Author: martin fierz
Date: 13:55:07 09/11/02
Go up one level in this thread
On September 11, 2002 at 13:04:52, Robert Hyatt wrote: >On September 10, 2002 at 07:31:29, Rolf Tueschen wrote: > >>On September 09, 2002 at 20:32:22, martin fierz wrote: >> >>>On September 09, 2002 at 11:53:42, Jeroen Noomen wrote: >>> >>>>On September 09, 2002 at 06:08:08, Côme wrote: >>>> >>>> >>>>>I don't really agree here, Remember tinsley retired from the match after 6 draws >>>>>because he was litteraly dying from cancer ! so chinook draw a dying man :-) >>>> >>>> >>>>I disagree. You cannot blame Chinook here, Tinsley was still the >>>>very best in the world and in the six games Chinook was putting >>>>the pressure. >>>> >>>> >>>>>Then he was replaced by Don lafferty and Chinook drew the match ! >>>>>Remember Lafferty despite being the second best player in the world he was >>>>>really a LOT weaker than tinsley ! I think tinsley rating was +2800 and Lafferty >>>>>like 2650 ! >>>> >>>> >>>>You still remember how Lafferty did accomplish this? By playing for >>>>a draw in every game! It is like playing soccer using 6 goalkeepers :-). >>>>Anyway, it is not fair to state that Don was a lot weaker than Marion, >>>>you wouldn't say that Anand is A LOT weaker than Kasparov, won't you? >>>>Number two in the world is number two. >>>> >>>> >>>>>So I do believe a full strenght Tinsley would have kicked chinook badly. >> >>This is what I also thought from the few informations I got. But let's discuss >>such a problem. What I can do is making conclusions of all the statements here >>and then I found something interesting. So I have to thank all of you. This is >>how the internet works! Thanks Ingo for the high-classed "troll", thanks Jeroen, >>Uri, Bob, Martin and others. >> >>Folks, what is a "full strength Tinsley"? To keep the debate as lively as >>possible I make comments straight across the former contributions. > >I think Tinsley was "full strength" for the six draws, but he knew he was >ill and stopped when he thought it was the right time. IE knowing him >fairly well, he would _not_ have gone into such a match not ready to play, >it just wasn't in him any more than it is in a GM player to intentionally let >someone win... > > > > >> >> >> >>>>>Laffert played hundreds of games with tinsley and he said he only beat him ONCE >>>>>and it was very late at night and tinsley was tired :-) >>>> >>>> >>>>You can believe this, but we will never know. All we can see are >>>>the 6 games played and in those games Chinook was never in trouble >>>>and once came close to winning. >> >>We have inconsistencies. Bob says that Tinsley had a close relationsship to the >>Schaeffer team. Martin says that Tinsley could have done bettter if he had had >>the chance to play more games against Chinook. My question then: is it the same >>situation as Kasparov had it when he had to play the completely secret DB2? >> >> > > >Tinsley was an unusually open person, regarding checkers (and to me, chess). >Jonathan knew him quite well and they regarded each other with a world of >respect. I can't speak to whether he played any real games against the full- >blown Chinook system prior to a match or not. That is better discovered by >either asking Jonathan or reading his book, "One Jump Ahead"... hi bob, i can answer that for you: tinsley didnt have the opportunity to play against chinook before the match. IIRC they also made some changes to the eval, specifically to make it play different so that his preparation would be for nothing, not because it would play better. >However, checkers is the kind of game where I don't think that secrecy would >have made one scintilla of difference. i disagree with you. there were very few games of the "full chinook" or anything remotely close to it. one main aspect of checkers is that if you win a man, you usually win the game. this makes us checkers programmers all implement relatively aggressive forward pruning. if you don't do that, you are hopelessly outclassed by a program which does that. however, every now and then a position pops up where you *can* sac a man without obvious immediate compensation (getting a king in return), and in these positions, all programs i know have trouble. cake lost two games in las vegas, because my pruning was too aggressive in these situations. i've changed it a bit now, so it finds these things earlier, but it still has trouble seeing it. the other programs in vegas had this stuff in book. they definitely also had trouble finding that stuff without book. there's a famous chinook-tinsley game where tinsley sacced a man and nearly won. that was long before the final match. the game chinook lost against lafferty in the follow-up match after tinsley-chinook was cancelled? yup, lafferty sacced a man and won... this is THE achilles' heel of the checkers programs. and i'm not sure how aware tinsley was of this. maybe, before the match, tinsley thought that they had fixed the thing. and that was why he was pessimistic. but the