Following a number of the threads on diversity and, in particular, on whether the effort to get a better demographic view of participation will lead to quotas, I have been increasingly uncomfortable with some of the arguments which appear to have some presumptions about how diversity and meritocracy relate.
To describe the issue, I'd like to start with a different situation and then draw a parallel. The different situation I'd like to use is a startup company experiencing growth. In my experience, startups that succeed tend to have a very strong core group that comes together early on; they tackle the hard work and develop a great degree of understanding of and trust in each others' capabilities. As the company grows, it's very common for that core group to continue to rely on each other whenever a difficult problem arrives. That can manifest in those folks moving up to be the top of a hierarchy and individually handling delegation; it can also manifest itself in severe bottlenecks as the individuals remain critical resources to solve an increasingly large number of problems.
In both cases, it's common for the individuals to pull in their own networks of trusted folks as support. Another way of expressing this is that a particular human network is the basis of an enterprise, and the scaling of that human network tends to work by each one of the humans pulling in additional folks from their personal networks whose skills are personally known to them. The result of that is that the start-up *is* a meritocracy as it grows (because the individuals are chosen based on their abilities), but its diversity is initially limited to that of the personal networks of those who end up in critical positions. As the company grows and recruiting becomes more formalized, the overall make-up may become more diverse, but key positions may remain less diverse as the human networks remain in place or are renewed. Note that the scope of this diversity may have nothing to do with race or gender, but may instead be about schools, disciplines, or ages. (When it's schools, we even get to reuse the phrase "old boy network" in its original sense).
In the IETF, things are slightly different, in that attendance and participation are completely open (there's no hiring gate), but many of the same human networks are in play. As a working group chair, when I stare out at a sea of faces looking for a scribe, the chances of my asking someone I know produces good minutes is much higher than my asking someone whose work I don't know. But that also translates into the pool of candidates being *only those people I know*, because that's the only pool whose merits I have assessed. In other words, even though I'm selecting on merit (good note takers),
the way in which merit is determined (personal knowledge) results in my
not using the whole pool.
If there were an objective measure I could use instead, the WG's pool of potential scribes would go up and the allocation likely would be fairer--if I could say: "please tell me which potential minute taker (with a score of 70 or above) was tapped for the work least recently" and then tap that individual, things get better for those who are otherwise tapped too often. Note again that the increased diversity in that pool may have nothing to do with race or gender or even age, but it might instead be in technical interest area (since I came from APPs into RAI, my background is focused in certain areas).
The individual impact of my limited human networks may be small (I hope so, anyway); in the best case, the limitations of mine would be overcome by the scope of my co-chairs' and ADs'. But it can easily be a self-reinforcing instead; if all the chairs come from the same backgrounds, they may know and trust the same people. Those people likely are being selected for merit--but not from the total available pool.
As folks worry about quotas and its impact on quality, I think we must recognize that the effort to promote based on merit alone is subject to the limits by which merits are assessed. The more human those are, the higher the likelihood that network limitations or cognitive bias will have an impact on our best use of the volunteers we have or could attack.
So, given this very human problem, what can we do? Suresh and I happened to be at a different SDO meeting yesterday, and we sat down briefly and discussed this. Two things emerged as possible concrete actions from that brainstorming, but they both boil down to this: institutionalize methods for getting technical input from newcomers/less well-known participants, so that technical input can be basis of assessments. In other words, give those outside the human networks an institutional method for being part of the technical community. That should align with chairs' chronic need to find technical energy to assess and complete work.
The first suggestion is a "Newcomer's directorate". After an individual starts participating in the IETF, either by attending a meeting or joining a mailing list, offer them the opportunity to become part of a review team. That review team would have templates (as does AppsDir, the GenArt team, and others) and example reviews to guide the new participant. The newcomers would be assigned last call reviews that would then be sent to the relevant working group chairs and ADs for review. Those who produced good work would then be known to others relatively quickly. At the end of their "newcomer" period they might transition to a relevant directorate based in an Area.
The second suggestion is a simple tool that at WG call time (be it last call or call for adoption) randomly selects a set number of participants from the mailing list, and then asks for a review or commentary. So 5 folks off the mailing list are directly asked for their opinion, without regard to preconceived notions of the chairs about who would be a good reviewer. If someone declines, the tool would select a new random person to fill out. The working group as a whole thus gets a chance to understand someone's technical viewpoint, without that person having to fit within one of the established human networks.
There are other methods that may well be better than the two Suresh and I discussed, but I put these forward as a potentially concrete step that may help those struggling with this to understand that the end result of this need not be quotas. It should be a better environment for all of our volunteers.
best regards,
Ted Hardie