At 12:54 PM 9/16/2014, Dave Crocker wrote: >On 9/16/2014 9:40 AM, Stewart Bryant wrote: >> It would be useful to run the stats to see how close we have ever come >> to that. The affiliations are not on the site so it's difficult to do a >> quick check. My recollection is that we have never seen worse than 2+2, >> and that does not happen every year. > > >I'll suggest that past statistics aren't all that relevant. The issue >is the sampling model and its potential for problematic outcomes. > >A design that permits 5 companies to /ever/ fully populate the Nomcom is >a flawed design. Actually, it's worse than that. The number is 3 companies to get 6 slots. At that point, those three companies control the outcome of the Nomcom. Of course, being too obvious will result in interesting backlash. >d/ > >ps. It's worth applying that same analysis approach to questions of >competence. No matter how bright or well-intentioned everyone might be, >would Nomcom be competent if all of its members only met the minimum >selection criteria and, for example, /no/ member of Nomcom had /any/ >experience actually participating in a working group, writing any RFCs, >or chairing any WGs? The current rules permit that outcome. Yup. I considered this as well, and then reasoned that if you capped the pool proportion (the input to the selection process) rather than (or in addition to) the output you might get companies not doing the "everybody volunteer" thing and instead actually having people volunteer who have a track record of interaction with the IETF. Mike >-- >Dave Crocker >Brandenburg InternetWorking >bbiw.net