lafferty game showed otherwise, and he just might have changed his mind. a chess example would be: for the first deep blue - kasparov match, somebody might have said: "wow, X million nodes per second, kasparov will have no chance". and then he would have seen that one terrible game where in the end everything of deep blue was tied up, where it looked like a patzer. and changed his mind immediately. of course the 8-piece db makes these programs real strong. but they have their problems, and if you know how to look for them, you can still beat them. the game cake lost to nemesis is a good example, BTW. murray saw an old version of my program lose this exact game against nemesis without book, and put it all in his user book. it was a man-down-line... aloha martin > > > > > >> >>>> >>>> >>>>Jeroen >>> >>>we will never know is the right answer, but we can see two more things besides >>>the 6 games between tinsley and chinook: >>>1) immediately after the match against tinsley, chinook played 20 games against >>>lafferty and the match ended +1=18-1. chinook made two serious mistakes in this >>>match, losing a drawn position, and failing to win a winning position. >>>2)tinsley was a stronger player than lafferty. >>> >>>these are both facts. obviously (unfortunately), you cannot make any kind of >>>conclusion like >> >>Of course we can make conclusions. >> >> >>>a) chinook was better than tinsley >>>b) tinsley was better than chinook >>>from these facts. >> >>So what means "was better"? Proven by concrete results? No, we have no such >>data. But still we can say that conclusion a) is false. We are in the similar >>position in chess, where we know for sure that even DB2 was a beginner in >>certain positions. So this alone disqualifies the achine to "be better" than a >>Wch.Both Tinsley and Kasparov habe proven that they could draw against the >>achines. Lafferty was even more successful in drawing. Couldn't we make >>conclusions with some logic here? I fear we can't do that! >> >>We had to know much more about the personality of the players. Bob gives us a >>ultifaceted character of Tinsley. He was honest, serious, and on the other side >>at least in chess a player, if not a gambler, who tried to make it with his >>perhaps 2100 against the CRAY of Bob. In nightlong events. >> >>I have some ideas for the reasons why Lafferty, whose personality is even >>completely unknown to me, succeeded apparently even better against Chinook than >>Tinsley. It is a completey different task to be the best (even worse if the best >>by far!) in human checkers and to play such a machine. Because you don't have >>too much experiences with equally strong resistance. You simply don't know how >>to fight in such situations. Typically Tinsley said - after very few games only >>- that Chinook were the better player. But how then could Lafferty hold the >>score open over so many games, over sixty in sum? Isn't it more the depression >>that took hold of Tinsley when he "felt" the strong resistance on the machine's >>side. What he never before had felt in his very long career? > >Don't forget the _difference_ between a chess match and a checkers match. > >In chess, decisive games are relatively common. In checkers, not. I once >followed a match with Tinsley where there were so many draws I thought it was >pointless for the match to be played. IE 60 draws in a row or some ridiculous] >number like that (don't hold me to 60, this was a long time ago and it probably >seemed bigger to me than it really was. IE in a WC match, if I saw 30 >consecutive draws that would seem so large I might remember it as "big" rather >than 30.) > >So drawing many games in a row does _not_ say the two players are equal. IE >the Elo formula would not be a good for checkers because all those draws would >tend to pull the two player's ratings close together even if Tinsley won every >match they played... > > > > > >> >>So, let's ask the question in a different version. If the match would have >>lasted over dozens of games, would Tinsley by and by have learned how to play >>with the conscience of an almost equal opponent? We don't know. From the facts >>Bob reported I have serious doubts. Ok, he had strong stamina, at least in >>chess, when he tried and tried all night long. But we must not forget that there >>is always a difference between passtime fun, even in addition with some >>masochistic appearances. > >I would not call that marathon "fried liver" match anything but "fun". Tinsley >and Charles Walker (owner of the world checker hall of fame facility) played as >a "team" discussing moves, etc. They even asked me to help many times, which >I tried to do, without looking at the Cray output. :) So it was "fun" and not >"serious work". > > > > >> But it is something totally different to be frustrated >>in you major field where you were the champ for decades, yes, almost a century! >>Those who knew him should rethink all this. Perhaps they can tell us what could >>be probable in this case. >> >>For