Author: Christophe Theron
Date: 18:58:43 11/24/00
Go up one level in this thread
On November 24, 2000 at 07:46:10, Fernando Villegas wrote:
>On November 24, 2000 at 00:47:44, Christophe Theron wrote:
>
>>On November 23, 2000 at 15:31:19, Fernando Villegas wrote:
>>
>>>I perceive a note of anger in your post that is missplaced. Maybe understable,
>>>but missplaced. I am not and I was not shooting at you, but just wondering about
>>>how testing is done. Never I have tried to damage your program, in case you
>>>thought it was so. It is just that I was surprised to find out that lack of
>>>basic knowledege. No matter if something happens 0,001% of the time,
>>>conceptually speaking IS esential.
>>
>>
>>
>>Can you please mention something that is not essential in your sense of the
>>word?
>>
>>I happen to have a very different point of view about this issue, it seems.
>>
>>
>
>Something can be considered esential when it is on the base of something. To
>mate is esential to chess and so, to mate with certain known patterns is
>esential too. Then there are the so called "important" things. A matter of
>degree and place in the evolution of things. Once that basic thing is got, maybe
>it does not appear anymore as a problem in itself, but still is esential.
>Euiclides geo is esential for mathematicians even if they never more in his
>lives has to do with euclides propositions. And so and so.
Very theoric words, but you did not answer my question.
What is not essential in a chess program, in your sense of the word?
>>> Of course you have your reasons and you have
>>>given them to me and that's fine; that is precisely what I wanted to get since
>>>the first post: to know why. Firt I got the "whys" of testers, now yours. And
>>>certainly it is not me who decide if B+N is important. No ferdinan. It is an
>>>issue that has been always considered important in any fair ending book. In the
>>>classic book "Chess Endings, Esential knowledege", by Averbach, B+N kind of
>>>endings gets pages from 11 to 14. So it does other kinds of endings.So it does
>>>in any kind of bchess book. Then clearly is not my decision, but a common sense
>>>decision since ever, and fully documented.
>>
>>
>>
>>So what?
>>
>>The problem is not to know if it is documented or not, the problem is to know if
>>it is important or not.
>
>
>I did not say that it is important because documented, but it is important AND
>documented. In a book of endings I suppose to document these mates is considered
>important beceause it is.
But the guys who write books have no idea how to write a chess program. I don't
see why I should rely about their judgement of what is essential and what is
not.
The only essential thing about a chess program is that it must follow all the
rules of chess. At least all the rules it will have to obey to.
The rest is a matter of playing strength, a matter of priorities.
>>You are surprised by the fact it is not important. I confirm: it is not
>>important at all.
>>
>>
>No. I am not surprised by that. Surprise me a lot more your stubborness in this
>matter. I can understand that becauise of statiscal reason you put that thing at
>the end of the list, but then that does not give you ground to decree that is
>not important.
It is less important than other bits of knowledge I had to implement first. It
is not important for me because there are more important things in my opinion.
You are the one to decree that it is "essential", on the ground that it is
considered important in books.
I don't really care what's written in books. If I like what I find in a book, I
do it. If I don't like it I ignore it. If I like it but consider it is less
important, I do it later.
What are you trying to tell me? That every guideline that is written in a chess
book must absolutely be implemented in any decent chess program because it is
"essential"?
By this definition there are no decent chess program.
>>>Respect your priorities, you are, of course, the best judge to sort them out
>>>according your goals, but respect your conception of esentials I cannot agree:
>>>if everything is esential, as you say, then nothing is. Just simple cartesian
>>>logic. Finally, Christophe, nobody is laughing. At least in my case you could
>>>just see in me a face of wonder, but not even a smile.
>>>In any post you answer about Tiger, you say that comments are welcome. I hope
>>>mine are, also, welcome.
>>
>>
>>
>>Your initial post was a critic against the beta testers.
>>
>>Unfair critisism I would say.
>>
>
>Why unfair? Only good critics are fair?
I consider they have done a good job. Their job is not to test all the endgame
configurations and to tell me which are the one which are not programmed.
I don't care about this.
What I want is that the beta testers put the program in the conditions of real
games and tell me which real games the program would blunder.
That's exactly what they did, and it has been useful and has taken all their
time. It allowed, for example, to find flaws in my KBPP/K knowledge. This ending
appeared several times in real games, and the first beta version of Tiger 13
blundered in these games.
Is this ending covered in your books? I would bet it is covered less often than
KBN/K.
The KBN/K ending never appeared in any of their games. Which is a proof that it
is a RARE case (no matter what books tell you).
That's the way I work. There is already an INFINITE amount of work required to
cover the cases that appear the most often in real games. Why should I look for
the cases that happen less often and work them out first?
On the ground that they are considered essential in books? You must be kidding!
So I consider the beta testers did exactly what they had to do, and in this
regard your comments were unfair.
You should have asked first how the beta testers were supposed to work, before
saying that they did a poor job.
Christophe
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.