Author: Albert Silver
Date: 13:32:12 07/14/00
Go up one level in this thread
On July 14, 2000 at 16:13:29, blass uri wrote:
>On July 14, 2000 at 15:51:31, Albert Silver wrote:
>
>>On July 13, 2000 at 17:05:46, Christophe Theron wrote:
>>
>>>On July 13, 2000 at 08:53:22, Jari Huikari wrote:
>>>
>>>>On July 12, 2000 at 14:02:16, Christophe Theron wrote:
>>>>
>>>>>Well you are right, but it is a proven fact that a K6-2 450MHz has much more
>>>>>chess knowledge than a 486dx2-66MHz has.
>>>>
>>>>>If you don't like the above idea and don't want to consider CPU speed as being
>>>>>part of the "knowledge", then your remark does not refute my definition...
>>>>
>>>>A good common definition for knowledge is hard to find. CPU speed is not
>>>>programmed into a chess program, which the original question was about.
>>>>
>>>>Knowledge IMHO is some piece of information somehow included into code.
>>>>Program knows something is good without having to do a deep search to
>>>>find it out.
>>>> Jari
>>>
>>>
>>>What you are saying here is that knowledge can only be found in the evaluation
>>>function. I think this is totally wrong. A lot of chess knowledge is in the
>>>search itself.
>>>
>>>
>>> Christophe
>>
>>Sort of. I agree that many decisions are determined by the search, and often, as
>>a player, I will work by process of elmination in order to play a move. In other
>>words (this is common practice) I will analyze moves and discard them until I
>>hit on the one that does the job. I may be able to then justify it by positional
>>rationalizations, but essentially, the decision came out of pure calculation.
>>While there is no question that the search (calculating) is a huge part of the
>>decision making process, I wouldn't call this knowledge.
>>
>
>I think that you can learn about the evaluation based on the search and you can
>learn from your search to do the positional rationalizations for future
>positions.
>
I agree, but here I differentiate my understanding of a position with my chess
knowledge. For example, I look at a given positions and see that there are a
number of important positional considerations. These considerations are all
there in no particular order, and only exist because of my chess knowledge. As I
begin to calculate, the importance of these factors begins to order itself, and
the more I calculate, the more I refine it, until it clearly defines the move I
choose. Perhaps I see a pair of doubled pawns, a backward pawn, and an open
file. All look important, but as search, I realize that I must attend to the
open file first because if I waste time playing around the pawn weaknesses, I
will be overrun by my opponent on the open file in the long run. I may not
calculate all the way into this imminent invasion, but just enough to see this
is clearly so, thus the search may help in my understanding of the position, but
not necessarily in my understanding of chess in general. Though, as I write
this, I can see this isn't absolutely true unless the person walks through life
(and chess) with their eyes closed. All the same, you probably understand what I
mean.
Albert Silver
>I think it is knowledge and the most important knowledge that humans have in
>chess and computer programs usually do not have.
>
>programs usually use a static evaluation function and do not learn to change
>their evaluation function based on previous search.
>
>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.