Author: Robert Hyatt
Date: 06:46:17 12/09/03
Go up one level in this thread
On December 09, 2003 at 07:36:14, Darse Billings wrote: > >I have been asked to contribute my views regarding the Shredder vs >Jonny game in Graz. (I was in Graz during the WCCC, and I've been >involved in similar 3-fold repetition situations in the Computer >Olympiad. FWIW, I have the highest arbiter certification awarded >by the Chess Federation of Canada: National Tournament Director.) > > http://www.chessbase.com/newsdetail.asp?newsid=1335 > >This is an interesting situation, but the ruling was entirely correct. > >The actual circumstances made the decision clear. Anyone who cannot >see this needs to check their logic or their knowledge of the rules. What exactly are you basing this on. I am going by this: "If the operator types in an incorrect move, the TD must be immediately notified. Both clocks will be stopped. The game must then be backed up to the point where the error occured. ..." That seems clear Another rule: "An operator can only (1) type in moves and (2) respond to request from the compute for clock information." How can you claim that an operator refusing to make a claim displayed by the program is "OK"? Arbiter or not, you simply don't know about the rules used in _computer_ vs _computer_ play... This is black and white, _not_ grey... > >The hypothetical issue is more interesting: whether the operator has >the right to decline an opportunity to draw. There is no hypothetical issue. If the program claims something, the operator can _not_ overrule. > >Some people have asserted that the operator does not have that right. >They are wrong. > >Since the operator is given the right to claim a draw on behalf of >the program, the natural corollary is that it is *not obligatory* >for the operator to do so. That is _utter_ _garbage_. > Note that this discretionary privilege >can also lead to a *win* for the operator's program. The operator >is *not* a completely passive entity, nor has that ever been the >case in computer chess competitions. Yes it has. The rules are _clear_. The operator can not even accept a draw. If the program can accept draws, then all is OK. If the program can not accept draws, the operator is only allowed to do so _after_ the TD approves it. Where have you been??? > >The rule in question dates back to a previous era when computer chess >was a friendly competition between gentlemen. If that is no longer >desirable, then the whole process of claiming a draw (as well as >resigning on behalf of the program) must be revisited, and be taken >out of the hands of the operator. It _is_ out of the hands of the operator. It has _always_ been out of the hands of the operator. There has _never_ been any exception made to this policy. And I do mean _never_. > >The exact procedure for claiming a draw by 3-fold repetition is >covered in the FIDE rules. If a program follows those steps, then >the operator has no say in the matter. Most programmers have better >things to do than encoding every niggling detail of the FIDE rules >(which were developed for human players). The program said "this is a 3-fold repetition". The operator _had_ to click OK to continue. From that point forward, it is the operator's responsibility to notify the opponent and the TD. If he fails, he makes a correctable mistake. The onus of the game is _not_ on the operator, it is on the _computer_. To think otherwise is ridiculous. > >Personally, I prefer to allow the programmer to do what he believes >to be right. If I were the arbiter, I would rule accordingly. If a >third party suggested or demanded that a programmer do something he >believes to be less than honourable, I would hope it was a bad joke, >and would dismiss it summarily. Didn't you just say that what happened was OK? And now you are saying it is _not_ OK? > >It is a sad statement that some non-cooperative participants prefer >to use the rules as a weapon, forcing increasingly complex rules to >handle minor quibbles (which is an impossible task in the limit; at >some point judgement and reason must come into play). > The rules are _not_ "increasingly complex". They have always been very simple and easy to follow, with the guiding principle "the computers are playing the game, but since the ICCA is so incredibly inept and refuses to mandate automatic interfaces to eliminate human operators, the required operators are there solely to move pieces, input moves, and operate the clock. Everything else, most particularly any "decision-making" is left up to the programs. This _is_ "the world computer chess championship" not the "world computer+human chess championship." It has _always_ been that way. That's why any operator error is corrected, rather than allowed to stand, to make sure it is the _computers_ that are competing, with _no_ outside influence. >Regardless, the case at hand is clear and unambiguous: Jonny did not >follow the exact steps for claiming a draw, and the operator's choice >to continue the game was legal. Those who have criticized the ICGA >on this matter should rethink their position. No need. However, in light of the very precise computer chess rules, you might rethink yours. > >As a side note, this situation would not have arisen if the programs >were required to use a direct communication protocol, like that used >for Go competitions. We could also dispense with the physical clocks, >leaving the time enforcement (and other technical details, like draw >claims) to a referee program in the middle. This places a greater >burden on the programmer to satisfy the protocol, and I wouldn't >recommend it for friendly events like the Computer Olympiad, but >it is long overdue for the World Computer Chess Championship. There we agree. But it probably won't happen anytime soon... > > - Darse.
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.