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

[Christian Huitema <[e-mail]>]

On 11/23/2023 11:43 AM, Eric Rescorla wrote:
> On Thu, Nov 23, 2023 at 11:29 AM Brian E Carpenter <
> [e-mail]> wrote:
> 
>> On 24-Nov-23 06:18, Eric Rescorla wrote:
>>>
>>>...
>> Having women *consistently* present in NomCom will have a subtle long
>> term effect, by persuading women that it's worth volunteering for NomCom.
>> And that will have a further long term effect in persuading women that
>> they should participate in the IETF (which is our underlying problem).
>>
>> This is indeed something we should have done years ago. Visible
>> diversity enhances diversity.
>>
>>> As Christian (and I earlier) noted, because the nomcom itself is quite
>> small,
>>> it is possible for women to be very underrepresented in a given nomcom
>>> with respect to the underlying pool. You may believe that's not an issue,
>>> but to the extent to which it is an issue, I don't think "try to get
>> more women
>>> into the pool" is a plausible response to that issue.
>>
>> What I think my argument shows is that we have a feedback loop, and
>> Mallory's proposal is the easiest way of inserting an amplifier in
>> that loop.
>>
> 
> Hi Brian,
> 
> I largely agree with your analysis here. The point I was trying to make was
> that even in the absence of such a feedback loop, if you think it's
> important
> to have women on the nomcom, then "let's try to get more women into the
> pool" is not a sufficient response.

As I wrote, the maths work for reserving one nomcom seat for women does 
work, and the effect on the reminder of the population is fairly 
minimal. The "reservation" process would kick in about every five or six 
years, and would only affect on seat out of ten that year -- so maybe 
one seat out of 50 overall.

But a caveat: the math only works because women are a sizeable minority, 
about 15 to 20% of the nomcom volunteers. If we tried that for smaller 
minorities, we would get some nasty side effects. And for classes larger 
than 25%, the random choice already provide the desired results.

Our random choice system is actually pretty good at providing seats to 
minorities -- it does that much better than, say, plain elections. It 
also does that without explicitly tagging who belongs to what group, 
which I believe is a great property. But the size of the nomcom is 
small, which implies great year over year variations.

Maybe those variations are in fact an asset. Successive nomcom have 
different sensitivities, will values different aspects of different 
candidacies, resulting in significantly diverse candidate selections 
over multiple years. If we believe that, the proper action would be to 
do nothing.

If on the contrary we want to guarantee that every year sees some 
non-zero representation of various minorities, we have two options. We 
could do a variation of what Mallory proposes, or we could just increase 
the size of the nomcom to 15 or 16.

Like Mike StJohns, I find this "increase size" alternative quite 
seductive. It would reduce the risk of "zero representation" for women, 
and also increase the chances of representation of other minorities. And 
it would do that without requiring a system of cataloguing who belongs 
to what group.

-- Christian Huitema