Author: Robert Hyatt
Date: 07:00:56 05/17/00
Go up one level in this thread
On May 17, 2000 at 04:14:27, Adrien Regimbald wrote: >Hello, > >>THis shows you know nothing about how a program plays chess. The program >>_always_ is trying to win. Even when down in material. It is never just >>"trying to hold on". > > >I know nothing about how a program plays chess? I find that statement >particularily amusing as I have made my own program from scratch. I may not >know all of the latest and greatest pruning methods, but I certainly know the >basics. > >I admit I was making a mistake in thinking of how Fritz is playing in terms of a >human disposition. You are also making the same mistake - the program is not >trying to win, it simply picks what it deems is the best move from a given >position. Playing the best move in each position is one definition of "trying to win." > >Resorting to personal attacks is inappropriate. I am not going to continue >debating the issue of Article 10 with you. I didn't state that "a player would >get a draw in that position, end of story". I said that it is likely that they >would get one. You however seem to take an absolute stance saying it would >never happen. Without taking a poll of a number of international arbiters, the >truth can't be settled, as this is a decision that is made individually by the >arbiter. I am certain the draw would not happen. There is no twist of the rule I can see, other than paying off the arbiter, that would cause an arbiter to rule that the game is a draw. Arbiters don't have a lot of flexibility here, as the rules are pretty clear. Which is how it should be... > >We have lost track of the point of the argument in the first place, and rather >than debating the main issue, are nit-picking over side points. > >The main issues (as I see them) are: >1. Was the draw offer an appropriate gesture? >2. What are the overlying consequences of this confrontation? > > >Addressing 1: >- Tiviakov was playing to win, and thus a draw offer should not have been made. >- I don't personally think that the operator was trying to somehow upset >Tiviakov. I think operators should simply be more careful of the idiosyncracies >of chess players. He was about to lose on time. How do you _know_ that he wouldn't accept the draw offer without offering it? How do you _know_ that he isn't worried about the clock? How do you _know_ that he wasn't simply being polite by not offering a draw himself, because he could see the clock and knew he was about to lose? In light of all those points, how can it be wrong to "ease the pressure" and offer _him_ a draw, which he is free to accept or decline, without any loss of concentration or anything... >- I also don't think that the operator technically should be offering draws on >his own. This being said, I think the operator should also consider the impact >of winning a drawn/lost game on time .. there is a very delicate balance here to >be tread - and the onus is on the computer operator to find where the line is - >as it is the GMs who have what the programmers want. You yourself have said on >numerous occasions that you have resigned won games for Crafty to ensure that >you get more of them. Why is it that you follow this policy yourself, but think >it is completely rediculous for Fritz to resign in a LOST game? From the point >of view of learning things from the game (which is what I would hope the motive >for having computers in human tournaments is) - if you have reached a won >position, you have the data you need - your program really doesn't care whether >it wins the game or not. If you have reached a lost position - you really have >nothing at all as a computer operator to resign - you already have the game up >to that point which will show you how your program got in trouble - and if I >understand correctly the programs aren't eligible for prize money - so point >totals should be irrelevant to the programmer. Resigning such games seems like >a good policy to me: a) everyone in the know will recognize the true strength of >the program if it achieved a winning position and b) the GMs will hold less >animosity towards playing a silicon based opponent. Then shouldn't they play _without_ a clock, if the clock isn't going to count when the human gets into time trouble (but of course it would count if the computer got into time trouble). The clock is part of the game, and is included in the rules of chess. This means that in addition to trying to manage your own space, material, positional advantages, you must _also_ manage your time so that you can take advantage of any 'edge' you manage to create in the game. Saying that Fritz should have resigned at -2 is ridiculous. Humans don't do that very often, although I have seen many GM players resign vs computers on ICC when they hang a pawn. But I have seen GMs play on a piece down, hoping to draw, since endings like KRB vs KR are dead draws and the extra piece is of no use. > >Concerning 2: >- this incident (even if everyone were to agree that Fritz' operator was in the >right) has likely already widened the rift between GMs and computers. Possibly. But it is a ridiculous stand to make. It would make _far_ more sense to complain about the clacking keyboard, the whirring fan noise (a quad is not quiet), the green LEDs on the power supplies on the back of the machine, the faint hum coming from the monitor, etc. All are nonsensical points... If humans don't want computers in the events, they ought not be in the events. But the reason should be "because they are computers" and not "because the operator insulted me, or the operator didn't resign when he should have"... >- whether rules are set or not, I think that in the interests of appeasing GMs, >computer operators should follow a set of guidelines when such occasions arise. That is the point. I have been in that _exact_ position many times. I have made crafty offer draws or resign even though it could have won on time. And in every case the opponent said "thanks". Not "hey, this draw should be a win for me because your draw offer interrupted my thinking" or some such nonsense. In 99% of the cases, what Frans did would be applauded as good sportsmanship. This one sour-grapes GM turned good sportsmanship into something else. Or at least he _tried_ to. I think _he_ ought to be banned from future computer events, period. :) >I'm not going to say what these guidelines should be, but GMs should be told >what they are, and the operators should follow them strictly. Then at least the >GMs have idea of what will occur in such situations - for example (if such a >guideline was part of the set), the GM would know that in a dead drawn position >if the GM gets below 5 minutes in a sudden death time control that he can count >on the operator offering a draw (I'm not saying this should be a policy, this is >just an example). > > >I hope that we can all work together on finding a reasonable middle ground on >this issue. Please don't get me wrong - as an author myself, I would love it if >when my program matures enough that I would want to test it against humans that >there will be humans still willing to play a computer in an OTB setting (other >than strong players who would do it as a personal favour to me). > >How about we avoid bickering over small details of everyone's opinions, and >instead work in concert towards this goal! > > >Regards, >Adrien.
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.