Author: Uri Blass
Date: 03:28:34 12/17/00
Go up one level in this thread
On December 17, 2000 at 04:17:54, Christophe Theron wrote: >On December 16, 2000 at 21:27:59, stuart taylor wrote: > >>On December 14, 2000 at 20:43:46, Christophe Theron wrote: >> >>>On December 13, 2000 at 20:11:01, stuart taylor wrote: >>> >>>>On December 13, 2000 at 12:53:11, Tom Glenn wrote: >>>> >>>>>Here's a fun game . My old (10-12 yrs?) Fidelity Elite Avant Garde , which >>>>>used the Mach III engine vs Gambit Tiger running on 1.0 GHz thunderbird chip >>>>>EAG got 30 mins to make all moves , GT got 60 secs....... >>>>>A lot of carnage in this opening line , settles into a fairly even ?? game >>>>>by mid 30's , then GT crushes him in next 10-12 moves...... enjoy >>>>>Sorry about the spacings...... still trying to figure all this out .... >>>> >>>>This result is obvious to me. In fact I'm surprised it took 57 moves. >>>>I think that GT would win every single game in 1 min. even against that same >>>>opponent taking tournament timing over his. >>>> >>>>But we know all that! The problem is that nothing is tearing the top programs of >>>>the last 3-4 years to pieces in any way at all. >>>> I don't even know if there has been any tangible improvement at all, during >>>>last 3 years, besides the hardware. >>>>S.Taylor >>> >>> >>> >>>Of course there has been. I even wonder how you can ask this question... >>> >>> >>> >>> Christophe >> >>By tangible improvement, I mean that in any game, you can see that the slightly >>older program just disintegrates, and can never really get anything good going, >>but that it gets overpowered, unless very lucky. >> >>S.Taylor > > > >To be 70 elo points better than your opponent, you have to beat him by 60%. > >That means that in 40% of the games you will... loose. But still you will be 70 >elo points stronger. Because you win the other 60%. > >I think 70 elo is a tangible improvement. But you will not see this improvement >in ANY game. In 40% of the games you will actually get the opposite impression. Not exactly. 1)There are draws in chess so and if you assume 30% draws you will see the improvement only in 45% of the games if the result is 60%. 2)70 elo difference suggest by the linear formula less than 60% The exact number is 470/800=58.75% 3)I think that the improvent in software is more than 70 elo. Fritz5 was the best program 3 years ago. The ssdf list suggest that fritz6a is more than 70 elo better than Fritz5.32 when Fritz6a is not the best because Deep Fritz is probably one of the programs that is better. I guess that the improvement in the last 3 years is more than 100 elo(my estimate is 150 elo improvement and it suggest getting 68.75% by the linear formula). 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.