Author: Rolf Tueschen
Date: 04:45:59 09/17/02
Go up one level in this thread
On September 16, 2002 at 21:22:23, Robert Hyatt wrote: >On September 16, 2002 at 15:53:06, Rolf Tueschen wrote: > >>On September 16, 2002 at 14:05:49, Robert Hyatt wrote: >> >>>On September 16, 2002 at 12:46:34, Rolf Tueschen wrote: >>> >>>>On September 16, 2002 at 11:30:29, Thorsten Czub wrote: >>>> >>>>>Rolf Tüschen claims that i cheated when playing autoplayer games. >>>>>In fact he says that the openings have been chosen or selected, and that >>>>>i maybe preselected special games. >>>>> >>>>>This is a personal insult. >>>>> >>>> >>>>Wait a minute. Shall I get the personal insults you wrote to me? The last ones >>>>here again to these threads? Being ill, needing doctor and such nonsense? >>>> >>>>It's better than a satire what you are doing here. >>>> >>>>NB I didn't say that you cheated. I said that you were a creative tuner. And >>>>then what I said next, that was the following. I showed clearly that your >>>>roughly 30 games you reported - attention - _do not prove_ what they apparently >>>>shouldm that Rebel Macheide were stronger than Fritz or Shredder. Period. >>> >>>I personally don't like _any_ of this thread. >>> >>>Why? >>> >>>Because it is _impossible_ to prove a negative. IE it is very difficult to >>>prove that someone _did_ cheat or manipulate results. It is impossible for them >>>to prove they did not. >> >>You are right and excuse me that you must deal with such trivialities. >>But please let me give a last explanation. Bob, I hope that my English is not so >>bad that one can understand that I - at first - asked questions. Questions can't >>be insults here. At first I asked specific questions for the five examples I >>gave. What did Thorsten do. He insulted me. I did the necessary here in CCC >>without a reaction. Then after Thorsten's insults I began to reflect why he >>reacted so confused. And then I realised the main point. >> >>NB please that I did never say he faked or cheated data. All I said was: These >>30 games including the five games I posted could_not_ show something relevant >>about this styled Rebel. >> >>I thought that this was a real discovery and therefore a reason to publish it >>here in CCC. I expected applause but notthe critic that _all_ should be taken >>with care because all could be cheated (what Ed explained to me) or that tests >>should always taken out of interest but never with too much reflections, in >>short either one loves them or one leaves them, but no discussing about them >>(what you said right now). Honestly, I don't understand it. >> >>But since you both (Ed and you)know CC better than I do I will try to follow the >>guideline. So here in this group test results should not be discussed. > >I don't think Ed said that, and I am certain I didn't. Indirectly. Ed said that nobody could prove cheating which was always possible. I read this as not a forbiding but making any analysis meaningless because you never know exactly what's going on. And you simply didn't give yet he valuable explanations you added now below and for which I am very thankful! I missed such clarifications for years now! Thanks so much. >Test results are >_always_ interesting, and (IMHO) perfectly acceptable to talk about. So >long as everyone remembers that they are not "absolute". Strange things >happen. Someone posted in r.g.c.c today that Crafty tore Fritz up in a >match he played. That is within the set of possible outcomes, knowing full >well that Fritz is generally better than Crafty overall (not in some parts of >the game, but overall). > >The result was believable. The dependency on this result as "absolute" was >wrong however... > >In the case of this match, strange thing can happen anywhere. As Thorsten >said, it is possible that for a particular program, its first non-book move >could be horribly slow and shallow due to quirks in how it allocates and >deals with a "new" hash table. Programs vary due to timing issues. Programs >vary due to outright bugs. Any (or all) of which could cause strange moves >that might never be reproduced... > >This is similar to your walking outside and a bird hitting you with a direct >shot of bird crap. Not very likely. But certainly far from impossible... Two things. Not too seriously the first. When I was in Amsterdam for the first time I went to the post near the Dam and wanted to give my parents a call. Before I could enter I got such crap right onto my jacket. Fortunately I can't remember how I survived the holidays. Perhaps this gave the reason for my long friendship with Holland. :) But back to the point. Why Thorsten presented _these_ games, since he didn't make any statistics but he was tuning and wanted to see the effects of his work? That was my question! He's fooling me now with answers like "But Rolf, should I really have left out these games, which would have been a gross wrong doing, didn't you know that!?" as if suddenly he were doing straight forward testings... as if I had insinuated that he _should_ hand pick single games! So please