now we must see that Tinsley surely have had the strength to win such a long >>match, but perhaps Lafferty was only so successful because for him it was not >>such a frustration to only draw in dozens of games. Checkers is too unknown for >>me as that I could say how far such series of draws matters anyhow. >> >>I must also add as always that the aspect that Chinook was a machine forcedly >>also influenced Tinsley. Had he enough informations and experience with >>computers? Martin seems to disagree as far as games are concerned. > >Tinsley was impressed by Chinook. And he followed the database issue closely >as he could see how it improved as they tacked on another layer of endgame >databases. Remember that Jonathan reported that in the last match, something >like 70% of the opening position moves were terminating in database hits, >which means it was close to playing _perfect_. Note that again, that 70% is >from memory, so his book would be a better source for the precise number... I >never attached a lot of significance to checkers since I was not doing checkers >myself, and that 70% could be off a ways... > >> >> >> >>>but i think it's clear that if lafferty could draw a match >>>against chinook, then "by b)" tinsley might have managed that too. >> >>That is the big question. With logic alone this can't be said. Did Tinsley ever >>meet a player equally strong? > >He said "no". Specifically. But that was his words, so only he could respond >to what specifically he meant, had he been asked prior to his death. > > >> >> >> >> >>>and if >>>chinook made a serious mistake once in ten games, who knows - it might have lost >>>a match to tinsley. >> >>To a Tinsley who must have digested the many draws like Lafferty. >> >> >> >>> or it might have won. >> >>No. This could not happen because of its strength but only because of Tinsley's >>depression. > >I don't agree there. Tinsley was absolutely convinced that Chinook was better >than he was. There was no "almost" or "almost certainly" or anything like a >qualification to his statement. Of course, it could have been a "gentlemanly >statement" or "a prepare everyone for the worst, knowing that it really wasn't >_that_ good" but speculation is pointless since we will never know. I never >found him to be a "hyperbolic" type person, so I would personally take his >statements at face value. Unfortunately I have a counter-example. He was >convinced that the Fried Liver was winning for white. It isn't. :) I knew >the thing inside-out because I personally liked to spring it on unsuspecting >people in our university tournaments. After CB smashed him repeatedly, I >even played him a couple of games and won them easily since white is a piece >down if the attack fails, which it always does against correct play. So there >is one "ego-produced statement" that was wrong. Were there others dealing with >the skill of Chinook? I don't know, but I personally doubt it. > > >> >> >> >>> you cannot tell. but there is no >>>evidence that tinsley was really outclassed by chinook. >> >>This is clear. But being outclassed brings the same results as being >>"frustrated" e.g. > >When you play "near perfect" checkers, as Tinsley did, then _nothing_ is going >to "outclass" him. Beat him? Yes. But that close to perfection, he wasn't >going to be outclassed by anybody or anything. > > >> >> >> >>> all we have to go on is >>>that according to bob, before the match, tinsley was really afraid of chinook. >>>but i think in part this is because he had no chance to play against it or to >>>study it's games. >> >>Thanks for giving the point! > >I don't personally know that that is true. I know he was convinced the machine >was "stronger" although I don't attribute "fear" as the underlying emotion he >felt. More "concerned" than anything else, because he _knew_ it would be a >"mental workout" to play the thing... Probably the same sort of mental workout >Kasparov discovered when he got into the 1997 DB match when he found out it was >not just "fritz on steroids".. > >> >> >> >>> for example, what would he have said after he saw chinook's >>>loss to lafferty, which was all book play? would he have changed his mind? >> >>Surely! But if that could have changed hispersonality's reactions? We simply >>don't know. It depends on the past experiences. Does anyone know if Tinsley >>really had stamina? Know what I mean? Please read about the young player Karpov >>when he was still under ten! Unbelievable!! Who ccan still remember the Wch >>match between Karpov and Kasparov with the long series of draws!? > >Tinsley had _incredible_ mental stamina. Checkers requires that since the >WC matches can go on and on and on and on, like that damned pink rabbit. :) > >> >>(Therefore don't miss today the *FREE* LIVE reports from Moscow on the FIDE >>site! LIVE games of all the Wch's. Karpov, Kasparov,Kramnik, Ponomariov, Anand, >>Khalifman etc.) >> >>Rolf Tueschen >> >>> >>>aloha >>> martin
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.