Life In 19x19
http://lifein19x19.com/

PGETC 2018/9
http://lifein19x19.com/viewtopic.php?f=47&t=16039
Page 1 of 4

Author:  Javaness2 [ Wed Sep 05, 2018 9:50 am ]
Post subject:  PGETC 2018/9

Is there any news on this year's season of the PGETC, specifically
We seem to be missing a manager for leagues C and D?
Is the event going to be rated or not?
What will happen to the play-off stages, will the all be at the EGC?

Author:  KKP [ Tue Sep 18, 2018 12:15 am ]
Post subject:  Re: PGETC 2018/9

When will PGETC 2018/9 start ?
PGETC started in 2017 at 26th of september, in 2016 at 20th of september, in 2015 15th of september, in 2014 9th of september.
Every time in september. It is time to start this year 18th or 25th of september. But, there is not informattion in https://pandanet-igs.com/communities/euroteamchamps

Author:  Javaness2 [ Wed Sep 19, 2018 10:58 am ]
Post subject:  Re: PGETC 2018/9

KKP wrote:
When will PGETC 2018/9 start ?
PGETC started in 2017 at 26th of september, in 2016 at 20th of september, in 2015 15th of september, in 2014 9th of september.
Every time in september. It is time to start this year 18th or 25th of september. But, there is not informattion in https://pandanet-igs.com/communities/euroteamchamps


It seems that it will start 'soon', we can guess that means in October. .

Author:  Vesa [ Wed Sep 19, 2018 8:07 pm ]
Post subject:  Re: PGETC 2018/9

Hello,

Sorry about the delay, but the announcement is now up https://pandanet-igs.com/communities/euroteamchamps/425:

The 9th Pandanet Go European Team Championship will start soon. We expect to have 37 teams, League A with 10 teams, League B with 10, League C with 8 and League D with 9.

League A: Russia, France, Ukraine, Poland, Romania, Czechia, Israel, Hungary, Italy and Germany.
League A Manager is Flavien Aubelle.

League B: Serbia, UK, Netherlands, Austria, Switzerland, Norway, Sweden, Turkey, Denmark and Belgium
League B Manager is Michal Zubalik.

League C: Lithuania, Finland, Slovakia, Spain, South Africa, Croatia, Slovenia and Greece
League C Manager is Amir Fragman.

League D: Georgia, Bulgaria, Portugal, Cyprus, Belarus, Ireland, Morocco, Kazakhstan and Iceland
League D Manager is Janez Janza.

The schedule will be sent out 20/21 September and the the leagues will start in the beginning of October (2 Oct).

The teams are requested to form their team by 25 September, and each player is requested to send an acknowledgement that they follow the rule of fair play and don't use any outside help (computer or any other assistant) in their PGETC games and accept the decisions announced by the PGETC Arbitration Board.

Cheers,
Vesa

Author:  Uberdude [ Wed Sep 19, 2018 8:43 pm ]
Post subject:  Re: PGETC 2018/9

Vesa wrote:
and each player is requested to send an acknowledgement that they follow the rule of fair play and don't use any outside help (computer or any other assistant) in their PGETC games and accept the decisions announced by the PGETC Arbitration Board.

I'm not going to do this without seeing some evidence of the pro panel's accuracy as Leela detectors. The Carlo Metta case has shown the importance of an appeal procedure against badly constructed prosecution cases and initial decisions, so scrapping this and "we will get it right first time and you cannot disagree" is a step backwards not forwards in justice.

Author:  Javaness2 [ Thu Sep 20, 2018 12:10 am ]
Post subject:  Re: PGETC 2018/9

Thanks Vesa - Can you confirm if the pgetc leagues are no longer rated?

Author:  Javaness2 [ Thu Sep 20, 2018 1:43 am ]
Post subject:  Re: PGETC 2018/9

Uberdude wrote:
Vesa wrote:
and each player is requested to send an acknowledgement that they follow the rule of fair play and don't use any outside help (computer or any other assistant) in their PGETC games and accept the decisions announced by the PGETC Arbitration Board.

I'm not going to do this without seeing some evidence of the pro panel's accuracy as Leela detectors. The Carlo Metta case has shown the importance of an appeal procedure against badly constructed prosecution cases and initial decisions, so scrapping this and "we will get it right first time and you cannot disagree" is a step backwards not forwards in justice.


What is the pro panel, and who will assess the pro panel's accuracy?
Given that the 'fair play statement' is only a formality, I don't really see why you care about it. It is not as if we demand that the EGF appeals procedure is regulated properly.

Author:  John Fairbairn [ Thu Sep 20, 2018 3:20 am ]
Post subject:  Re: PGETC 2018/9

Quote:
Given that the 'fair play statement' is only a formality, I don't really see why you care about it. It is not as if we demand that the EGF appeals procedure is regulated properly.


Is it only a formality? It seems to put people between a rock an a hard place.