let's be careful. He was not presenting the games from _matches_ or a single match. He posted games. But my question then was what these games could show. In my eyes nothing about the growing strength of Macheide Rebel. Only Uri proposed how that could be tried to research. By comparing games of both versions. Default and Macheide against the _same_ opposition. BTW do you know how many games one could play in say 1 day, 1 week and 1 month? That was my next question. Why Thorsten did play only so few games! > > > >> Also >>because (that was what Jonas told me) could be an attack on the "human" approach >>of the specific tester. All what I understand now is that the testing seems to >>be a delicate thing, like a love in real life, and there we would also never >>criticise someone for his taste. Beauty lies in the eyes of the observer. >> >>On the other side it's a bit difficult for me to oversee that perhaps business >>interests might cause me to simply neglect clear biases and other weaknesses of >>a practice. Because in science there are no such taboos. I thought that testing >>was object of free critic. >> > > >There is no doubt that testing in the past has raised the spectre of >suspicious circumstances producing suspicious results. And that will continue >to happen when someone has a particular agenda to promote. I don't see a >solution, and so I choose to avoid the discussions. But other than Ed I would say that those who post exact descriptions of their experimental design, complete games of complete matches, exact records of the failures perhaps due to autoplayed games during the night sessions (Thorsten suggested that all must have such happenings but only few reported them) and absence of too boasting headlines, those results then have more credit IMO than such 30 games from Thorsten with headlines like (I quote by heart) "Wow! Macheide is crashing Fritz and now Shredder!" When his games contain such examples I had reported and analysed. Let's talk about agendas. Thorsten is famous for his aversion against beancounters. He's conten-oriented. So far so good. But then he should give us analyses of his games. But he didn't say too much. Therefore I was so shocked when I discovered that at least in the 5 examples (of only 30 games) there was surely _not_ something that could be interpreted as a quality of Macheide. So statistically it's perfectly ok, to "bean"-count these games. But in the understanding of Thorsten they meant nothing at all. And he already confirmed that judgement. But then he writhes around the crucial question why he had presented the data at all! Know what I mean? I'm not insulting Thorsten. I'm just taking him by his own words and wanted to remind him of his high claim. Please keep that in mind for later below when you discuss the question what motivation I might have had in the eyes of neutral observers! :) > > > >>So, in the end I begin to understand why also the SSDF topic demands >>a lot of tact apparently. > > >It is easy to find flaws in _any_ testing methodology, which is a shame. >But ponder=on, ponder=off, auto232, winboard protocol, UCI, the broken >winboard adapter from ChessBase, programs hogging the CPU when not thinking, >special books, custom books, hidden books, cooked books, borrowed books, >stolen books, all lead to different results that could be challenged on >any of the above grounds... > >So perhaps, rather than challenging, which could be done for every event >reported here, it is easier (for me) to simply read them, say "aha" or "oh" >and save my writings/musings for things that won't escalate into a meltdown. For these two paragraphs more so for the first of them I want to thank you! For years now I'm thinking without success about the question why you never commented on all the big issues SSDF and similar. Any snip I could find was always of importance so far. Your precise rejection of the claimed already GM status is just an example. I really thought I had to convince even the most famous experts here in CC of such trivial things from methodology. Jonas was the first who made me think! He mentioned the human factor! And I thought that that could be the reason for many experts to keep their silence when others go on tangents again. Honestly exactly these points however are for me the most important motivation and also fun for my participation in the communication since 96. I still think that we have no alternative to the strict rules of science. At least if we must judge what is ok and what is not ok. Then there is no room for such excuses like ""But the SSDF is a private organization with no demands." Of course we must show tolerance if something spooky happened again. But we cannot change the rules for private hobby groups. Or we must define the complete activities having no relevance in nuce. I doubt that this is what the members want. But then the tolerance of clearly false methods is a crime in my eyes! At least known academics should not participate in such wrongdoing. I mean is it not for CC as a whole we hold up the true rules