Author: Robert Hyatt
Date: 06:00:26 03/10/00
Go up one level in this thread
On March 10, 2000 at 08:59:18, Robert Hyatt wrote: >On March 10, 2000 at 04:57:42, Torstein Hall wrote: > >>On March 09, 2000 at 20:58:09, Robert Hyatt wrote: >> >>>On March 09, 2000 at 16:58:07, José Antônio Fabiano Mendes wrote: >>> >>>>[....]Three Studies >>>>Eleven studies had long since been prepared for testing >>>>the program---several years earlier I had written in an >>>>introduction to a collection of studies by G.Nadareishvili >>>>that it was with studies that one should begin experiments. >>>>My reasoning was simple---in studies there is forcing tactical >>>>play,positionnal evaluation is not needed,and since positional >>>>"understanding" was to be the last thing to be put into the program, >>>>one should therefore begin with studies... >>>>We began with a famous study by Réti. >>>>[D]7K/8/k1P5/7p/8/8/8/8 w - - >>>> White to play and draw >>>>What could be simpler,and at the same time cleverer than this >>>>composition?[....] >>>>And so,during December 1976 to January 1977,"Pioneer" solved >>>>Réti´s study.We thought that it would all be very simple,but >>>>it proved to be highly complicated.Without any positional >>>>evaluation,and without the attachment of the endgame library, >>>>the tree "disintegrated".The computer had little other work, >>>>but hours went by,and still no result.It became clear that >>>>"Pioneer" needed help! >>>>We took the rule of the square,programmed it in three modifications, >>>>put it into the library,and at each node of the tree "Pioneer" >>>>received from the library the necessary information.The effect >>>>was staggering:the study was solved within 70 minutes,and in the >>>>search tree there were only 54 moves.This small "human" tree was >>>>first obtained on 28th January 1977---without doubt a significant >>>>date in cybernetics.[...] >>>>Source:"Selected Games 1967-1970",M.M.Botvinnik,Pergamon Press,1981, >>>>pages 299 and 300 >>>>So it seems that,contrary to widespread belief,Botvinnik thought his >>>>computer-chess work to be very important. JAFM >>> >>> >>>He thought it very important... most of the rest of us didn't... as it >>>seems that most of the results were faked... >> >>What do base your conclusion that the relults where faked on? >> >>Torstein > > >Some PVs he published in the JICCA that were obviously 'faked'... he would >make a comment that pioneer decided "this variation can be stopped as obviously >white is winning" but in another almost identical position, pioneer would keep >searching. Hans Berliner had a very good (if very abrasive) rebuttal article >that he published in the JICCA in response to Berliner's article. His analysis >was concise, accurate, and devastating to Berliner's statements... That last "Berliner" should have been Botvinnik of course...
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.