Author: Ed Schröder
Date: 22:02:21 01/06/00
Go up one level in this thread
On January 06, 2000 at 05:30:55, Graham Laight wrote: >On January 06, 2000 at 03:25:28, James Robertson wrote: > >>>This seems like a very funny thing for you to say. The allusion you have made >>>in the field of logic(especially Toulman logic) is called an "appeal to >>>authority". And this type of argumentation can be effective, however when >>>people make an appeal to an Authority on ACTUAL chess play like larry Kaufman an >>>International master who claims that programs are around as strong as the top 5 >>>U.S grandmasters(on mere p400s at that) you dismiss it out of hand. >> >>You know? While I respect Bob's and Ed's opinion, I really don't care what it is >>as long as the only evidence I see - Rebel's performance rating - continues to >>stay below the GM level. This is not some mere mortal's OPINION, it is actual >>fact. >> >>James > >It's also an actual fact that the recent versions of Rebel do not have an SSDF >rating. Rebel might be like Genius - reached a certain level and not improved >from there (even if people think that it plays more stylishly now). > >As for EVIDENCE - I would have thought that an objective observer would have >said that the balance of evidence comes down strongly in favour of those who >believe that the computers are into the grandmaster range. > >So - why do I think that programmers refuse to say this? > >The following are not intended to be seen as accurate analyses, but rather as a >demonstration that "reasonable doubt" exists about the programmers' >impartiality. > >* In the case of Crafty, maybe it really isn't at GM level - or maybe Bob is >demoralised by the number of times it's been beaten on the ICC, forgetting that >that the ICC conditions, while useful for research, are not truly representative >of tournament chess conditions > >* In the case of Ed Schroeder, I think he suffers from low self esteem. He seems >to have developed a fear of competition. Also, maybe the long years of >proclaiming great results, only to be knocked back down by a humiliating defeat >at the hands of a GM have created within him an automatic reaction of excessive >modesty when asked about his programs relative to strong human players > >* In general, every programmer who has spoken to "positional" IMs and above will >have heard expressions like, "What a terrible move that was positionally". >Certainly many GMs said that many times about DB during its successful match >with GK in '97. But DB still "brought home the bacon". > >* Chess programmers are often timid people, who prefer to have other people heap >credit on them, rather than shouting out their achievements for themselves > >* For so long, computers have been worse than GMs, and GMs emphasise the >computers' weaknesses so strongly, that it is easy to see how, in the absence of >truly compelling, utterly indisputable evidence, most people could easily miss >the moment when the computers really do reach the GM level. Think about this: it >is conspicuously clear that the vast majority of Wall St traders miss the >moments when the market's primary trend turns up or down sharply - it's almost a >truism by definition! > >But if Bob Hyatt wants to laugh, I say "go ahead" - laughter is known to be good >for your health! > >-g Graham, to think about: Rebel-9 beat Yusupov convingly at 5/all. Yusupov did clearly in the longer time control games. In the six 5/all games Rebel-10 scored 4.5-1.5 against Anand, TPR far above 3000. Rebel-10 did win with 1.5-0.5 in the two 15/all games. Rebel-10 lost with 0.5-1.5 in the two 40/2h games. In comp-comp (SSDF for example) there is not much difference (+/- 50 elo) playing blitz or 40/2h games. In human-comp I estimate that difference at 400-600 elo points. Get the point? Ed
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.