Author: Ossie Weiner
Date: 11:58:40 09/23/98
Hi Bruce [Moreland], <<< Since this was sent to me, I'll feel free to respond to it. I'm not sure how it came about, but it seems like you guys all ended up supporting Auto232 at about the same time. You all could test against each other, and nobody had the advantage of being able to autotest against people who couldn't autotest back. I'm not sure how this fragile balance was maintained, but perhaps it was because everyone wanted to make sure that their thing was testable by the Swedish folks, who want to do testing via Auto232. >> So far, so good. << But now Fritz 5 has upset this balance, they have cut Auto232 support. And Ossi and Ed would seem to be mad because Fritz 5 can test against their stuff but not vice versa, and also there is concern that the unreleased autoplayer might be doing some funny stuff (if I am wrong about their motivations, then please substitute the phrase "Bruce, assuming he ever creates a product" for "Ossi and Ed"). And rather than saying, "no, we won't put up with this funky autoplayer", the Swedish folks have accepted this thing. >> Well spotted! The Swedish folks made a mistake by accepting this (devlish?) thing, without thinking thoroughly about the consequences. Now we have a big mess, most of us have lost our trust in their rating procedure. << You each have to choose from between three main courses, one with two variants: 1) Support Auto232. 2) Provide your own autoplayer to the SSDF, or provide an Auto232 version to the SSDF but not to customers and hence not to competitors. 3) Cut all autoplayer support ... a) ... and get mad at the SSDF. b) ... and not get mad at SSDF. I think that Ossi and Ed would like the SSDF to try to force people to do #1, and it sounds like Thoralf doesn't want to be in this position. >> Wrong! We don't want to force anybody to do anything. That's not the point. What we do want is *EQUAL CHANCES* for everybody. Everybody knows that it's technically possible to cheat with a secret, non-standard autoplayer, simply because there is no control. Fair competition can only be secured if *all competitors play under the same rules*, and follow the same standards. No exceptions, no privileges! I don't mind if everybody agrees on Auto232 or any other standard (yet to be defined). Anyhow it can't be tolerated that one particular competitor has the privilege of defining his own rules. That's just unfair, immoral and leads to totally worthless results. << It's true, ChessBase could just as easily gone for course #3b, which would still have gotten Fritz tested, but resulted in more tired people in Sweden. >> .... and certainly not in a top spot on the SSDF rating list .... << I think the only stable course is #2. Otherwise, even if Fritz comes back into line and supports Auto232 in its commercial version, someone else is going to come along who doesn't. Perhaps Fritz should support Auto232 but not release this version. That way, they still prevent others from autoplaying against them, but there is no question about whether the unreleased autoplayer is bogus or non-standard or a creation of Satan or whatever. Then the rest of you can do this as well, and while it is true that none of you can autoplay against the others, at least you are all equally blind, and there are no tired Swedes. >> That makes sense, Bruce. There is no logical reason not to support Auto232, other than going for an (undeserved) advantage. Secret autoplayer stink ... << There is still the problem of you guys pirating each other's Auto232 version, but doesn't this problem already exist with the private autoplayer version of Fritz? >> That's of lower relevance. Much more important is getting the smell of dishonesty out of this business. Best regards Ossi Weiner **************************************************************************** From: Thoralf Karlsson[SMTP:thoralf.karlsson@mailbox.swipnet.se] Sent: Friday, September 18, 1998 11:13 AM To: Ossi Weiner Cc: Chrilly Donninger; Eric Hallsworth; Marty Hirsch; Komputer Korner; Richard Lang; David Levy; Rainer Menningen; Stefan Meyer-Kahlen; Bruce Moreland; Peter Schreiner; Ed Schroder; John Stanback; M. Uniacke; Mark Uniacke; Helmut Weigel Subject: SV: SSDF Rating List << *We* don't give Chessbase the advantage. They have taken it themselves. >> For heaven's sake - what's the difference?? Thoralf Karlsson publically admits that Chessbase has an (unjustified) advantage. That means everybody else has a disadvantage ...... But why is the SSDF tolerating this unfair situation? Why are they knee-falling from secret, hidden, non-available, supposedly cheating autoplayers? What have they done to keep fair competition under control? << We examine how good the chess programs play against each other. If now Fritz 5 has become stronger against other chess programs thanks to the fact that they more easily and cheaper can play games against their competitors, why should we try to hide it for the customers?? >> Wrong! Fritz 5 has *not become stronger* against other chess programs, it's just *scoring more points* (supported by a secret, hidden, non-available, supposedly cheating autoplayer) in Sweden. That's quite a big difference, in fact it looks more like "simulated plying strength". If F5 would be nearly as strong as the unconvincing SSDF results, it should be able to prove it also without "secret weapon". Shame that it never did ...... << And in connection with this argument concerning that it's easier and cheaper for ChessBase to play against the competitors than the reverse, I don't see why it would be better for SSDF to play manual games. If we accept that there is no inherent cheating in the autoplayer, the result would be the same anyway. The only difference would be that it would take so much longer for SSDF to reach 100 games. >> Wrong again! How can Thoralf assume that there is "no inherent cheating" in the autoplayer? What has the SSDF done to make that theory sure? The F5 results in manual games could *never* be the same, more likely they would be closer to Eric Hallsworth's hand-tested results, possibly somewhere around 5th place with about 60 - 80 Elo less than in Sweden. For the customers (to whom Thoralf is referring) the Swedish results have become most deceiving, if not practically worthless. The SSDF was supposed to test the strongest chess programs - instead they are now comparing the "smartest" autoplayers. What is that good for? **************************************************************************** On Sept. 19, 1998 Ed Schroeder wrote: << Until you come up with good answers your results are suspect. Why put your list in such a position? Here is my advice to SSDF: - Do not allow secret / hidden autoplayers no one can check. - Allow only public available autoplayers. - If a program has no (public) autoplayer test manually. >> 100% agreed! Secret / hidden autoplayers are not only unfair competition. As any secret device is uncontrollable by definition, it will always be a *permanent source of potential cheating*. Such results are suspect indeed. How can anybody demand from a professional programmer - who makes a living based on the reputation of his software products - to accept such crazy conditions? Wouldn't it be rather silly for any programmer to enter such an unbalanced competition, as long as the SSDF fails to define clear rules and make sure that they are fair and equal for all competitors? Best regards Ossi Weiner
This page took 0.01 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.