Author: jonathon smith
Date: 09:20:01 02/07/00
Go up one level in this thread
On February 07, 2000 at 11:54:37, blass uri wrote: >On February 07, 2000 at 09:01:40, jonathon smith wrote: > >>On February 07, 2000 at 08:56:44, Robert Hyatt wrote: >> >>>On February 07, 2000 at 02:35:58, Bertil Eklund wrote: >>> >>>>On February 07, 2000 at 02:13:19, Will Singleton wrote: >>>> >>>>>Quite unusual, I guess Crafty was all worn out from its ICC tourney win. This >>>>>is a typical kside attack, I'm surprised Crafty allowed it. My first win >>>>>against Crafty. >>>> >>>>Hi! >>>> >>>>Yes, king-security is probably one of Craftys weaker sides. I have seen Crafty >>>>17.01 castle queen-side, right into an enemy attack in cases where a human or >>>>another computer don´t even consider the possibility. >>>> >>>>Bertil >>>> >>>>> >>> >>>If it does, something is broken. Crafty has a _lot_ of code to analyze the >>>king safety on _both_ sides of the board >> >> >>ChrisW told me that Crafty has not much code and what it does have demonstrates >>a total lack of understanding of king safety. >> >>ChrisW told me there are two types of chess programmers. Programmers. And chess >>players. ChrisW told me that despite manifest protestations to the contrary, >>"programmers" really have no idea. Which is why they bean-count. Like Crafty >>does. >> >>I believe ChrisW. > >I know that programmers talk with chess players. >I know that Amir ban talked with grandmasters. >Do Chris think that the grandmasters have no idea to explain him the problems in >the evaluation function of Junior? ChrisW says he have no idea about Mr Ban. He never tried to work him out. If he is a "typical chess programmer" (which is defined further on), then he and a GM will find it difficult to communicate because they are on very different levels, and few GM's have put their thoughts into words. Fewer still into words that are machine-programmable. Better is if the programmer himself is a strong player. The ideas can flow as pictures in the head, no need for word translation. > >I am sure that Amir knows about many problems in the evaluation function of >Junior(I told him about some problems that he did not fix). >I guess that the reason that he did not fix them is not that he has no idea but >because it is not easy to explain the computer everything that you know. ChrisW told me that since he didn't fix the under-promotion bug for some time (did he ever?), then it was always going to be unlikely that deeper issues were ever going to be resolved. > >It is also possible that he found that giving Junior more knowledge is going to >do it weaker in games because the speed of the search is also important. >I do not believe that you can learn from the fact that programs have no idea >that the programmers have no idea. > ChrisW told me you can learn from mnay small things. No self-respecting chess player could possibly allow his own silicon creation to play with such holes in it, if he knew what the holes were. > >I know that Robert hyatt also talked with grandmasters and that he is not a weak >player relative to most of the programmers. ChrisW says Dr Hyatt is a very weak player relative to chess players. >What is the minimal rating that a programmer need by chris's opinion to be a >chess player? ChrisW told me that the crucial boundary to pass is that of thinking that chess is tactics. ChrisW says that some people refer to the stage beyond this boundary as "playing positional". There are more boundaries beyond this one, but the typical chess programmer didn't even reach this first stage. The typical chess programmer found chess difficult, realised he wasn't going to progress as much as he liked, and turned to chess programming instead. The typical chess programmer sees himself as a silicon-human combination that plays chess. Hence the confusion you will often see when the programer talks of "we". "We" can mean the program, or the program-programmer combination. Serious confusion arises in the mind of the typical programmer when he begins to say "we" when he means "I". After "playing positional" is mastered (and it is *not* mastered by knowing some positional concepts - only by actually doing/thinking it), the next step is psychological - the idea of working out what your opponent is thinking. Translated into chess programming "what your opponent is thinking" reduces to "what the programmer is thinking" and "what certain types of programmer think". This comes under attack from programmers who never got to this stage. Hence you will find occasional posts from Mr Moreland to ChrisW telling him to "quit psychoanalysing me". Mr Moreland never understood what was going on and reached the wrong conclusions. >Is chris a chess player by his definition? Yes. > >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.