ACCA PanAm kibitz rule

[sje]

The ACCA web site says that participants in the upcoming 2009 PanAm event are required to kibitz move selection data (e.g., predicted variation and score).

Will this rule be enforced for all participants and for all rounds?

Will there be a penalty for non-observance?

[bob]

The ACCA web site says that participants in the upcoming 2009 PanAm event are required to kibitz move selection data (e.g., predicted variation and score).

Will this rule be enforced for all participants and for all rounds?

Will there be a penalty for non-observance?

It is probably time for penalty = disqualification. if someone forgets to turn it on and does so when asked, I don't have a problem with that. If an engine or interface can't kibitz, then it should not participate.

One note, I do think it better to "whisper" rather than "kibitiz" analysis. One _could_ use that information in their program if kibitzing is done. And in the past, custom interfaces have had occasional problems with the opponent kibitzing, although winboard/xboard have not had a problem.

[BubbaTough]

It is probably time for penalty = disqualification. if someone forgets to turn it on and does so when asked, I don't have a problem with that. If an engine or interface can't kibitz, then it should not participate.

I am fine with this, assuming minor slip ups (like missing a PV here or there) is forgiven. I don't mean whole games, just a move here or there for engines that get PV from hash which occasionally will lead to slips.

After all, all my hard work using opponent PV to help guide my search will be wasted if all these violations continue

-Sam

[sje]

After all, all my hard work using opponent PV to help guide my search will be wasted if all these violations continue

Likewise my efforts with feeding your program a bogus PV,

[sje]

One note, I do think it better to "whisper" rather than "kibitiz" analysis. One _could_ use that information in their program if kibitzing is done.

Cheating could still be accomplished even with whispering; it would take only a little bit more effort.

[CRoberson]

We have had some issues at every tournament. Most of the issues
fall under " I forgot to turn it on". In that case, it is up to the opponent
as to if he want's the offender to drop connection and restart with kibitz.

We have had some issues with people trying new GUI's and kibitzing
not working, but they get it fixed in a round or two. This predominately
happens to people using UCI protocols.

The best two options for UCI that I've seen are Polyglot and ChessPartner,
however, Aquarium is supposed to hanle it.
I no longer recommend Polyglot as it is too buggy. My current
recommendation for UCI engines are ChessParnter and maybe
Aquarium. I am basing Aquarium recommendation on info
from others. I use ChessPartner myself.

Also, we usually have several complaints that turn out to be the
viewers fault. They complain that the opponent is not kibitzing but
he is. The problem is due to the viewer not having his ICC variables
set correctly.

I handle each issue on individual basis as the complaints come in.
Most get fixed quickly. Some take a round. Invariably, somebody
tries to use Arena which claims it can kibitz, but all that have tried it
fail. So, that have to change interfaces, books and such.

[sje]

I handle each issue on individual basis as the complaints come in.

This is good. But ever better is to avoid most complaints before the event starts. This can be done by requiring some kind of proof by demonstration that an entrant's kibitz facility actually works. Waiting until Round One is waiting too long.

Perhaps a better way of kibitzing would be to use the web page server idea mentioned in another thread. Instead of relying on the ICS and a third party interface, a chess engine could simply output HTML to a web server with the kibitz data. The page would be updated after each move and maybe more frequently. The URL to the page would be provided to the organizer prior to the event and publicly posted for all to see. Human observers would open up one or more browser windows and watch without having to even be registered on the chess server.

The idea can be extended to have an entrant's web page include embedded, real time audio (maybe video, too) to enable author telepresence and discussions.

[bob]

It is probably time for penalty = disqualification. if someone forgets to turn it on and does so when asked, I don't have a problem with that. If an engine or interface can't kibitz, then it should not participate.

I am fine with this, assuming minor slip ups (like missing a PV here or there) is forgiven. I don't mean whole games, just a move here or there for engines that get PV from hash which occasionally will lead to slips.

After all, all my hard work using opponent PV to help guide my search will be wasted if all these violations continue

-Sam

Crafty has a quirk where it will on rare occasions display something like "pv: N/A". That happens when the last move fails high and there is not enough time to resolve the PV, so I have no PV to actually display. It will play the correct move of course... I do not see how someone could fix that since failing high and running out of time is not that uncommon.

[sje]

Crafty has a quirk where it will on rare occasions display something like "pv: N/A". That happens when the last move fails high and there is not enough time to resolve the PV, so I have no PV to actually display. It will play the correct move of course... I do not see how someone could fix that since failing high and running out of time is not that uncommon.

Symbolic keeps track of the last valid PV and expectation. That's what it outputs at the end of the search regardless of the reason for search termination.

Symbolic's PV postings are occasionally only one move long: opening book moves, tablebase moves, and a few other cases. But with each PV post, Symbolic also outputs the reason for search termination so there is no doubt as to the authenticity of the analysis.

[Zach Wegner]

One note, I do think it better to "whisper" rather than "kibitiz" analysis. One _could_ use that information in their program if kibitzing is done. And in the past, custom interfaces have had occasional problems with the opponent kibitzing, although winboard/xboard have not had a problem.

This doesn't work for me. I only have one ICC account, so I chat and see kibitzes with my computer account. I really don't think there is any risk of anybody using kibitz information.