You can sign up, play, and be falsely accused and disqualified under an improperly regulated procedure, with no recourse to recover your reputation.

Or you can refuse to sign up and accept the possible stigma that you have something to hide.

I don't really know what the solution is, but I'd have thought that at the very least it must include proper regulation of a properly qualified panel. If that is impossible to achieve with current technology or in the time available, well, tough. Two wrongs don't make a right.

As I understand it, Carlo Metta still stands in limbo, with possibly a reputation in tatters because of an improperly regulated procedure. I don't know whether he cheated or not, but the PGETC people still don't either.

I still have every sympathy for the organisers, though. It's a depressing issue all round. The AI bots have been a classic example of "be careful you don't get what you wish for."

FWIW, if I was uberdude, I'd take the risk in signing up. I don't know him personally but from what he's said here, I'd back his reputation over any PGETC accusation and I'm sure others here would. But I'd also expect that in practice the PGETC people would be so apprehensive about the fallout that they would be unlikely to risk acting on any but the most ridiculously egregious cases.

Author:  Javaness2 [ Thu Sep 20, 2018 4:03 am ]
Post subject:  Re: PGETC 2018/9

As I imagine it...
Last year the process was, claim, judge, if appeal judge again.
This year the process is probably still, claim, judge, if appeal judge again.

Now perhaps my imagination misleads me, but in my head nothing has changed beyond the fact that we are asked at the start to promise not to cheat. Personally I will still take the risk to play and be accused of cheating.

Author:  Uberdude [ Thu Sep 20, 2018 5:43 am ]
Post subject:  Re: PGETC 2018/9

Javaness2 wrote:
What is the pro panel, and who will assess the pro panel's accuracy?

According to an email to team captains (but I've not seen it publicly announced) the pro panel is Yoon Youngsun 8p, Zhao Baolong 2p, and Antti Tormanen 1p. To pre-accept their conclusions a blind test on 100 games with a false positive rate under 2% should satisfy me (a higher false negative rate would be acceptable). I could help with that test. As a warm up they could demonstrate their ability to judge player strength from moves with my little 4 or 6 dan test at viewtopic.php?f=10&t=15832.

Javaness2 wrote:
Given that the 'fair play statement' is only a formality, I don't really see why you care about it. It is not as if we demand that the EGF appeals procedure is regulated properly.

I care because I occasionally beat much stronger players (e.g. Victor Chow 6/7d several years ago, almost Pavol Lisy 1p one time) so if I am accused of cheating to do that and the panel decide I cheated I would like to be able to dispute their conclusion rather than having signed a declaration I agree with them.

I can understand that with the poor use of statistics in the Carlo Meta case the organisers have shied away from using them again and instead opted for human experts, but I don't think it's obvious that being strong at playing Go makes you strong at detecting Leela-cheating. I think a well designed and applied analysis with statistics and comparison to bots a-la Ken Regan in chess would probably be a better judge than 3 pros, but would take time and effort to make. Perhaps the EGF should consider spending some money for a suitably able person/team to create such a tool if volunteers can't do it for free.

Javaness2 wrote:
As I imagine it...
Last year the process was, claim, judge, if appeal judge again.
This year the process is probably still, claim, judge, if appeal judge again.

My understanding is there is no longer the chance to appeal, the PGETC Arbitration Board's initial decision is final.

Author:  Vesa [ Thu Sep 20, 2018 6:10 am ]
Post subject:  Re: PGETC 2018/9

Javaness2 wrote:
Thanks Vesa - Can you confirm if the pgetc leagues are no longer rated?

Hi Java, no rating of PGETC online games anymore. The decision is written in the AGM 2018 notes.

Cheers,
Vesa

Author:  yakcyll [ Thu Sep 20, 2018 6:12 am ]
Post subject:  Re: PGETC 2018/9

Uberdude wrote:
My understanding is there is no longer the chance to appeal, the PGETC Arbitration Board's initial decision is final.

Had the same impression upon reading it the first time around; I don't think this requirement, worded this particular way, could be understood differently, unless I am unaware of a way you can officially and fruitfully 'disagree' with an Arbitration Board's decision closing an appeal process.

Author:  Javaness2 [ Thu Sep 20, 2018 6:36 am ]
Post subject:  Re: PGETC 2018/9

Okay, that's rather different to my imagination. So the professionals will be ruling on all matters - mundane stuff like whether or not somebody showed up on time - or only on cases of suspected cheating?

Author:  bugsti [ Sat Sep 22, 2018 8:58 am ]
Post subject:  Re: PGETC 2018/9

I think It is interesting to ask to the Panel Board what do they think about Metta case last year. He will probably play at the same level also this year and I bet every one of his moves will be scanned with every AI at our disposal.

So what the Panel Board is going to do if he will play like last year (when he was acquitted at the end)? They agree with that acquittal?

How many false positive will be among 600 games in PGETC 2018/2019?

