Social media posts

I post on different social media. Sometimes the posts contain information that might be useful to reference later, so I will include them here. List of topics:
  1. How long does it take to complete a case
  2. Status of work from volunteers in November/December 2020

How long does it take to complete a case"

July 6, 2021.
For a "complete case" - it is sufficient information for a trier of fact to determine if a player or pair is guilty. I typically provide all examples, including both suspicious deals and anti-cheating deals. For example, if it were Brian Platnick's RKC example (see another thread) then I would provide all such auctions for the trier of fact. Compare a "complete case" to the written report that ACBL produces for its hearings.

As part of the data I provide, I always provide all data that I have on a pair in a private URL. This helps the accused search the data as well. I always provide a separate URL which identifies all of a pair's leads for the same reason. Thus everyone gets all information (that I have).

I do include some, but not all, statistical data. This can be useful but I am not suggesting that anyone is convicted based just on their statistics. Yet.

I include enough deals in the written write-up to justify - in my opinion - what is needed beyond a reasonable doubt. Some times I provide a little more just in case. But I don't take this to excess. There is no need to provide a human written analysis of every board this pair has played. This is not something ACBL does for its cases at the moment. One of the most prolific cheating pairs I track has played over 18,000 boards. To sate our combined curiosity, I took at a look at this pair to see how long it would take to write a "complete case". 15 minutes is the answer. Can send if you want. It will probably take 30-60 minutes for someone to review the data, the cited examples, and confirm they are a cheating pair.

I will offer to send you what I recently sent someone else who reported how long it took them to review. Or, if you do not want to personally review them, PM (private message) me the contact information (e.g. email) of someone you trust. I will email them an example of a "complete case" and they can tell you how long it took them to read, review and decide. I can send either the one I referenced earlier in this post, or can send you the most recent one (I will probably anonymize the data in either case).

To be clear: there are some cases so blatant that you can quickly decide. I am not suggesting that all cases are this way.

You may know a little more than me if there are hearings where the party is found not guilty. These should not be made public, but you may have been on the panel.

Status of work from volunteers in November/December 2020

July 6, 2021.
I was asked
"Nicolas, Late last year you posted links to hands played by 50 pairs (mostly suspected cheats but a few controls). I'm curious about how many of those pairs have actually been disciplined at this point?"
My reply:

None.

The work was done in Nov/Dec last year. Unlike an ACBL recorder form of a single possible instance, these were complete cases that the volunteers had each spent some number of hours on.

I looked up where these pairs are on my current cheating list (it is updated daily) for ACBL BBO events (some pairs from Nov/Dec may have been from the VACB database).

This is where these pairs currently rank in my most-likely-to-be-cheating list of players with 1000+ boards. I have 10,800 pairs in this list:

59
71
72
74
87
91
93
106
111
119
120
121
137
142
154
181
193
227
232
271
318
357
418
555
613
653
822
931
1034
1689

This is 30 pairs. In the original list there was also some top pairs from FTF play (e.g. Fisher/Schwartz) as well as some honest placebos and honest top players from FTF play. I do not know how many cases were filed with the ACBL by the volunteers.

Looks like, for the last four at least that they may have stopped cheating or were honest placebos. Or these may have been pairs that were on the threshold of automatic detection and human review was needed. I was curious, so did a quick look at the current data, and the data back in November.

1689: This was an honest pair. Volunteers did not find them to be a cheating pair. No case filed. I did have to put honest pairs in the data to avoid confirmation bias. 1034: This was a cheating pair. But they did not cheat all the time. I do not know if a case was filed. 931: This pair was picked to test a threshold. The volunteer(s) thought they were cheating but may not get convicted. I do not know if a case was filed. I remember this pair because it was right on a threshold I had predicted of being difficult to convict.

The top 25 pairs on this list have played 178,546 boards combined.

As you can see, I did not pick the most blatant cheating pairs. The blatant cheaters are very obvious to catch and there is little need for human review, the computer can write up these cases. I did pick some pairs that were on the threshold to find where volunteers believe there is not enough evidence. I have used this to tweak the parameters and thresholds I now use.

Before you condemn ACBL for being so slow... these cases were filed in Nov/Dec of last year. We do not know how many cases ACBL had at the time, or the backlog. Each case has to be verified and a decision made to investigate. Each case will take time. Hopefully the evidence that the volunteers created would be shared (after all identifying information from the volunteer is removed) with the accused just so the accused understands how strong the case is (or isn't).

Then you have to allow time for the accused to prepare a defense. Set up a hearing etc.

However, it has been at least six months. Once would have hoped for at least one negotiated resolution from these pairs.

If/when there is the first conviction from the volunteers from last year, I will let you know.