of methodology? We are not nitpicking or being pretentious, no? What is the reason why people like Thorsten always gets new credits for the odd defaming of all what could be called science and methods? I had no problem with that if at least the official level would be defined from time to time. It's the complete opposite in the internet. Here he can write 100 times against beancounting and then readers might think that his way might be a correct alternative to often boring maths. IMO that shouldn't be supported in the case of grown-up adults of over 30 years... > > > > > >> >>It's a pity that positive and constructive criticism seems to have no value >>here. Because since when the truth could be a question of tact? Nobody is >>mourning here? > > > >It is more a question of personalities, the nature of "virtual interaction" >and a natural outcome of dialogue. The problem is that what is "constructive >criticism" to some might be a "outright defammation or flame" to another. >That's the way it has _always_ been on network facilities like this. That >is the way it will probably always be... Perhaps we have a typical cultural problem here. Bob, the fact, that critic as such _basically_ is _always_ a good thing. And that is not a question of estimation but this has been judged some _thousands_ of years ago! So that is not open to democratic compromises. Of course it's also true that only for really stable personalities that truth could be really meaningful! And it's also true that already thousands of years ago people in rage or in a 'zwick' took the first they could get to make him responsible for their fate. That's where the saying comes from "DOn't shoot me, I'm just the reporter!". ;) Let me give you my general thanks for making such a prudent statement about the actual conflict. Reminds of certain events in the past when you showed that same class. I never had the luck to meet something comparable in the German groups for CC. If I only could reach a certain improvment with my own writings and case! Civil courage is so important in the question of decency of man. > > > > >> >> >>But back to Thorsten. The next time when I should find such a strange happening >>(these three games with the opening book bug) and I ask a question here I should >>tolerate the insults of Thorsten and realise that now 'tact' is the only >>guideline possible? Since any further critic or substantiation could cause too >>much pain for Thorsten? I'm sorry, but this logic was new for me. But I will try >>my best. If only a first question is possible in future. Or is that also >>unwanted? Ok. But when do you tell Thorsten that _he_ then should also >>concentrate on what he calls tuning or testing? Why he has the right to talk >>about mental illnesses and beancounters and other such insults. I think I am not >>the only one being Thorsten's victim? Look what he wrote to Chessfun. I thought >>that personal attacks were _completely_ impossible here in CCC? > > >I didn't follow this in the beginning. But if you look at it through neutral >"eyes" you could interpret your post in at least two different ways. (a) as >you claim, a simple intent to point out that there seems to be a problem that >needs investigation to see what caused the strange move(s); (b) a direct >assault implying that something dark and sinister happened. > >As is usually the case, the truth probably lies somewhere therein... Usually but not here in the case. What perhaps might have struck you as well, the continuing story of the five examples had a most trivial cause. And I have described the reasons. It was simply a self-runner the moment I once discovered the nice feature of the scrolling analysis for a single variation in Fritz engine. I saw many analyses sent here into CCC but I always thought that this was from special settings for the testing experts. I had always the one-line analysis on the display and also posted it long-time back in rgcc. Suddenly I discovered this "My Town Stuff" and could get it from clipboard and then paste itto my posting. In about two hours I had the 5 example ready one after the other of course. So, it wasn't a flame war or such, it wasn't some strange addiction or a suffering stroke, it was simply the sudden using of a new tool. So, I must pleade innocent so far. But in the end I'm happy for having survived the insultive climate and I must admit that many good explanations on the basic content were sent. Let me thank all of those who took part. Rolf Tueschen > > > > > >> >>But now back to something about chess. >> >>Rolf Tueschen >> >> >> >>> >>>And as a result, all that is going to happen is that tempers are going to flare, >>>debate is going to heat up, and this will erupt into a full-blown nuclear >>>explosion. >>> >>>My advice to _everybody_. Take results with a grain of salt, knowing that >>>another event will probably produce a different result. Program A being better >>>than program B is _not_ an absolute thing, ever. The "weaker" program can win >>>three in a row without any outside "help". >>> >>>What's the point in