Computer Chess Club Archives




Subject: About Proofs and Wishful Thinking

Author: Rolf Tueschen

Date: 02:00:38 05/30/05

Go up one level in this thread

On May 30, 2005 at 01:26:15, Alex Schmidt wrote:

>>If someone accuses an engine to be a clone, he must collect A LOT of PROOFS
>>before exposing his feeling to the world. Asking an author to proove his
>The proofs shown would be enough for every court.


>We have the same moves in _all_ tested positions where several moves lead to a
>mate or a draw. (~2% of the UCI engines did the same)
>We have a very special feature which is known only from Crafty, no one of the
>programmers here said "I have the same feature in my engine", no one even told
>what the feature is used for. (I would guess less than 1% have this)

What a nonsense! Indeed you turn logic upside down. Because no one "here said"
something, it is proven that it doesn't exist? A claim like that would fail in
every court for sure.

>We have an opening book which works also with older versions of Crafty. (I would
>guess less than 1% of the engines would use the same book format coincidentally)

Also of strange logic because in Crafty itself - as far as I've understood it -
"new" books dont work in older versions. So what should it prove if a specific
book works with all Crafty versions? This isn't sound what you are concluding or

>We have the same behaviour if kings are missing in the fen string, Crafty thinks
>there are kings on e1 and e8. (~10% do this)

NB that the higher such percents the more unlikely youre right with your theory.

>We have 2 strings which are _very_ seldom, after searching many engines we found
>only 1 engine where one of the strings is used. (let's say 5% for each string)

How many percent of engines you could (re)search?

>We have the same behaviour in most illegal positions. (less than 5 %)
>The probability that an engine have the same behaviour in all of the above cases
>is 0,0000000005% (one of 2 Billion), and it's calculated in order to be on the
>safe side.

From all the above I know that this final blow up is wrong, absolutely wrong.
You simply cant "calculate" certainties out of many uncertainties. I would also
say that you can have reasonable doubts on the base of the above, but you can't
present the case as if you had "calculated" the "proof" of your suspicion.

>I think this is enough to ask the programmer, especially if he wants to go
>commercial. It's up to him to show the source or not, but as long as he even
>didn't try to explain one of the similaries this are enough proofs for me.

Sad enough. It's simply wrong.


This page took 0.02 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.