Author: Bruce Moreland
Date: 01:01:59 04/06/00
Go up one level in this thread
On April 06, 2000 at 02:01:18, KarinsDad wrote: >On April 05, 2000 at 18:28:14, Bruce Moreland wrote: > >>It is pointless to argue with you KD, you drag every aspect of this into minutia >>or non-sequitor. > > >Ok, I won't argue. I'll ask questions instead. > > >> >>I am arguing from a very practical point of view, as someone who competes in >>person at live computer chess events. In case anyone is listening, here is a >>summary of my arguments: >> >>1) I think that the availabilty of 5- and 6-man tables is a ridiculous reason to >>change the laws of chess as they pertain to computer vs computer play. > > >So, is it your belief that a rule like the 50 move rule that was added just in >order to not drag out a drawn game for humans who get tired or humans who want >to attempt to weasle their opponents into making a mistake should not be >re-examined in relationship to computers who a) do not get tired and b) can have >EBTB endings that require more than 50 moves per side for the win? The live ICCA tournaments are run basically according to FIDE rules involving blind players. There is a human intermediary who serves as operator. That person gets tired, and the rounds are scheduled and need to start on time. Using sudden death time controls solves part of this problem, but I think brings the operator into the game a little too much, since operator speed would have a more definite effect upon the game than it currently does. If you subtract out the human operator, there is not a lot of reason to have the 50-move rule at all, other than that the human game has it, and it is not necessary, certainly, and probably isn't desirable to have a seperate game for humans and computers. It's my goal to build a program that plays either computers or humans, and I'd rather not have to have some boolean that I set depending upon which rules are in force. > But if >>this is to be done, the specific endings that may be excepted need to be stated >>well in advance of any tournament. > > >So, if we have exceptions to the 50 move rule, you want each and every one >spelled out ahead of time as opposed to just dropping the rule, even though >there doesn't seem much reason to have it in computer/computer games except for >human historical tradition (and the fact that it is already coded into a lot of >programs)? If the rule is dropped altogether, it's different than having someone show up at the site with a KRR vs KBB table, making an unverifiable assertion about the optimax for this ending, and asking for an exception to the FIDE rule. I don't favor dropping the rule in whole or in part, although the ramifications are different depending upon how much of the argument I lose. I just don't want to lose the argument on site, at a tournament. People should be given the opportunity to know what game they are trying to program, and changing the rules on-site doesn't allow people time to prepare. We've had this same problem at ICCA events in other cases, for instance when trying to deal with the issue of operator time, and when you can set your computer's internal clock. It's hard to make a change on-site, because suddenly a program that was conformant with the old rule is not conformant with the new rule, and some other guy who was doing something non-conformant is now conformant. This isn't fair. > It should not be possible to argue that >>since I show up with a weird 7-man table, everyone should make allowance for me >>since I still want to win any 300-move mates I might happen to steer for. If I >>am going to argue that this ending should be excepted, anyone else should be >>able to argue that they need ample time to handle the ending specifically. >>Personally, I think this is no fun and generally stupid. I've operated with >>5-man tables for several years and I've never felt the need to try to bother my >>opponent with such esoterica, if I had ever lost a half-point due to the 50-move >>rule, I would have congratulated my opponent and added the event to my >>collection of stupid computer chess stories. From the point of view of someone >>who has the tables, this is all esoterica, but when people who have the tables >>try to make others change their programs in order to allow for these weird >>cases, that's more serious. > > >You have mentioned several times that you think that a reason that people should >not consider rules changes is that it would force computer programers to modify >their code. Is this why you appear to be so strongly against examining the >differences between programs and humans and coming up with a set of rules for >human/computer and computer/computer games that may be different than >human/human rules? You don't want to change the code? I don't want to change the code on-site, certainly. If I show up at a tournament and someone demands that the FIDE rules by amended for computer play, and that the changes be in force during that tournament, I'm going to balk big-time, which is a sensible thing to do. I don't like the idea of changing the rule beforehand, either, because I think the change is bad. I don't think you are going to get the ICCA to agree to drop the 50-move rule. If you do, I think it would be a bad thing, since you've effectively created another dialect of chess, another distinction that would cause it to be difficult to have humans play against computers. A great many of these programs are designed to play any opponent, human or computer, and I think that it makes most sense to have one set of rules. The humans will not change their set of rules. I think it makes most sense to use the human set of rules. And no, this has nothing to do with your bullshit argument (hey, you called something I said, "tripe") about computers being non-conformant because they can't press the clock with their finger. As I've said previously, computers operate according to the "blind" rules, which include an operator. >Also, advances in technology have allowed the addition of elements of chess like >Fischer time controls. Why would a time control change be acceptable as a new >way to play chess whilst the removal of a human-centric rule like the 50 move >rule not be consider acceptable for computer/computer games? A significant on-site time control change is not acceptable. People should come and expect to deal with some small variation, but if you go 3000 miles expecting to play 40/2, you shouldn't have to play a Fischer time control. There is a very good chance that you haven't even coded it, if your program is new. Fischer time controls, as a concept, are a little revolutionary but are perfectly fine, in my opinion. >>2) I disagree with the stated idea that one program should be awarded a point in >>a given position, while another one must take a half-point in the same position, >>based upon what the author says about how the program plays chess. > > >I am not quite sure what you are referring to here. Could you please elaborate? One of the proposed solutions to the > 50 move mate bug in endgame tables, is to allow extension of the 50-move rule if the program is announcing mate. If you have tables, and play into a mate in 80, 50 moves in you are showing mate in 30 and are allowed to continue. If you don't have tables, and play into a mate in 80, 50 moves later the game would be declared drawn, even if the exact same moves were played. Someone might argue that the program without tables is unlikely to play perfectly efficiently, but I can point to cases where programs with tables don't play perfectly efficiently either. A program that uses distance to conversion tables isn't playing for the shortest mate. It would be difficult to prove or disprove this on-site, but it is certainly possible that this situation could happen. So poof, the Thompson tables are no longer allowed. You can also poof the Nalimov and Edwards tables, by pointing out that a program that is minimizing distance to mate may not be minimizing distance to conversion. It is theoretically possible that a mate in 60 may involve a path that is more than 50 moves without a non-reversible move, but could be executed in few more moves (or even the same number of moves), without involving such a path. So it's not even completely certain that endgame tables are the "word of god". I find this whole topic repugnant. People who didn't do any of the work involved in creating a table generator are trying to change the rules of chess in order to remove any possibility that their perfect play engine, which they got for free and at practically no effort, could cost them a half point if it messes up. I wrote my own, which I have been using since 1994, and I never had to nerve to make this argument. bruce
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.