[sanst]

I tried to put this post in Changing Laws first, but that went wrong. The same happened in this forum. I will try the Changing Laws forum again.

Here’s my text:

Per January 1st the Laws 73 and 89 will change. See http://www.worldbrid...plicate-bridge/ for the text.
There’s already a discussion going on on Bridgewinners. Regarding the new Law 89D1 and 2 I’m quite curious how your average club director is to establish foul play with a statistically significant correlation. Although I studied a lot of statistics - well almost fifty years back, so that knowledge is rusty at best - I wouldn’t know how to do this. At least there should be standards for method, accuracy, precision and reliability.

[axman]

pescetom, on 2023-December-14, 06:44, said:

The forum is broken - but this one got through. I can't post to my thread in I/A though, was so all yesterday too.

Perhaps BBO objects to the topic.

[pescetom]

sanst, on 2023-December-14, 06:41, said:

Regarding the new Law 89D1 and 2 I’m quite curious how your average club director is to establish foul play with a statistically significant correlation. Although I studied a lot of statistics - well almost fifty years back, so that knowledge is rusty at best - I wouldn’t know how to do this. At least there should be standards for method, accuracy, precision and reliability.

There was fussing about this in the BW thread too, but even an outspoken statistician like Richard W. suggested that the language was clear enough and that the main purpose of this new sub-Law was to enable the use of such analysis rather than to define operational requirements.

In any case, the existing Laws and associated procedures like Polls often require the TD to make quasi-statistical decisions, often on distressingly small sample sizes and non-standardised criteria, so putting a club TD on the spot is nothing new in that sense.

I think it does change things to have it spelled out in Law that we can use a correlation as grounds for determination of wrongdoing, though. Especially if we have access to electronic play records or audio-video (don't hold your breath for that in the club, in Italy at least).

[pescetom]

axman, on 2023-December-14, 08:01, said:

Perhaps BBO objects to the topic.

Humour aside, the problem remains, at least for me: I can make a new post, but if I make a reply then when I try to send it I often get the "Unable or Server down for maintenance" page.
The bypass is to make a new post and then edit it to insert the quote and reply.

[blackshoe]

Something weird was going on. When I logged in just now I found two empty topics on this subject in the "changing laws" forum, and one with text and replies in "laws and rulings". So I moved the one to "changing laws". Whatever weirdness was going on seems to be over now.

[hrothgar]

pescetom, on 2023-December-14, 11:41, said:

There was fussing about this in the BW thread too, but even an outspoken statistician like Richard W. suggested that the language was clear enough and that the main purpose of this new sub-Law was to enable the use of such analysis rather than to define operational requirements.

I don't claim any special knowledge regarding what the the folks who drafted Law 89 were actually thinking, but I wouldn't be surprised to discover that this Law is responsive to some specific issues that that cropped up in the not so distant past. For example, there's been at least one case over in Europe in which the defense (approximately) claim "Well, you know, the Laws never say that you can't use out of band signalling, so you can punish us for doing so". Hence we have Law 89, section A+B

In a similar vein, I think that Law 89 Sections C+D are best interpreted as address recent debates around the necessity to "crack the code" with respect to how pairs are signally. We've seen cases were the Defense alleges that the Prosecution doesn't have a perfect description of the signalling methods that the Defendant are allegedly using. They can point to individual hands that are inconsistent with cheating method foo or bar or what have you. The defense claims that these discrepancies are sufficient discredit the broader claim.

I wouldn't be surprised to discover that this section of Law 89 is intended to put this to rest.

There's a bunch of discussion going on on Bridge Winner's right now wrt "false positives" and the possibility that an innocent pair might get swept up by some anomaly detection system that is use drawing inference based on Statistical correlation. (To me, at least, the discussion around parameterization feels part of this broader discussion).

I don't consider that discussion to be particularly interesting (and I wouldn't be surprised to discover that Law 89 isn't really talking about these sorts of issues)

[pescetom]

hrothgar, on 2023-December-17, 10:47, said:

In a similar vein, I think that Law 89 Sections C+D are best interpreted as address recent debates around the necessity to "crack the code" with respect to how pairs are signally. We've seen cases were the Defense alleges that the Prosecution doesn't have a perfect description of the signalling methods that the Defendant are allegedly using. They can point to individual hands that are inconsistent with cheating method foo or bar or what have you. The defense claims that these discrepancies are sufficient discredit the broader claim.

I wouldn't be surprised to discover that this section of Law 89 is intended to put this to rest.

IIRC, the CAS absolved FN mainly on this basis (some positions were inconsistent) even without overwhelming arguments by the defence or much respect for statistical theory. So kudos to WBF here, agreed.