of knowledge have? I know it slows it down, >>>>but what is the relationship (a rule of thumb) between the 2? It is said that >>>>adding knowldege in hardware does not exact a speed penalty, but can this really >>>>be true? Or is the speed penalty simply less apparent? >>> >>>It may be possible to double knowledge and NPS at the same time. >> >> >>By knowledge I meant chess knowledge as it pertains to evaluation. Another type >>of chess knowledge is that pertaining to move ordering. Also, non-chess >>knowledge to increase the efficiency of a program is another type knowledge and >>the one the one you appear to allude to. All of these I think they need to be >>considered separately. However, the one I'm most interested in is the knowledge >>used in evaluation. > >Yet another type of knowledge is search depth. So I don't think it is possible >to separate the two cleanly. I'm actually very interested in not separating them cleanly. I'd like to know the relationship between search depth knowledge and eval knowledge. It is useful to know which one to try to emphasize. > >> >>> >>>>How much attention should a programmer give to lowering the EBF of his program >>>>versus adding more knowledge? >>> >>>Just pick a single move. EBF is one. In other words, any reduction in EBF >>>must be smart or it may do more harm than good. >> >> >>Which is one of the reasons why I asked, "How much attention should a programmer >>give to lowering the EBF of his program versus adding more knowledge?" I think >>it is possible to answer my question a little bit intelligently ;-) > >That would require a more intelligent person than myself. > >>>>I'm curious about the different opinions programmers have on this topic. My >>>>assumption is the answers to these questions can have a big impact on computer >>>>chess program design.
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.