|
Hello John See PPE below On 10/1/18, 7:01 PM, "ietf on behalf of John C Klensin" <ietf-bounces@xxxxxxxx on behalf of john-ietf@xxxxxxx> wrote: Padma, I'm reasonably familiar with that literature and some of the underlying studies and have been quite impressed with both the diversity outcome and general quality ones by other measures. Whether it is usefully applicable to the IETF nominations and selection process is another matter; see below. >> Today, AFAIK (I may be wrong or unaware of unwritten rules) >> it seems that company affiliation is the main diversity >> factor. > > Affiliation is part of the (written) selection criteria. More precisely, it is, at least AFAIK, the only easily-measured factor in the rules other than the basic "Nomcom eligibility" ones. <PPE> Agree this metric is easily measured. Would you agree ”is it enough?” is an interesting worthwhile question? >> Undoubtedly, this is a good start but surely it can be >> augmented for other diversity considerations as well such as >> gender parity... >... >> If we are committed on diversity and enlarging the pool of
>> candidates then possible starting points could be >... >> 1. More diversity on the interview panels and other levels
>> contributing to key selection criteria >... > The first point would be challenging. There is an existing > mentorship programme (point 2). As for point 3, please see > draft-klensin-nomcom-incumbents-first-00. Since one of my efforts to sort through pieces of this (it went nowhere, and so did all of its relatives) was cited, let me address that first, challenging, point. I think what has gotten lost in this piece of the discussion is that the Nomcom isn't an "interview panel", it is a selection mechanism and panel. Perhaps more important, when the mechanism of selection by a Nominations Committee was established, there were two key assumptions about that committee. <PPE> As there is no specific literature here, so the context might have been unclear. IMO, the selection of a candidate is typically the result of the debrief of an interview panel. First, the Nomcom membership was expected to represent a random sample of IETF participants. At least implicitly, part of that expectation was that the workload and duration would be sufficiently low that the people who would volunteer would be a good enough sample of the community that a random selection of them would closely approximate a good random sample of the community. That is almost certainly no longer the case: Nomcoms are a lot of work for their members, requiring very active involvement for many months. That, in turn, biases the pool toward people who have unlimited time to devote to the IETF, who do not have significant other IETF responsibilities (e.g., for technical work, as WG Chairs, etc.), and, for people with significant organizational/employer commitments, whose organizations are willing to support Nomcom-type work.
<PPE> We are in agreement here. Second, there was an assumption --again, at least implicit-- that Nomcom members would actually know, and often would have worked with, most of the plausible candidates for IESG and IAB positions and therefore be able to rely on their own knowledge and make selections by discussing that knowledge and experiences. Perhaps at least in part because of the changes above and because the IETF has simply gotten bigger with more participants who are specialized in one particular area, that is no longer the case. As a result (I think) Nomcoms are increasingly dependent on questionnaires (another problem, IMO, interviews, and recommendations/endorsements rather than first-hand knowledge of candidates by members. Those approaches are inherently time-consuming, reinforcing the statistically-biased membership described above. <PPE> Thanks for giving some context – indeed that shrinks the pool of viable candidates Would you agree that this assumption also brings with it unconscious bias? Given that backdrop, I doubt that one could tinker with selection criteria for the Nomcom (beyond the "no more than two from a company" rule) without making the Nomcom even less representative of the community. In other words, one might increase one type of (easily-measured) diversity there at the expense of decreasing overall representativeness and range of experience and perspectives. <PPE> Hmm … not convinced here … We are mostly in agreement. If no change is made then we back to square one. Reading through the thread, diversity also may mean something different depending on your personal experience or interest: technical diversity, diversity in companies by business, geographical location. large/small companies, gender
… May be the first step is to define what type of diversity would be beneficial.. However, after the very long thead(s), I am pessimistic on getting consensus on this …. Now I think that it may be time to rethink the Nomcom model, including the random selection idea, entirely, but I have seen little or no sign of energy in the community for dealing with that ... and no good ideas for consideration either. <PPE> As the thread is already long, taking discussion offline best Regards Padma best, john |
