Author: Eugene Nalimov
Date: 10:08:58 01/30/02
Go up one level in this thread
I don't have journal with detailed description of KAISSA with me in the USA
("Problemy Kibernetiki", 1974, 29, pp. 169-200). But from the memory it's
more-or-less "standard" brute force.
In the addition, KAISSA is several times mentioned in the 'Games Programming'
(in Russian, 1978). Yes, they described several methods of forward pruning (e.g.
something like futility pruning, attempts to avoid Q-search by reusing
information from the Q-search 2 plies ago, etc.), but they use those methods
only in the last 2 plies. Closer to the roots they are full width. They also
wrote "Two strongest today's programs -- Chess 4 and KAISSA [...] avoid pruning
methods that can change tree search results".
In 1974 KAISSA run on ICL 4/70. In 1977 on IBM 370/168 (Blitz 5 was running on
Sigma 9 the same year). Very fast computers for the time.
Eugene
On January 30, 2002 at 10:36:12, Robert Hyatt wrote:
>On January 30, 2002 at 00:55:28, Eugene Nalimov wrote:
>
>>Kaissa was a brute force program. I have a book 'Games Programming' (in Russian)
>>written by Kaissa authors and published in ~1977.
>
>
>I don't believe it was a brute force program by any reasonable definition.
>IE look up their "method of analogies" in how they did forward pruning. That
>doesn't sound like today's "brute force" idea..
>
>They had a clever algorithm that could be used to forward prune based on the
>idea that if in position X, a tactical flaw existed, then in many derivatives
>of position X, the same flaw existed because non-useful moves had been tried
>by the search...
>
>They discussed this at the 1977 event in Toronto (which they didn't win, btw).
>
>
>
>
>
>>
>>Eugene
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.