Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What's Fritz's IQ?

Author: Peter Berger

Date: 14:54:07 12/30/01

Go up one level in this thread


On December 29, 2001 at 14:57:29, Miguel A. Ballicora wrote:


>First, let me rephrase my english since I introduce a double negation:
>
>"NOT even amateurs are taught to think like a tree nowadays, and the
>best book about it is "How to reasess your Chess" by J. Silman."
>
>So, the influence is less and less to think like a tree. Still, there is
>tree component in the teaching but is limited to purely combinatorial
>situations.
>
>>I have read "How to reasess your Chess" and my impression was that the main
>>procedure Silman explains that is different to Kotov is how to get the candidate
>>moves and how to evaluate the resulting positions ( imbalances etc).
>
>I think is more than that.
>All the analysis with the imbalances is a non-tree approach and that is the core
>of the book. After that, you come up with a very good plan that will be tried in
>an analytic fashion (extremely reduced tree). Still, there is an emphasis in
>when NOT to calculate. See for instance part III (calculations)
>page 37 that the first two examples of this chapter is how NOT to calculate!!
>"... To calculate any variation at all would be a complete waste of time".
>Diagrams 28 and 29. Then, it goes to analyze a complex analytical situation
>but see what is says at the end:
>"Old experienced dogs like myself, will often avoid calculation even in this
>type of complicated situation and just play Rc5 cold turkey (of course, we would
>still use the thinking technique to find the correct plan! [MY NOTE: which is a
>non-tree approach])"
>The he explain why he would do that an adds
>"...Normally, you may want to look a couple of moves ahead to make sure that
>everything is in order..."
>
>The point is, sometimes you need to calculate, sometimes you need NOT to.
>Silman quoted a funny remark by R. Reti. Whe he asked how many moves ahead
>a GM usually calculates, he replied "one move". This is an exageration but
>sometimes it is true.
>
>>Or to put it in different words: I thought his work was mainly about reduzing
>>the tree size and getting more selective.
>
>Kotov's is supposed to be very selective!
>I think is the opposite, set your mind to a "non-tree thinking way" to see the
>truth of the position and later, if needed, calculate as in Kotov's.
>See that this book if for beginners, Nunn's and Tisdall's particularly, present
>a more elaborate way of calculation. Even in wild positions they are far from
>Kotov's "selective alpha-beta" approach.
>
>>I remember the first position in the book with the Qa7 move. It has been my
>>impression that it should be possible to teach a computer to think exactly like
>>this.
>
>What position is that? I cannot find it.
>
>Regards,
>Miguel
>

I am talking about the very first position in the book ; now I think it was
something like Qc2-a4-a7 but it should be easy to find as it is the first one.

My problem is I have borrowed the book to a friend a few years ago and never got
it back. I have saved the rest of your message to think about it more thoroughly
later.

Thanks and kind regards,
pete



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.