2007 World Computer Rapid Chess Championships: Final Results

[hgm]

The Hiarcs operator did a mistake, because he could follow the game and change the contempt after seeing the game becomes drawish. That was taking human influence in the game. But I'm sure a lot of worse stuff happenens during such tournaments, and not only online ones.
I think it was a "small" oversight from Harvey. But apparently generosity is not the peculiarity of the chess programmer.

You correctly remark that Harvey could and should have set the contempt factor before the game. The point is not that he used information from his own game: you have no advantage from changing the contempt factor when it gets drawish, as that is the only situation in which the contempt factor matters.

The point is that, doing it during the game, you can do it in response to the results of your competitors. Before the game you don't know yet if you need a win, or if a draw is sufficient.

Of course you could wonder if it is really desirable that the engine should be kept in the dark about his competitor's results. In a Human tournament the players have access to the board position of the other games. I can imagine, though, that if you want an engine to make use of that situation, we should require that it is _programmed_ to interpreted it, rather than needing the operator digestion of it in the form need-a-win vs need-a-draw.

As for programmer generosity: I think that Harvey made an honest mistake, and that he has recognized that. As far of I am concerned, that should be enough. I would think it would be really sad if he would not be allowed, or would not allow himself, to operator HIARCS in future similar events. Provided, now that he knows the strictness of the rules, he will not do a similar thing again.

I think that what infuriates some critics here is not so much that he has done it, but his early reactions to justify it. Like "so in the future I will not write about it" or "other rules were violated as well" or "others violate the rules as well". This is entirely a wrong response, because it suggests that, even now that he has been informed about the rules, does not see a serious need for abiding by them. While this time it was an honest mistake, the next time it would of course be intentional cheating. But I am entirely satisfied that Harvey has long since abandoned that point of view.

That others are allowed to break the rules is very reprehensible, but a totally different matter.

[Rolf]

Changes during the games should be not allowed, in my oppinion, as I aready stated. Or only after consulting the TD.

Maybe it's a question of speech, so, I am sorry already in advance, but if I read your statement right the message you offer is wrong.

It is wrong to hypostate that the possibility of change during a running game could still be discussed. It cannot. It was decided 37 years ago, a damned long time to diffuse into the common memory of the community.

Therefore it was also very wrong what had happened in Graz with the assistance of the arbiter. A threefold repetition is a draw. Now it is irrelevant what a friendly opponent says. If he might offer a continuation of the game, so that the "normal" or "expected" result came out, so that then the "stronger" program, that should even win the whole tournament, gets its suited result, it's a factual cheating, nothing else. In special this is influencing the tournament issues of many other programs. Which is the main reason why this is called cheating. Otherwise two opponents could do what they wanted if they are in a friendly relationship. They could do with their game what they wanted. However this consists cheating in all thinkable sports.

It was hgm who led my attention into a different direction than what I had seen still yesterday. I was convinced that Harvey had acted in emotional stress when suddenly his wished "tie" disappeared in the distance if the draw would become factual in his game. hgm's version is more convincing.
Depending of the outcome of RYBKA's last game, HIARCS could have drawn or must have won. After RYBKA had won, HIARCS had to win too. With the pre-setting this was almost impossible in a drawish position. So, Harvey changed contempt factor in cold blood. Now I understand Uri much better who already claimed that this decision had impact on the following progs in the final ranking. Here we have the aspect again that some might think that nothing really matters than the first or second place. But that offers a different contempt factor.

Remember. In Graz it was different. Only through the intentional continuation of a drawn game, what could only happen through a wrong programming of SHREDDER, because his position was clearly better, SHREDDER got the tie against FRITZ that he won and hence became Wch.

All this is an offense against the well defined spirit of computerchess. Human intervention is neccessary as far as no automatic activities by the program itself are available. But this means the technical side of the game, not the chess related outcome. If Harvey is correct with his statement that human intervention is done in the chess direction then this should be researched. If this is possible without that it's directly discovered, we must find automatic tools to prevent such cheating. If for instance a re-booting is happening because then the program is able to continue the game with a different setting, this can only mean the direct loss of the game.

So, opposed to the meant subject line, I would want that the debate about Harvey should be extended to a clarification how the eng-eng games could be protected against human intervention. Actually I see a trend towards a slow adaption of intentional cheating tricks, which is completely against the historical traditions of computerchess. It is also beyond me why anyone could participate in a tournament and imagine a game that would be paractically irrelevant to the rest of the participants so that any form of at the instant interventions by operating personal could be condoned. Is there no fair-play basic conviction anymore. In the end it seems so that the critics themselves must excuse their criticism because they could have hurt the sensitive spots of someone who has terribly gone wrong. That should be thought over despite a golden rule that no matter how gross the mistake was we dont exclude or exterminate anyone. But it's not up to the one who failed to threaten that he might leave if the critics would become too intensive to bear...

[CRoberson]

As ACCA President, I can safely say that changings parameters
during a game is not allowed. This is a general statement in that
it could be allowed in certain conditions:
If your engine crashes and you can't get it up without a parameter
change then that is fine.

