--On Tuesday, August 23, 2022 20:48 +0000 Warren Kumari <warren@xxxxxxxxxx> wrote: >... > It still feels like there is a tiny tweak that can be made > that somehow magically solves this without having to rerun the > algorithm (minus the selected people), but everything I think > of is either wildly baroque, or relies on secrecy, or > similar… I have a horrible feeling I'm going to wake up at > 3AM with the perfect solution, just to realize once I'm fully > awake that it is completely, obviously and hilariously wrong. The tiny tweak is one I suggested in passing some days ago and to which the part of your note I omitted above gets very close. If actual problem we are concerned about is to prevent Company X, who has people picked in slots 1, 5, and next in queue at 11 from playing games, we decide that, as soon as Company X has two people selected for the Nomcom, they are finished and all of their other employees are dropped from the pool before any other selections are made. At that point, if someone drops out (whether from Company X or not) we can just pick person 11, who is guaranteed to not be from that company. As far as "customer of Sponsor X" or "Corporate Partner of Acme Tool and Die" are concerned, we already have that problem because, if we wanted to eliminate it, we'd have to figure out how to exclude more than two people from a company, its partners, its customers, and anyone else able to be influenced by it. Down that path lies madness and unfair treatment even if it would possible in some cases. Note too that, even if the randomization is repeated (especially if it were repeated without excluding everyone from the "two selected already" Company X, the same problem applies that applies to the first pool. A company that wants high odds of having one or two people on the Nomcom can control the likelihood of selection by being sure it has enough volunteers for the pool (from the company or, as above, those it might be able to influence). Borrowing a bit from PHB but not going quite as far, the entire Nomcom system relies on almost everyone who volunteers being of high integrity and making choices based on what they know and what they learn during the process, not on either preconceived preferences or company inference. We may be able to lower the risk of certain attacks being successful but, if we cannot mostly count on individual integrity once the Nomcom starts looking at candidates (and that includes other Nomcom members noticing bad/biased behavior among one or two of their peers and ignoring the recommendations that come with it), we'd better start rethinking the whole model, not putting energy into tiny tweaks for selected odd cases. best, john p.s. Don, I think I did see someone suggesting starting the whole process over starting with redrawing the entire pool and selections. Could be wrong about that even though I can make a statistical/combinatorial argument for it if we are concerned about people looking at the initial pool and playing games. But, either way...
