Re: [Gendispatch] New Version Notification for draft-knodel-nomcom-gender-representation-00.txt

[Brian E Carpenter <[e-mail]>]

On 22-Nov-23 17:12, Martin Thomson wrote:
> On Wed, Nov 22, 2023, at 14:25, Brian E Carpenter wrote:
>> One nit: if the volunteers are asked to select their own gender
>> category, that isn't a binary selection. So is the algorithm male vs
>> female, or male vs non-male, and how do we handle "prefer not to say"?
>> Maybe that is more than a nit.
> 
> To be precise, Mallory's draft suggests to limit to no more than 9 of the one gender.  As others have noted, that could be 8 in the same way and remain similarly equitable.  How we collect this information is relevant here.  This is an area where offering many choices is not necessarily going to provide the best outcome, but I consider the attack profile such that we can afford some flexibility in the interest of being respectful.

Yes, but if (hypothetically) we had four possible responses: Male, Female, Other, Prefer-not-to-say,
I really don't know what to do with Prefer-not-to-say.

We are probably not the first organisation to face this question.

     Brian
  
> On balance, I think that Mallory's proposed approach is worth trying.  I'd prefer the 9 person limit marginally over 8 for reasons related to those below, but I wouldn't oppose a move to be more proactive in this area.
> 
> Like Brian, I would also caution against trying too hard to hit diversity targets across other axes.  We have many things to consider here (geographies, areas of technical focus, ...).  Some of those things are not always obvious or measurable.  Constraining the random selection is only likely to regress diversity on those axes we don't constrain and an overconstrained selection would create other problems.  The NomCom is not always a perfect representation of our community, but I find that it does tend to provide a reasonable sampling in the aggregate.