How do they recognize and treat false positive?

I wonder how many player will accept this silly new procedure :scratch:

Not to mention the fact that player cannot appeal against the Board decision, this is the thing that makes me more doubtful. Is that legal under EGF regulation?

Author:  Pangolino [ Sat Sep 22, 2018 9:59 am ]
Post subject:  Re: PGETC 2018/9

bugsti wrote:
So what the Panel Board is going to do if he will play like last year (when he was acquitted at the end)? They agree with that acquittal?


He was not acquitted.

"We do not think that it was proven without a reasonable doubt that Carlo Metta broke any rules" means that the first accusation was not solid enough, not that he was proven innocent.
The fact that it was presented in your way in the italian go world was sketchy at best.

Author:  Tryss [ Sat Sep 22, 2018 1:32 pm ]
Post subject:  Re: PGETC 2018/9

Pangolino wrote:
bugsti wrote:
So what the Panel Board is going to do if he will play like last year (when he was acquitted at the end)? They agree with that acquittal?


He was not acquitted.

"We do not think that it was proven without a reasonable doubt that Carlo Metta broke any rules" means that the first accusation was not solid enough, not that he was proven innocent.
The fact that it was presented in your way in the italian go world was sketchy at best.


Really ? Usually, when the accusation don't find you guilty, you're acquitted. How does it works here?

Author:  Pangolino [ Sat Sep 22, 2018 2:54 pm ]
Post subject:  Re: PGETC 2018/9

Tryss wrote:
Really ? Usually, when the accusation don't find you guilty, you're acquitted. How does it works here?


When someone is acquitted (with a "Sentenza di proscioglimento") in Italy there are two possible results:

You can have a "Sentenza di non doversi procedere" = the accusation was invalid or there was not enough proof of guilt.
Or you can have a "Sentenza di assoluzione" = the accused was proven innocent.

Prosciolto & proven innocent -> assolto

I am not entirely sure how it works in other legal systems and if there is a clear cut difference in terminology, considering the online sources I checked on the fly.

In the official news on the FIGG website (archived version) Carlo was presented as "assolto" in the title, of course, giving the idea that he was proven innocent by the appeal commission.
The proper way to translate the news was to say that Carlo was "prosciolto", that is more generic. The accusation is not up anymore but we can't say anything about his guilty/not guilty status.

Maybe it was in good faith and whoever wrote the news is not very good with words. Maybe. I doubt it, because those are terms that we hear regularly, on tv, on the newspapers. Those are not technical terms anymore.
The point is that that page was circulated and quoted in that form, leading most of the italian players to believe that Carlo was proven innocent after the appeal.
From the context it was quite clear to me that bugsti was thinking about this version of what happened, where Carlo was declared innocent just to be accused again next year by the new commission.

The only italian source to report what happened in the proper way was is the Go Club Milano, here (archived version).

Quote:
L'appello di Metta è stato vinto in quanto non è stato possibile provare senza ragionevole dubbio l'uso del programma.

The Metta appeal was won because it wasn't possibile to prove beyond reasonable doubt the use of the software.

Author:  bugsti [ Sat Sep 22, 2018 4:28 pm ]
Post subject:  Re: PGETC 2018/9

Pangolino wrote:

He was not acquitted.

"We do not think that it was proven without a reasonable doubt that Carlo Metta broke any rules" means that the first accusation was not solid enough, not that he was proven innocent.


But here a big issue arises: in such a case it is impossible to prove one innocence! How can a referee possibly find a prove of "not cheating" for any possible move? It is clear to me that every possible form of absolution can be written only in the form you mentioned.

Fun fact: also the opposite claim is impossible to prove. How can one prove that cheating occurred? One can think that cheating occured, another can say that it is unlike, another that it is suspicious, another that is fifty-fifty, but nobody can prove it (unless showing some physical evidence).

Am I right?

Author:  Bill Spight [ Sat Sep 22, 2018 4:51 pm ]
Post subject:  Re: PGETC 2018/9

bugsti wrote:
I think It is interesting to ask to the Panel Board what do they think about Metta case last year. He will probably play at the same level also this year and I bet every one of his moves will be scanned with every AI at our disposal.


Curiously, ben David's plays agreed with Leela Zero's (at the 100k setting) more often than Metta's. :shock: How now, brown cow!

Author:  Bill Spight [ Sat Sep 22, 2018 4:52 pm ]
Post subject:  Re: PGETC 2018/9

Pangolino wrote:
bugsti wrote:
So what the Panel Board is going to do if he will play like last year (when he was acquitted at the end)? They agree with that acquittal?


He was not acquitted.

"We do not think that it was proven without a reasonable doubt that Carlo Metta broke any rules" means that the first accusation was not solid enough, not that he was proven innocent.
The fact that it was presented in your way in the italian go world was sketchy at best.


Acquittal is not the same as exoneration or exculpation.

Page 1 of 4 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/