Author: Roy Eassa
Date: 14:11:57 05/23/02
Go up one level in this thread
On May 23, 2002 at 17:05:46, Robert Hyatt wrote: >On May 23, 2002 at 16:59:02, Roy Eassa wrote: > >>On May 23, 2002 at 16:34:04, Rolf Tueschen wrote: >> >>>On May 23, 2002 at 14:20:09, Roy Eassa wrote: >>> >>>>On May 23, 2002 at 13:41:05, Rolf Tueschen wrote: >>>> >>>>>Kramnik wants FRITZ (sic!) 6 months in advance and some more rules. The FRITZ >>>>>even Eduard Nemeth with Elo 2100 can beat almost at will? Strange. >>>>> >>>> >>>> >>>>I thought Eduard was quite a bit stronger than 2100. Also, I'm pretty sure he >>>>canNOT beat Fritz 7 "almost at will". He can win more games against it than >>>>might be predicted by his rating, and far more than most humans can, but I think >>>>he would be the first to tell you that he loses a lot of games against it too. >>> >>> >>>Roy, that's true. Two points. >>> >>>1. Do you believe that Kasparov would lose a game against Fritz at all? I mean >>>not in a match or in public for some PR, there even POCKET Fritz drew with Leko. >>>I mean in real. And no Blitz or fast controls. Although Eddie wins 5, 10, 20, 30 >>>minutes games. I have seen about 100 games in total. >>> >> >> >>I believe Fritz 7b running on a fast PC can indeed win *some* 40/2 games against >>Kasparov, unless Kasparov always plays for a draw. I believe Kasparov would win >>a lot more games than Fritz would, if the match were of any significant length. >> >> > > >I believe Frans would be happy to win one of every 10 games at 40/2hr. Of >course it would take a lot of money to get Kasparov to invest that kind of >effort into a bunch of games as well... > > > >>>2. If Fritz and Junior are presented as 2700 machines a little hot air might be >>>allowed in my position too, only into the opposite direction to inspire >>>thinking. It's not so funny if here except Dr. Hyatt nobody showed interest in >>>my thought experiment with the 5 human GM in a group who trained on the "new" >>>chess directly designed for computers. In that case only I believe the computers >>>fall down to 2200-2350. Instead everybody is happy to declare that I must be >>>wrong with such numbers. BTW I must read the old posts you have written about >>>the topic before my time. The offline readers are not yet very comfortable, >>>alas. >>> >>>Rolf Tueschen >> >> >>My estimates were around 2550, which is halfway between your 2350 number and the >>2750 some claim. > > >That is in line with my estimate for computers as well... assuming that the >opponents play the computer rather than the board. > > >> >>But it's not my estimates that matter, it's the concept of uncertainty. IMHO, >>nobody can be really certain of the true strength of these machines yet. Only >>when GMs have spent a few years with them as frequent opponents can we start to >>know the computers' real strength. >> >>I think most GMs over 30 or 35 may already be too old to "re-wire" their neurons >>to the different approaches required (e.g., avoiding all possible tactical >>inaccuracies at all costs, even if it means forgoing certain promising >>strategies) against computers as opposed to the approaches they've spent years >>and years refining against humans. > >I disagree there. Roman has "rewired" and wrecks computers right and left >at blitz time controls on ICC. > >Others do it to. They are slowly adapting... I know about Roman. That's why I said "most". He's one of the exceptions. But I can reword: all other things being equal, the younger the GM the more likely he/she is to be successful against computers (mainly because the younger ones have spent a greater percentage of their years in the presence of such strong machines, but also because most people learn and adapt to change more readily when they're younger).
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.