Author: Darrel Briley
Date: 19:28:00 08/28/05
Go up one level in this thread
On August 28, 2005 at 20:01:12, David Mitchell wrote:
>On August 28, 2005 at 13:42:40, kaqs.1662@bumpymail.com wrote:
>
>>On August 28, 2005 at 12:44:29, William Kerr wrote:
>>
>>>Where are the versions of MacHack on the web?
>>
>>I stumbled across them last night. I would have mentioned it in the previous
>>mesage, except I just sort of assumed that it was already common knowledge and
>>that it was just new to me.
>>
>
>I'm sorry I can't help you "resurrect" MacHack. I'm amazed that 1). You seem to
>have found it, and 2) You thought anyone else would have surfed into some dusty
>tape storage archive of PDP computers. :)
>
>Has Allan Turing's "hand" chess algorithm ever been set to play on a modern
>computer? That would be interesting. Also on the list of interesting historical
>programs would be Kaissa, the world champ of so many years ago.
>
>Dave
Engines:
Turing Chess engine for Fritz by Mathias Feist and Ken Thompson which implements
the famous paper machine by Alan Turing. The orginal set of rules was incomplete
and unclear in several points, which are discussed below.
1. no mention was made of stalemate; detection was added with a positional
evaluation of 0.0, which seemed to be the most logical.
2. a move gets a bonus if threatening mate in 1; a threat obviously means a
nullmove for the other side which is illegal if the move was giving check. Still
it may be a mate in 1 move. This was ignored, the move just gets the bonus for a
checking move.
3. algorithm enhanced so that the engine may play with both sides, i.e.
everything logically negated if black to move.
4. iterative deepening added to allow higher search depths.
5. a recapture is not made entirely clear: it is considered to be a capture to
the same square as the previous move.
6. rule d) is not clear about checks/enemy pieces; a wKe1 and a wPe4 give the
vulnerability of e2+e3, but a bPe4 gives an additional vulnerability of e4.
Doesn't sound logical.
7. calculating the material value is impossible if one side is left with the
king only. A division by zero would occur. In this case the material value is
material+1.0 of the other side.
The algorithm is based on material value, at the root moves with equal material
values are resolved by positional values which are shown as the evaluation. This
is not completely satisfying since the positional values are rather "jerky" and
the more important material values are hidden.
You download a setup program which allows you to install the engine in the Fritz
directory.
Download Turing Engine
http://www.chessbase.com/download/engines/setup%20turing.exe
DB
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.