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

[Michael StJohns <[e-mail]>]

On 11/25/2023 4:33 PM, Ted Lemon wrote:
> Why would any demographic standard other than "the whole population" 
> make sense?

Small steps.  If we can get buy in from the IETF to make the IETF look 
more like the tech community, and if we can actually accomplish that, we 
can then try the next step building on success to try and make the IETF 
(and the tech community from which we draw people) look more like the 
whole world.    If we try to make the only and first goal to make the 
IETF look like the world, we will never succeed.  Or at least not as 
long as you or I live.

> I don't think anyone is proposing that we actually succeed, in the 
> IETF, in achieving parity, but clearly we are so utterly removed from 
> parity that some extra effort is reasonably called for to at least 
> make space for there to be a consistent voice for the part of the 
> population that is currently so far from parity.

You and I define parity differently.  For this Nomcom and the next and 
the next, we're dealing with a pool size and composition of possible 
volunteers that's mostly already decided.  That pool will come no where 
close to parity with the population at large. That is a fact.  I'm 
willing at least to try to get to parity with the current tech 
community, or even to a 5 year projection of the changes to that 
community.  It has the benefit of being a supportable argument, and 
mostly not annoying the rest of the folk who are seeing their chances to 
participate diminish (e.g. it is somewhat a zero-sum game).

> We make genuine (although I would say incomplete, since we never go to 
> Africa) efforts for regional parity, so why not do the same for gender 
> parity?

Actually - with respect to the Nomcom we do not.  With the sole 
exception of the max 2 per org rule, the Nomcom runs on randomness.  
With respect to the actual selection of leadership - I would argue that 
we've told the Nomcom to pick the best people, not "and make sure we 
have one from X and one from Y".

If you mean actual physical meetings - well sort of.   There are whole 
swaths of the world we avoid - and for good reason.  Not all of us agree 
on all reasons.  That's a whole other problem and one for a distinctly 
different mailing list.

I participated in the first ICANN Nomcom and regional diversity WAS a 
criteria.  And there was a great deal of horse trading done to 
accomplish that.  But the selection of that Nomcom was targeted rather 
than random and members generally represented communities. It was a 
different, but valid, way of doing things and led to not a little amount 
of cognitive dissonance on my part.   As of 2023 it was also much larger 
than ours so targeting made a bit more sense.

I hesitate to try and pigeon-hole IETF participants in buckets for the 
purpose of selecting Nomcom members or for the selection of group 
membership over technical competence.

