Author: Amir Ban
Date: 16:24:23 11/17/00
Go up one level in this thread
On November 17, 2000 at 13:39:36, Uri Blass wrote: >On November 17, 2000 at 13:02:04, Amir Ban wrote: > >>On November 16, 2000 at 15:19:38, Bob Durrett wrote: >> >>>On November 16, 2000 at 12:38:05, Uri Blass wrote: >>> >>>>On November 16, 2000 at 12:01:56, Bob Durrett wrote: >>>> >>>>>On November 16, 2000 at 11:50:29, Bob Durrett wrote: >>>>> >>>>><snip> >>>>> >>>>>>source code] dedicated to position evaluation, assuming one can find out this >>>>>>information. For example, a program with several lines of code devoted to >>>>>>position evaluation might be [perhaps arbitrarily] declared to be "knowledge >>>>>>based" whereas one with a few hundred or less might be called "bean counter." >>>>>>The problem here is that the source code may not be available. You would have >>>>>>to ask the programmer. He/she might be willing to answer that question even if >>>>>>unwilling to reveal the code itself. >>>>> >>>>><snip> >>>>> >>>>>Sorry about that. >>>>> >>>>>I meant to say "several thousand lines of code" yeilds "knowledge based" >>>> >>>>I disagree with the method of counting the number of lines of the source code. >>>> >>>>It is possible that one programmer knows to write the same evaluation by less >>>>lines of codes so it is possible that 2 programs with exactly the same knowledge >>>>will have different number of lines of code. >>>> >>>>Uri >>> >>>Yes, that makes sense. >>> >>>We are faced with the problem of trying to make a determination based on >>>inferences rather than direct evidence. >>> >>>One could simply ask the programmer what his/her intent was. If the answer were >>>to be "I intended for the program to be knowledge based," then we might be >>>tempted to declare the program to be "knowledge based." Similarly if the intent >>>were "bean counter." >>> >>>But, the rest of the chess-playing program programmers might say that the >>>programmer failed in his/her intent. >>> >>>There is also the closely related problem that there does not seem to be any >>>universally accepted precise definitions of "knowledge based" and "bean >>>counter." >>> >>>As for the importance of the issue: >>> >>>If a new programmer decides to make a knowledge based program, maybe simply >>>asking for advice from experienced chess-playing program programmers would >>>suffice. >> >>You are missing the point that there do not really exist two separate >>approaches. The term "bean counter" (as used in this thread, it originally meant >>something else) is derogratory. The term "knowledge based" is mostly hype. >> >>Amir > > >I think that it is possible to divide programs by the question what they >evaluate. > >Example: >Programs that evaluate unstoppable passed pawn and program that do not evaluate >it. >It is easy to chack it >[D]8/ppp5/k5K1/8/7P/8/8/8 w - - 0 1 > >If programs evaluate this position as negative for white at depth 1 then it is >clear that they do not evaluate unstoppable passed pawns > >I do not know about the new version of Junior but the old version of Junior >evaluates this position as negative for white. >It is not important in order to play the position correctly but it may be >important if this position is deep in the search. > >This is only one example. > >You can ask many similiar questions and check the evaluations of programs at >depth 1 to see which program evaluates more things(I assume that there are not a >lot of extensions at depth 1 so programs do not see it by tactics). > >I do not say that programs that evaluate more things know more because it is >also possible to know the right numbers in the evaluation and it is also >possible to know what to evaluate but it is possible to get better evaluation by >evaluating more things. > >If the design decision of the programmer is to evaluate more things then I think >that it is correct to say that he meant to develop a knowledge based program. > >Uri If your criterion of knowledge is based on accuracy of evaluation then I respectfully apply for membership in the exclusive "knowledge based" club (and IMO some members don't belong there). BTW, accuracy of evaluation is the best criterion for being knowledgable that I'm aware of. I've posted here in the past that, to start with, we don't have a real definition of what good evaluation means. This is the focus of my work with Junior for more than a year. Amir
This page took 0.01 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.