2007 World Computer Rapid Chess Championships: Final Results

[hgm]

Well, as we are all good friends again, I still want to defend Uri a little. His wording might have been a little abrasive, especially as I think I know Harvey well enough to be sure that he acted in good faith, but making such adjustments is indeed not allowed. The TD even explicitly pointed out (in reaction to a question) that it was not even allowed to offer or accept a draw on behalf of your engine. Tournaments like this are supposed to be fully automated: you are supposed to start your engine, and from then on it is fully on its own, and you are not allowed to intervene in any way.

(This seems a bit silly in the case of a KRKR end-game where it is obvious that none of the engines is going to blunder away his rook, no matter how little time it has left. But you have to draw a clear line somewhere. Otherwise things will deteriorate on a continuous scale, because there is always an intervention that is just a little bit worse, and you would indeed end up with people "coaching" their engines by telling them where to attack, vetoing moves that are strategically bad in a way they know the engine cannot see (e.g. trapped pieces), etc.)

As for the book: I don't really believe in books at all. Opening theory never interested me much, even when I still played as a Human. So Joker simply plays without a book, thinking up its own moves from move one. So I am 100% sure Joker can never 'get' a bad line, he will play only what he likes. And by sufficiently randomizing moves, it is not easy for opponents to 'book' Joker in advance. I might still force the first move by randomly picking from a table, to get somewhat more variability. One could call that a 1-move book.

I guess you were indeed unlucky then, if a line looks good at 10 sec, but bad at 1 min.

[Harvey Williamson]

Well, as we are all good friends again, I still want to defend Uri a little. His wording might have been a little abrasive, especially as I think I know Harvey well enough to be sure that he acted in good faith, but making such adjustments is indeed not allowed. The TD even explicitly pointed out (in reaction to a question) that it was not even allowed to offer or accept a draw on behalf of your engine. Tournaments like this are supposed to be fully automated: you are supposed to start your engine, and from then on it is fully on its own, and you are not allowed to intervene in any way.

(This seems a bit silly in the case of a KRKR end-game where it is obvious that none of the engines is going to blunder away his rook, no matter how little time it has left. But you have to draw a clear line somewhere. Otherwise things will deteriorate on a continuous scale, because there is always an intervention that is just a little bit worse, and you would indeed end up with people "coaching" their engines by telling them where to attack, vetoing moves that are strategically bad in a way they know the engine cannot see (e.g. trapped pieces), etc.)

As for the book: I don't really believe in books at all. Opening theory never interested me much, even when I still played as a Human. So Joker simply plays without a book, thinking up its own moves from move one. So I am 100% sure Joker can never 'get' a bad line, he will play only what he likes. And by sufficiently randomizing moves, it is not easy for opponents to 'book' Joker in advance. I might still force the first move by randomly picking from a table, to get somewhat more variability. One could call that a 1-move book.

I guess you were indeed unlucky then, if a line looks good at 10 sec, but bad at 1 min.

Hi,

You are probably right although I can find no specific rule against it. Indeed in some tournaments it maybe wrong to change things but I don't mind that. In fact the 1st time I met you at Leiden I enjoyed hearing about the changes you had made to Joker each day!

This change I made was very minor and one the software seems to allow with no warnings so I am sure others also do this.

Other rules were broken over the weekend - the late entry of Zappa and Rybka being alowed to play although it can not kibitz - as you pointed out Micro-max can! I did not ask for sanctions against vas or erdo for their small breaking of the rules. If my infraction was wrong then it cost me half a point and serves me right!

Best Wishes,

Harvey

[George Tsavdaris]

Hi Harvey,

I think that it is a general rule in computer chess not to allow changing setting during the game.

Without that rule you have not computer chess but type of advanced chess.

I agree.

If Harvey is correct(and i guess he is) that when you change an engine's parameter while it is thinking and press OK to apply the setting, then the engine plays immediately, then we have:

Imagine the consequences from a programmer to be able to
interfere during his engine's game:

-He is able to force a move for his engine.
If for example the programmer wants his engine not to spend more time in searching an obvious move he just opens the parameters change or not change something, press OK and the engine plays the move it is thinking.
Another example is when the engine is between 2 moves but only one is good(it happens many times) and the programmer knows this, so when the engine shows as the best move in the PV, the really better move, then the programmer does the same trick to force his engine to play this best move.

-He can program settings like "play more attacking now", "play for a draw now", "play more slowly now" in his engine and then during the game when he learns for example that the engine which fights for the first place with his engine loses and a draw for his engine is good enough to win the tournament he may change the setting to "play for draw now" so the draw will come easier.
Or if he learns that he needs to win for winning the tournament he may set the "play more attacking now" during the game. Or if he sees that his engine plays a bit slowly he may set "play faster" etc.....

