Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How did Alen Turning's program work?

Author: Robert Pope

Date: 09:17:51 04/21/00

Go up one level in this thread


On April 21, 2000 at 07:43:51, Alessandro Scotti wrote:

>On April 20, 2000 at 19:51:02, Pete Galati wrote:
>
>>I've seen Alen Turning's Chess program that only existed on paper mentioned a
>>few times, but I don't think I've seen any mention of how it actually worked.
>>
>>Has his program ever been published?
>
>I'm very interested in this subject of reviving old programs. After a little
>search I was able to dig out the original papers by Kotok and Bloom (in the form
>of bitmap scanned pages!) but it seems I'm missing a couple of additional pages
>which contained moves from actual games, a very useful thing for testing. If
>there's enough information available it could be possible to write a program
>that plays (more or less) the same even if the original code is not available.
>
>Sorry if I haven't answered your question... I'm looking for info too!

The Turing's article in "Faster than Thought" is also reprinted in David Levy's
"Computer Chess Compendium".  I was planning to recode some of the early
evaluation functions as options in my own program, but I haven't had a chance to
do any work on it recently.  It is similar to other current bare-bones programs,
though his selection of moves to look at in quiescence is a little unique.

Here is a quick synopsis:
P=1, N=3, B=3.5, R=5, Q=10, checkmate=1000

Search: 2 ply exhaustive, then quiescence.

Quiescence:
All recaptures, all en prise captures, all checkmating moves, all captures where
value(attacker)<value(captured piece).

Evaluation=W/B, white's piece totals over black's.

If there is a tie, look at Position-play value for white pieces and black king,
consisting of:
a) +sqrt(true legal mobility of each N,R,B,Q), capture moves count double
b) +1.0 for non-queens that are defended, +1.5 if twice-defended
c) +sqrt(t.l.m. of king, excluding castling)
d) -sqrt(mobility of queen if it were on king's square)
e) +1.0 if moves do not destroy castling possibility
   +1.0 if castling is possible on next move
   +1.0 if castle this move
f) +0.2 per rank advanced per pawn
g) +0.3 per pawn defended by a non-pawn.
h) +1.0 for threatening checkmate of black king (??)
i) +0.5 for checking black king

Square roots are rounded to nearest tenth.

Hopefully this is all right -- I am typing by hand and interpreting what he
says, which is sometimes ambiguous.

Rob



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.