Author: Bruce Moreland
Date: 09:45:03 03/19/99
Go up one level in this thread
On March 19, 1999 at 02:13:35, JW de Kort wrote: >Old ICCA Journals are very hard to get by. As far as i know only two libraries >in teh Netherlands have copies of this journal and they are in an other part of >the country! So could you be more specific about Mr. Berliner critism. > >AND: in some old posting to CCC a few months ago i think somebody mentioned that >he was going to de some rechearch into Botwinniks ideas. Maybe this person could >tell something more about implementing Botwinniks ideas! > >Thanks in advance The Berliner article is the Sep '93 issue of the ICCAJ, volume 16 number 3. I also mentioned another article, where I claimed that Botvinnik's son had asked for money to continue the project. I was wrong, it was his grandson, who hints fairly strongly that he'd like to do the work somewhere other than in Moscow. This article was on page 255 of the Dec '95 ICCAJ, volume 18 number 4. Back to the Berliner article. It is extremely harsh with Botvinnik. He claims that Pioneer was probably not even a D player, if it could play at all, despite Botvinnik's claims. He calls into question some of the search trees that Botvinnik published. I've seen the trees he is referring to, they don't look like normal search trees, they are not bushy at all, and they leave a lot unsaid about what is going on. The output of the program is also presented, and it is extremely concise. Sub-variations are presented that seem to demonstrate massive understanding. Berliner is disturbed that Botvinnik publishes trees that contain a mixture of egregious mistakes and positions whose evaluation and subsequent pruning are given no comments, even though the position requires great insight to understand. Here is an example, from Botvinnik's article in the Jun '93 ICCAJ, p 72. 5rk1/5ppp/p1Q1p3/1R6/q7/4b1P1/P2RPP1P/6K1 w - - 0 1 This is from Kasparov-Ribli, 1989. Botvinnik gives the variation 1. Rd8 Qxb5 2. Qd6 Bxf2+ 3. Kxf2 Re8 4. Qe7. Berliner finds three problems with this 7-ply main-line. First, 3. ... Re8 is a strange move for black, why not do 3. ... Qf5+ instead? It's an easy move for a computer to see, and it appears to draw. It doesn't draw, but the winning line is supposedly 31 plies. Second, in response to 3. ... Re8, there is a winning line that is easy to see for a computer, and that is 4. a4. Third, is that after 4. Qe7, *black* can win instantly with 4. ... Qb6+. This is supposedly the culmination of 20 years work by between three and ten people, plus Botvinnik (ICCAJ Jun '93, p. 71). I think that Berliner is right, that there was something extremely fishy about Botvinnik's results, and I think his article was restrained, actually. bruce
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.