There is a huge number of things/parameters/tricks one can do to help his engine, if the intervention between the programmer is allowed.....

That makes it advanced Chess and not engine-engine Chess.....

[Harvey Williamson]

Hi Harvey,

I think that it is a general rule in computer chess not to allow changing setting during the game.

Without that rule you have not computer chess but type of advanced chess.

I agree.

If Harvey is correct(and i guess he is) that when you change an engine's parameter while it is thinking and press OK to apply the setting, then the engine plays immediately, then we have:

Imagine the consequences from a programmer to be able to
interfere during his engine's game:

-He is able to force a move for his engine.
If for example the programmer wants his engine not to spend more time in searching an obvious move he just opens the parameters change or not change something, press OK and the engine plays the move it is thinking.
Another example is when the engine is between 2 moves but only one is good(it happens many times) and the programmer knows this, so when the engine shows as the best move in the PV, the really better move, then the programmer does the same trick to force his engine to play this best move.

-He can program settings like "play more attacking now", "play for a draw now", "play more slowly now" in his engine and then during the game when he learns for example that the engine which fights for the first place with his engine loses and a draw for his engine is good enough to win the tournament he may change the setting to "play for draw now" so the draw will come easier.
Or if he learns that he needs to win for winning the tournament he may set the "play more attacking now" during the game. Or if he sees that his engine plays a bit slowly he may set "play faster" etc.....

There is a huge number of things/parameters/tricks one can do to help his engine, if the intervention between the programmer is allowed.....

That makes it advanced Chess and not engine-engine Chess.....

The lesson I have learned here is after a tournament whatever the result just to keep quiet like all the other programers do.

Attempting to give some insight as to what was actually happening at the time is a waste of time - in future when we score 12/14 I will keep quiet.

[George Tsavdaris]

The lesson I have learned here is after a tournament whatever the result just to keep quiet like all the other programers do.

But why? I don't understand.....

And by the way all the things i've said in the previous post are not OF COURSE things i believe you or anyone else did in this tournament.
I just said what it is possible to be done when there is the option to interfere while an engine is playing.....

[Mike S.]

I can assure you that it WILL BE greatly improved.

Yeah! I must say, usually I am interested in the strongest engines only, and in freeware which is quickly rising to the very top, but microMax is an exception. I find it is fascinating. So much power in so small code. I think I am already a fan of it. 6.0/14 in a major tournament... go microMax go!

[frankp]

in my cases I believe the operator was simply trying to do the best for the commercial program - he was not the programmer - and did not set out to cheat at all. I have no reason to suggest this was not the case, in this case. And certainly believe that Mark Uniacke would not countenance (and does not need) improper external influence on his behalf.

Often it is the operators rather than the programmers who 'interfere' in my very limited experience of playing in these events - no doubt well intentioned and with no intention of cheating. We have had endless debate about the extend of interference that should be allowed (program locks, internet connection failure), whether only programmers should be allowed to operate, kibitzing lines, automatic operation only etc etc. In the end you have rely on those present.

[edited]

As for the time line of the posting I am afraid I did not read Harvey's post but was responding to Uri's.

[VP]

The lesson I have learned here is after a tournament whatever the result just to keep quiet like all the other programers do.

Attempting to give some insight as to what was actually happening at the time is a waste of time - in future when we score 12/14 I will keep quiet.

Hi Harvey,
Congrats on the wonderful performance of Hiarcs. I think it is one of the few engines that has been striving hard to keep up with Rybka and is being successful.
And thanks for sharing your insights. It is always fascinating to know what the operators are thinking.
To others, maybe in future, it is a good idea to have the rules clearly laid out before the tournament- maybe no user intervention at all - or allow operators to accept/ reject draws and modify the contempt factor etc.
Let there be no gray areas. I think before modifying the contempt factor, the TD should have been approached- who would have most likely rejected the proposal. Good thing is, it did not alter the final outcome.
A well deserved second place for Hiarcs.
Cheers,

[Steve B]

As for the time line of the posting I am afraid I did not read Harvey's post but was responding to Uri's.

ok Frank
the timing of your post was all i objected to

if you didnt read it then there is nothing further for us to discuss

Steve

[hgm]

Well, we could of course still discuss if it is a good idea, in cases where top programs break the rules in order to enlarge their advantage over their opponents, to deny others the right to object against this, just because their programs are mediocre...