Author: Uri Blass
Date: 13:31:12 04/28/05
Go up one level in this thread
On April 28, 2005 at 15:40:05, Rolf Tueschen wrote: >On April 28, 2005 at 14:12:05, Robert Hyatt wrote: > >>On April 28, 2005 at 13:19:07, Rolf Tueschen wrote: >> >>>>Because there is none. >>> >>>Look, the science question is "what is the chess strength of the machine". If >>>you ask that can you then measure the confusion of Kasparov? I would say no! And >>>this is already part of science we are in. >> >> >>No. That is something _you_ are injecting. The question was never "what is the >>chess strength of Deep Blue?" The question was "can the machine beat the best >>player on the planet at standard time controls, using the normal rules of >>chess?" The answer was "yes". > >The answer is a clear No. DBII didn't win a single game in match 2. Surprised? > > > >> >> >> >>> >>>We don't have real chess matches between machines and humans. All what we have >>>is exhibitional humbug. >>> >>> >> >> >>I don't know where you get that from. I've played GM matches that were >>broadcast live on TV and on the radio. I've sat across the table from these >>players and from my observations, they are treating the games as "real matches" >>and not an afternoon "lark". > >Chessmasters of human chess are very shy and proud people. In front of cameras >they can be confused. They lose concentration. > > > >> >> >> >>> >>>>No, I'm not "unaware of them". They simply do not exist... >>> >>> >>>Of course David Copperfield does also use science to prepare his spooky >>>delusions. Science isn't religion and you are not free to either believe it or >>>not. If you say that science doesn't exist in computerchess vs human chess, then >>>you are not a scientist... <gg> >>> >>> >>> >> >> >>Science exists in computer chess. But not in the matches that are played. >>Humans are too inconsistent and the standard deviation / variance is way too >>high to think in terms of "reproducible experiments". It just won't work that >>way so long as humans are involved. Even computers vary a bit due to minor >>timing changes, even if you try to play the same exact game over and over there >>is no guarantee that you can do that with a computer. > >I see no humans in chess who could be researched in computerchess but I see >machines. And human chessplayers can tell us how strong machines play chess. >Viewed from the human chess standpoint. > > > >> >> >> >> >> >>> >>> >>>Bob, we are NOT in politics here. I say that it doesn't matter who fired the >>>first shot - IBM/Hsu STILL were responsible for a sober science in the >>>experiment... Bad luck for you (as a scientist). Times are gone that you as >>>scientists make your weekends of fun in computerchess. You must create sensible >>>rules for matches between the two spheres. Computerchess and Human chess. >>> >>> >> >> >>How can one create more sensible rules for a match, than to tell your opponent >>"you think about this match, you explicitly define each and every rule you want >>used during the match, and we will go along with your rules exactly." >> >>That is what happened twice in NYC. The only difference in the two matches was >>the outcome the second time around, which seems to be unacceptable to some. >> > >Your insinuation is wrong that Kasparov could have proposed all the rules he >could imagine. Kasparov was not the one who wanted to lose his face by inventing >rules as if he feared the machine. He was certain that the team around Hsu would >respect all decent rules of science. He couldn't know that Hsu was planning to >lose his face... > > > > >> >> >> >> >>> >>> >>>You know that this is basically the death of serious computerchess as far as >>>competition with human chess is involved. Then make this clear in your field. >>>Or. Or find solutions to the problem. There is no "Doesn't work", "Never worked" >>>or "Won't work". This is solvable. If not forget about future competition with >>>human chess! >>> >>> >> >> >>Baloney. Because we also can not prevent the _humans_ from cheating either. Do >>I need to enumerate spectacular examples of such? Is human chess going to >>disappear because humans can find a way to cheat? Is human chess going to >>disappear because we can't find a way to prevent all forms of cheating until >>after they are exposed. Once we find a cheating methodology, it is possible to >>design countermeasures to prevent this. But even Las Vegas, with its years of >>experience in gambling, still can not prevent all forms of cheating, because >>nobody knows what "all forms of cheating includes" until they are discovered and >>then analyzed... > >Kasparov couldn't know that Hsu and team would suddenly apply undecent behaviour >after they had worked so hard on DBII over decades. A sane scientist wouldn't >risk his reputation. > > > >> >> >> >> >> >> >> >>> >>> >>>You mean you or the machine? Of course that is ok. But by all means I say: let >>>the machine decide things. Otherwise it's a hoax and impostering. And make exact >>>protocols to research things if wanted. >> >>Both. I prepared openings against specific humans when _I_ was playing. I did >>the same when my program was playing... >> >>Computers are _not_ human. Trying to make them such is not worth talking about >>today. > >You must respect the rules of FIDE. Machines have no extra rights... > > > >> >> >> >> >> >>> >>> >>>That is ok, but don't dream of the final matches after you gave birth to a new >>>baby... Go through the fire of improvement and success. >>> >>> >> >>DB was 11+ years old when it played Kasparov. It had grown, and been modified >>many times. Of course humans go through that same process as they age as >>well... Deep Blue 0 (deep thought) had played dozens of matches against GM >>players and won almost all of them. It was worthy of playing the world champion >>based on those games. Deep Blue 1 played Kasparov and lost, but it had its >>moments. And Deep Blue 2 beat him. It wasn't a "new star from out of nowhere" >>it was a process of evolution based on 11+ years of development. > >Fine by me, but not fine enough for players like Kasparov or any top human >player. You can't motivate them to create anti-computer chess for just a million >dollars... It's way better for them to continue with the draws so that companies >don't lose interest... No I think that it is the companies who do not want them to create anti computer chess. Otherwise they will give the following conditions: If you win a match of 6 games you get million dollar. If you do not win the match you get nothing and the next best player who agree to play against chess programs is playing a match against the program. Continue this process until the human win a match or until there are no more players with rating above 2500 that agree to play a match. 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.