Author: blass uri
Date: 03:35:46 03/14/99
Go up one level in this thread
On March 14, 1999 at 02:05:59, Oliver Y. wrote: >On March 13, 1999 at 21:19:29, James T. Walker wrote: > >>On March 13, 1999 at 21:08:42, Marc Plum wrote: >> >>>(snip) >>>>>To give only one example: As far as I know GM Kasparov is not able to >>>>>win with a Queen against a Rook when you use Ken Thompson's endgame >>>>>tablebases. Of course the position should not be mate in one of course, >>>>>but let us take just a position with mate in 27, for example, which >>>>>is quite a common thing. >>>>>Well believe it or not, neither Kasparov nor Karpov can win. They are >>>>>helpless! >>>>> >>>> >>>>Have you asked Kasparov or Karpov this? I've never heard either of them admit >>>>that they can't win Q vs. R. It is a difficult ending, but if a GM like Walter >>>>Browne can win it (and he can) then I'm sure Karpov and Kasparov are up to the >>>>task. >>>> >>>> >>>>--Peter >>> >>>For that matter, there is at least one computer on ICC that allows players to >>>practice various kinds of theoretically won endings. The Bishop and Knight mate. >>>Queen vs. Knight, and Q vs R. Unrated of course. Comparatively slow blitz time >>>controls with increments. The computer is using tablebases, and kibitzes the >>>number of moves until mate on each move. >>> >>>I once watched an IM using this program to practice the Q vs R ending, which he >>>won twice in a row (maybe more, but it wasn't *that* interesting to watch!). >>>According to the kibitzing from the computer, he missed the fastest line once or >>>twice, but he still did it without serious danger from the 50 move rule. >>> >>>I'm sure I couldn't win in under 50 against optimum resistance (unless there was >>>already some obvious short range tactic), but for Kasparov, Karpov, Anand, >>>Kramnik, and many lesser GMs, I think it would be a piece of cake. >>> >>>Marc >> >>Hello Marc, >>I think it depends on the individual and how much practice he gets on this >>ending. I rember when Walter Browne tried this and could not do it. He studied >>the ending and tried again. The second time I rember it was "ify" as to whether >>or not he won the bet. It seems to me he "showed Progress" to conversion which >>would have taken place about move 50/51 and was given credit for the win. In >>any case I'm sure it was not easy after a lot of preparation. I'm not sure he >>could do it on demand today. The problem is around the 18th move to conversion >>for some reason. There seems to be a "wall" there. When they get past that >>area it's not so bad. I'll bet there are many GM's like Walter Browne that can >>not do it without a lot of prep/practice first. >>Jim Walker > >I beat Taimanov in an untimed simul where I had Q vs R, and he had about 5 >opponents left at that time, in 1987 when my rating was about 1987. > >I had never practiced that ending before, and I didn't seem to have too much of >a problem with it. He didn't blunder, by the way. Did you analyze the game with tablebases to see that he realy did not blunder. If you had mate in 26 and after Taimanov's move you suddenly had mate in 24 then it is obvious that he blundered. winning against humans KQ vs KR is easy because humans do a lot of blunders in this endgame. The problem for humans is to win computers in this endgame. Uri
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.