Author: Robert Hyatt
Date: 18:38:27 07/19/02
Go up one level in this thread
On July 19, 2002 at 15:50:08, Rolf Tueschen wrote: >On July 19, 2002 at 11:50:51, Matthew Hull wrote: > >> >>It seems that most of the controversy can be boiled down to the axb5 vs Qb6 >>issue. > >Objection. > >That was Kasparov's reason for his question for the logs, yes. But it's not the >biggest challenge of the DB2 team. The biggest examination came from me with the >miles higher levelled questioning of their experimental design and ethics. >Please read all the exchange of arguments between Dr. Hyatt and me. The point is >that the logs seriously do not prove a thing. Nor would they _ever_ have proved anything. So the discussion again revolves around a "pointless issue". > > >> This was the ONE BIG evidence Kasparov had of possible cheating. It >>seems in fact that this is the only thread by which the cheating allegation >>hangs. If that could be demonstrated, then who knew how many more moves in the >>match would have been "corrupt", yes? >> >>But we find that logs from other chess engines show a similar DB2-like regard >>for the two moves over time in that the two evaluations show a trend toward >>intersection. This greatly weakens (if not completely destroy's) Kasparov's >>strongest evidence of cheating. > >Objection! > >Why do think that completely different entities could be taken for refutation or >corroboration of DB2 play? That is comparing apples and peas. (That is why I >demonstrated the importance of the deconstruction of the machine. It simply >destroyed all possible evidence. The rest is speculation. At whose costs? The >team of DB2 is guilty of.) Several reasons why your objection is overruled. 1. Kasparov said "no computer could play axb5 instead of Qb6" which he claimed was a larger material win. He was wrong. I demonstrated that _one_ program thinks the moves get closer and closer with deeper searches. One exception is more than enough to disprove any statement that says "no computer can..." Because if one can, it is is false... Just because Crafty _might_ switch to axb5 given enough time does _not_ prove that DB actually did that without human help. There is _no_ way to prove that. But it is easy to prove that _some_ computer might well play it given enough time. And it _has_ been proven that the two moves are within 1/2 pawn of each other, which is more than close enough for a few differences in evaluation to change the move. > >> >>All the rest of the arguments about courtesy, politeness, friendliness, supposed >>sinister motives (please provide Hsu's and Murray's brain logs for evidence!) >>and goodwill do not strengthen Kasparov's axb6/Qb6 contention, by which the >>entire cheating allegation hangs. That is the only thread Kasparov ever had. > >In science truth is not depending on some individual's limitations. As I said I >did make the main accusations! In your collection above you forgot to mention >ethics of scientists. And this is completely independant of Kasparov. It's a >'must', a duty, of the team. Unfortunately they failed. Perhaps you don't >understand it at first. But then read please my exchange with Dr. Hyatt. The >main reason lies in the early stages of their experimental design of the >machine. Perhaps the whole question could be led back to the early times of >tournament computerchess. At the time the protagonists simply missed the >question of documentation, because it was just an academic fun at weekends where >all parties had the same good status just by their participation in the new >research. The chess itself was not the most important factor in these days... >simply because it was very weak chess. Even the operators were sometimes >stronger than the machines. :) > >And another important point was the long-distance connection to the huge machine >power. So, the question of cheating would have led to nowhere - because anything >in fact was possible. But to what purpose? That's why the question wasn't top on >the agenda in those ancient days. That is why the "question" has not been an issue in 30+ years of comp vs GM matches... > >>And his contention has been scientifically undermined, if not completely >>destroyed. > >Objection. > >> >>_That_ is science. If it's good enough for other engines to find the evaluation >>trend, if not the move itself (given the un-avoidable technological >>disparities), then the DB2 log validity is only strengthened, not weakened. > >You must read what Amir wrote about it. The question is if DB2 would reject the >present of three black pawns! Of course PC machines can be instructed to reject >it, but facts speak a different language - that comps still are a bit too >greedy. > >>That's where the true scientific evidence points. End of story. Game over. >> >>Yes? :-) > >No, I'm so sorry. :) > >Rolf Tueschen > >> >>Regards
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.