Other than that, ACCA tournaments are fully automated. The only
interventions allowed fall under the context of engine crashes.

You are allowed to change parameters or even engine versions
between matches but not during them.

That said, I see don't see any reason to take any action against the Hiarcs team
as it would not have made any difference in the tournament results.

[gerold]

As ACCA President, I can safely say that changings parameters
during a game is not allowed. This is a general statement in that
it could be allowed in certain conditions:
If your engine crashes and you can't get it up without a parameter
change then that is fine.

Other than that, ACCA tournaments are fully automated. The only
interventions allowed fall under the context of engine crashes.

You are allowed to change parameters or even engine versions
between matches but not during them.

That said, I see don't see any reason to take any action against the Hiarcs team
as it would not have made any difference in the tournament results.

If they broke the rules should they be barred from the next
tournament.

[Rolf]

That said, I see don't see any reason to take any action against the Hiarcs team as it would not have made any difference in the tournament results.

That stands against what Uri said. Of course you were focussing on the two first places, right?

[Guetti]

As ACCA President, I can safely say that changings parameters
during a game is not allowed. This is a general statement in that
it could be allowed in certain conditions:
If your engine crashes and you can't get it up without a parameter
change then that is fine.

Other than that, ACCA tournaments are fully automated. The only
interventions allowed fall under the context of engine crashes.

You are allowed to change parameters or even engine versions
between matches but not during them.

That said, I see don't see any reason to take any action against the Hiarcs team
as it would not have made any difference in the tournament results.

If they broke the rules should they be barred from the next
tournament.

You can't exclude some teams and others not for breaking the rules. Crafty (and Rybka) broke rule 7. that is clearly stated on the ACCA website. But it's a bit harsh to exclude somebody for not kibitzing. You could say breaking rule 6 is more objectionable than breaking rule 7., but rules are rules.

[sje]

If there is to be a lesson learned from this event, it's that the rules should be clearly stated in advance along with the sure and most certain consequences of a failure to adhere to said rules.

--------

I learned programming back in the Old Days when it was all done on paper punch cards. And the cheap ass university where I was subjected to this for the first segment of my undergraduate career charged quite a bit for blank cards, although they did let students use the keypunch machines for free. Well, those keypunches and cards were a Write-Once memory system with no "delete" key -- make a mistake and the card had to be tossed. Unsurprisingly, people soon learned not to make mistakes.

If a breach of the rules in a computer chess event was immediately and unquestionably rewarded with a forfeit, of the game or of the entire event, then authors and operators would also soon learn not to make mistakes.

[frankp]

If there is to be a lesson learned from this event, it's that the rules should be clearly stated in advance along with the sure and most certain consequences of a failure to adhere to said rules.
<snip>
.

I think this is the most important rule - what happens when a rule is broken. Upfront, explicit, transparent and .... obvious.

[hgm]

You can't exclude some teams and others not for breaking the rules. Crafty (and Rybka) broke rule 7. that is clearly stated on the ACCA website. But it's a bit harsh to exclude somebody for not kibitzing. You could say breaking rule 6 is more objectionable than breaking rule 7., but rules are rules.

I don't agree.

Actually, I think that the fact that some engines systematically and consistetly refuse to implement Kibitz, shows a blatant disregard for the rules that is far, far worse than what the HIARCS team did. They obviously have no intention at all to comply with the rules. While I agree that the rules were not so clearly spelled out, the requirement to Kibitz was one of the rules that was widely advertized from the beginneing. (It would not hurt, for the benefit of novices, to distribute a mail with rules, plus a short overview of how to operate ICC during such a tournament. This "tell pear games" business was far from obvious.)

Not that I want to make a clear distinction here between programs that did not Kibitz because they merely forgot to switch it on, or directed the output to an unusual place, and those that simply did not implement it at all. If someone makes a mistake, it is no reason to forfeit a game. The TD can simply notify the operator of his mistake, and order him to switch the Kibitz on. If there are compelling technical reasons whythis cannot be done during a game, at least the offending side should be required to have proper parameter settings for the next game. But if not, there no longer is an excuse. So after the first warning, all following games should be forfeited on the bases of refusing to kibitz!

If we don't want that, the requirement to Kibitz should be immediately removed from the rules. As having rules that can be broken with impunity makes a joke of the tournament!

[Dan Honeycutt]

If there is to be a lesson learned from this event, it's that the rules should be clearly stated in advance along with the sure and most certain consequences of a failure to adhere to said rules.

--------

I learned programming back in the Old Days when it was all done on paper punch cards. And the cheap ass university where I was subjected to this for the first segment of my undergraduate career charged quite a bit for blank cards, although they did let students use the keypunch machines for free. Well, those keypunches and cards were a Write-Once memory system with no "delete" key -- make a mistake and the card had to be tossed. Unsurprisingly, people soon learned not to make mistakes.

I, too, date back to the punch card days but, unlike you, never learned the art of not making mistakes. Well done.

Best
Dan H.