Author: Mark Young
Date: 14:11:26 06/25/01
Go up one level in this thread
On June 25, 2001 at 17:01:56, Albert Silver wrote: >On June 25, 2001 at 16:41:58, Mark Young wrote: > >>On June 25, 2001 at 16:28:12, Albert Silver wrote: >> >>>On June 25, 2001 at 15:06:45, Mark Young wrote: >>> >>>>On June 25, 2001 at 14:05:58, Albert Silver wrote: >>>> >>>>>On June 24, 2001 at 15:24:53, Chris Carson wrote: >>>>> >>>>>>On June 24, 2001 at 15:02:54, Mark Young wrote: >>>>>> >>>>>>>On June 24, 2001 at 14:53:04, Chris Carson wrote: >>>>>>> >>>>>>>>On June 24, 2001 at 14:27:12, Sandro Necchi wrote: >>>>>>>> >>>>>>>>>Ok, but the computers are very good against chess players up to 2500 Elo, so >>>>>>>>>this may give the impression they play about 2600 as an average. >>>>>>>>>Against stronger opponents (2650 and above) they would score less. >>>>>>>> >>>>>>>>Deep Junior scored 2702 against opponents with average rating of 2702 at >>>>>>>>Dortmund 200. So they can score more. :) >>>>>>> >>>>>>>I find it funny people will give opinions, and that is fine, but I sometimes >>>>>>>wonder what the opinion is based on... can't be the computer's game record. :) >>>>>>> >>>>>>>> >>>>>>>>Best Regards, >>>>>>>>Chris Carson >>>>>> >>>>>>Mark, >>>>>> >>>>>>You will find this very funny. Here are the results for all programs and all >>>>>>games 40/2 against average opponents 2648 and above. :) >>>>>> >>>>>>Prog Avg Opp Avg Total Lowest Opp Highest Opp >>>>>>ELO FIDE ELO Games Fide FIDE >>>>>>2632 2645 30 2566 2795 >>>>>>2653 2671 22 2600 2795 >>>>>>2650 2730 10 2650 2795 >>>>>> >>>>>>Plus the DJ Dortmund 2000 performance 2702 vs 2702 FIDE over 9 games. >>>>> >>>>>Disregarding Dortmund (super performance) and the Dutch championship, my problem >>>>>with the above is that these total games are done over quite some period of >>>>>time, a game or two at a time, and on variable hardware and conditions. To >>>>>qualify for a Fide rating, one cannot just accumulate 9 games with one here, two >>>>>there, another here, etc. It has to be done with groups of at least 4 games from >>>>>a single event. >>>> >>>>As pointed out many times, no one is talking about earning the Fide title of >>>>GM...this is calculating the computers strength vs human GM's. >>> >>>That's my whole point. My Fide rating cannot be acquired as a result of a game >>>here and a game there. This isn't a question of titles, but merely the correct >>>Elo relative to my playing ability. Why do you think Fide won't consider 5 >>>groups of 2 games against Fide rated players (10 games) in order to give me a >>>first rating? It can ONLY be done with groups of at least 4 games from a same >>>event. This isn't done for the sole purpose of adding complications. >> >>Why disregard Dortmund...if we calculate the rating of tournaments disregarding >>spot games, the ratings of computers is stronger. So what is your point. > >I didn't say to disregard the result of Dortmund, but to disregard Dortmund >concerning the complaint I was making. In other words, regard it! :-) I realize >we must do with what we have, but if we are going to flash results and >statistics, then all the provisos regarding those results and statistics must be >considered and declared. In other words, show the results, but be honest about >the limitations or possible problems. Yes, Junior performed brilliantly, and >*perhaps* this represents its true strength, but unfortunately it is only one >tournament. Karpov had an absolute field day in Linares 1994 (I think it was 94) >but did this really represent his relative playing strength against the other >players? Yes, we have data, but it is a couple of tournaments, for two different >programs, and a collection of single games here and there. It's still a far cry >from nothing, but it's also a far cry from being conclusive. With all due >respect. Here's another tidbit: you'll sometimes read about some top board >performer in league play. Such as Christiansen did a couple of times in the >Budesliga with scores on the order of 12/13, etc. With such a result does this >mean he is a WC contender? Unfortunately not. It's a fantastic result but it is >very different from playing a tournament. Perhaps Junior wouldn't be any >different, but what about the players it played and the circumstances? No one is trying to determain the exact rating of any program...what we want to know is what standard of play does the data give us. We have more data coming in, tiger is doing well in its tournament. If it wins that tournament this also suggests a certain level of play for all top programs. > > Albert > > >> >>We have to look at the data that we have.... >> >> >>> >>> Albert >>> >>>>> >>>>> >>>>>> >>>>>>The performance goes up, not down due to improvemets in SW and HW over the last >>>>>>3 years. Pretty stiff competition. :) >>>>>> >>>>>>Best Regards, >>>>>>Chris Carson
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.