doing the finger-pointing here? If you don't like or don't >>>trust Thorsten's results, ignore them and use the ones you do like and/or trust. >>> >>>It certainly makes the moderators' jobs easier... >>> >>> >>> >>>> >>>>And now folks: Thorsten agreed with me on the judgement. He said that the game >>>>didn't prove anything. >>> >>>If you want to _prove_ something, chess is probably the wrong venue for >>>doing so... >>> >>> >>>> >>>>So far so good. >>>> >>>>But then my examples (I wrote about 5 games! out of 30) showed that Thorsten did >>>>not present his whole data but a pre-selected sample. For instance with 3 wins >>>>of Macheide Rebel against Shredder with the same opening blunder book line. >>>> >>>>Now all I am saying is that this doesn't prove that the style Macheide is >>>>anything because it can't prove it. Because Shredder was lost out of the >>>>opening. And not due to the play of anyone as opponent. >>>> >>>>So, I asked questions, this is what I did. I asked Thorsten how he could present >>>>such data. But he never replied with content. He preferred to fake a medical >>>>expert instead. >>>> >>>>Then yesterday Ed came and tried to defend his "tester". He wrote that everyone >>>>could pre-arrange data. SSDF and all. >>>> >>>>Then I showed that this wasn't possible because people would discover it. >>>> >>>>So far so good. >>>> >>>>Then "AB" explained to me that Thorsten is doing operator/program testings. >>>> >>>>Didn't interest me because we must see what Thorsten's intentions are. If he >>>>wants to find a style, then this is not testing and he can do what he wants. >>>> >>>>But then Thorsten is claiming now that he had posted data from tests and I had >>>>no right to criticise him for pre-arranging the data or wahtever. >>>> >>>>This goes too far now. >>>> >>>>It's one thing to insult "bean-counters" (that is the majority of all testers in >>>>CC) and to pretend to be a creative researcher/"tester" and then to claim that >>>>the chosen very small sample of data (NB that Thorsten tested from May 2002 >>>>until now Sept 2002 and all he had to show were 30 games!!!) should have the >>>>same respect as the many data from the other normal testers. This is impostering >>>>square *3. >>>> >>>>If I had to make my choice I would take all the "insulted as beancounter's >>>>results" data in the world before I would believe that the 30 games from >>>>Thorsten gave a complete overview of Thorsten's practice with Macheide. >>>> >>>>But that is so trivial that Thorsten shoud be able to understand it. >>>> >>>>The point of critic is not that Thorsten is doing things different to the normal >>>>testers, the point is, that Thorsten doesn't show his data and he doesn't >>>>explain either. All what he's trying to do is claiming that the thirty games >>>>proved that Macheide was stronger than the default Rebel. >>>> >>>>And all I did was to prove that Thorsten couldn't succeed with such unsound >>>>data. >>>> >>>>Thorsten tried a final trick. He asked me if I wanted that he should post not >>>>the data he had but data which I would find better, without these strange >>>>contents. >>>> >>>>No, I can only reply this, Thorsten should learn to present data without bias. >>>>When he does autoplay then he should post these games. And not only 30 games >>>>from May until September 2002 and for Shredder three lost games with the >>>>identical opening book blunder! >>>> >>>>I take for granted that all here in CCC have seen Thorsten's cabinet with the >>>>minimum of theree computers. So Thorsten had enough hardware to play many games >>>>in 4 months. Many more than roughly 30! >>>> >>>>What is insulting that is (besides Thorsten's own insulting of many here in CCC) >>>>his insinuation that the so-called beancounters did present useless data and >>>>that he could do it better. But the truth is that nobody knows what Thorsten is >>>>really doing, or what exactly he is presenting and what not. >>>> >>>>But then it's a hundred times better to have the other's data, where everybody >>>>can see what is game one in what match and of how many games. >>>> >>>>If SSDF presented such data Thorsten provided here, Thorsten would run amok. But >>>>since he's something better than just a bean-counter, he has the right to do >>>>what he wants. That is what he's saying. >>>> >>>>When people like me try to differentiate and reflect and ask questions, he comes >>>>and insults again because that is not computerchess in his eyes. God! May CC be >>>>protected against Thorsten, the better than mere bean-counter, who does nothing >>>>a all that could be analysed. And who runs into his insult-mode when someone >>>>tries to judge his practice. >>>> >>>>I close with this summary. Thorsten has all the right to do tunings the way he >>>>wants. But he can't force people to take his little data sample as if it were >>>>the usual data the defamed "beancounters" present. That would be the >>>>impossibility. The squaring of the circle as we say it in German. >>>> >>>>Rolf Tueschen
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.