Later, Mike

> On Sat, Nov 25, 2023 at 2:33 PM Michael StJohns 
> <[e-mail]> wrote:
>
>     Hi Ted - thanks for the thoughtful note.  Comments.
>
>     On 11/25/2023 1:11 PM, Ted Lemon wrote:
>>     I feel like the problem that you aren't representing here that I
>>     think is quite important, Mike, is that most of us live in
>>     societies that clearly bias in favor of men being in positions
>>     that allow them to travel to the IETF. And I think we've seen
>>     that the IETF is also a fairly unwelcoming environment for people
>>     who are not willing to handle disrespect by responding in kind.
>>     (In cast it's not clear, I DO NOT mean you here—I'm just saying
>>     this about the social structure we dwell in generally).
>
>     Agreed. And there are non-societal structural issues we've started
>     to address - e.g. provision of child care that should start to help.
>
>     (Let me address "non-male community" - Mallory's document uses the
>     term gender and that's got a number of 2020era edges and hooks
>     that don't exactly match to a simple binary choice between male
>     and female.  "non-male community/class/?" was the best I could get
>     to trying to identify the class of have-nots in this discussion.
>     Community may have been a bad choice.  Offense if any was given
>     was not intended.).
>
>
>>     And women are generally, in our societies, socialized to accept
>>     disrespect without responding in kind. They are generally seen in
>>     a poorer light than men if they do respond in kind. They
>>     generally experience negative consequences most men are not
>>     familiar with if they respond in kind. Indeed, they also often
>>     experience these consequences merely by disagreeing, even when
>>     they do so politely and constructively.
>
>     I believe you're correct that the literature says that. The
>     question is how do we - the IETF - mitigate this characteristic of
>     conditioning?
>
>
>>
>>     As a result of this, it is pretty much inevitable, particularly
>>     to the extent that we measure IETF participation on the basis of
>>     who shows up and who talks the most on mailing lists, that women
>>     are going to be underrepresented in the IETF, and consequently on
>>     nomcom.
>
>     Underrepresented based on whole population?  Based on equivalent
>     populations in the tech world? Based on membership on the mailing
>     lists?  Based on WG chairs?
>
>     Based on whole population, pretty much every tech community  has
>     women substantially under-represented.  And mostly not something
>     the IETF can directly address.
>
>     On a SWAG basis I'd guess women  already in the tech world are
>     only moderately under-represented in the IETF. I don't know the
>     answer for mailing lists, and I once did the math for WG chairs
>     and our leadership was actually overrepresented based on the
>     proportion of women WG chairs at that time.
>
>     I started through the volunteer list for the current Nomcom and
>     gave up trying to gender the entries by name. Some are people I
>     know, some I guessed based on the organizational/national 
>     association, but after Linkedin dinged me for looking up a number
>     of the non-european derived names in quick order, I gave up.
>
>     What I would like to know is whether the proportion of women
>     volunteers vs eligible women is similar to the overall proportion
>     vs eligible.  There were 254 volunteers this year - I seem to
>     remember that worked out to be about 10% of the eligible pool.
>
>>
>>     For what it's worth, I don't really like the notion of a
>>     "non-male community." I feel like this draws a line that doesn't
>>     exist. Women are members of our community: they are just treated
>>     differently, and because of that fewer of them participate in
>>     ways that make them eligible for nomcom.
>
>     Attend 3/5 meetings either in person or virtually is all that is
>     needed to become eligible.  Do we want people on the Nomcom
>     without that minimum requirement?  Mallory's document does not
>     suggest that and I wouldn't support it.
>
>>
>>     So the question is not how we address the needs of the "non-male
>>     community," but how we do a better job of including those members
>>     of our community who are not currently being fairly and
>>     effectively included.
>>
>>     I don't actually know how to do that. I'm not convinced that
>>     Mallory's proposal is the best approach. However, I don't have a
>>     better suggestion, and none of the suggestions you've offered
>>     feel to me like they are better suggestions.
>
>     I keep asking the question and no one says more than "it couldn't
>     hurt" - does Mallory's proposal fix a problem other than the math
>     occasionally omitting a woman from the Nomcom?   Does having a
>     guaranteed member of the Nomcom lead to more women being selected
>     to IETF leadership?
>
>>
>>     In an ideal world, I would much rather make the changes that we
>>     need to make so that we have appropriate proportional
>>     representation of women and men in our community, which I think
>>     is roughly 51-49 based on my limited demographic knowledge.
>
>     And you would be wrong.  To the extent that googling gives you
>     anything, take a look at
>     https://www.luisazhou.com/blog/women-in-tech-statistics/ It's
>     roughly 3:1 men to women - so about 25%.
>
>     The rest of the document matches my limited understanding of the
>     overall tech women problem.
>
>>     I think the fact that we could implement Mallory's proposal now
>>     is actually evidence that we've made some progress in that
>>     direction. But this feels very much like a chicken-and-egg
>>     problem. I don't know how to get men in our community to be more
>>     respectful and welcoming of women. Even if we were (and I think
>>     the majority of us are!), I don't know what we can do about the
>>     problem of representation that isn't directly our fault, but is
>>     simply the emergent property of our societies' biases, over which
>>     we have very limited agency.
>>
>>     So having said that, it feels to me like what Mallory has
>>     proposed is something that's actually within our power, and does
>>     address a real problem. And so maybe until we can identify an
>>     approach that we can agree is better, this proposal is a
>>     reasonable place to start.
>>
>     And I'm still disagreeing with "address(es) a real problem".   
>     This is an artifact of how we do things nothing more.  The problem
>     is actually disproportional representation in the leadership if any.
>
>
>     Let me suggest a more radical approach that is evidence based and
>     not triggered by the math and the volunteer pool or a weak
>     statement of a problem:
>
>     1) Resolved: The Nomcom voting membership should reflect the tech
>     industry population, not just the IETF's current population.
>
>     2) The selection algorithm for the Nomcom shall select 2 women for
>     the first two slots of the Nomcom.   The number of women shall be
>     capped at 3. In other words, each Nomcom  will have either 20% or
>     30% of the voting members as women.  Selection for the first two
>     slots shall only be done from volunteers who self-identify for
>     inclusion.
>
>     3) The remaining selections for the Nomcom shall be from the
>     entire pool of volunteers discarding any that would violate a cap
>     restriction.
>
>     4) If the volunteer pool for women falls below 10% of the pool or
>     below 10 women, the guarantee drops to 1.  If the pool contains 
>     <= 10 women then in addition, the cap drops to 2.
>
>     5) The numbers in section 2 shall be readdressed by the IETF
>     community every 8 years and shall be adjusted based on the
>     population of women in technology irrespective of the IETF'c
>     current participation to give guaranteed proportional representation.
>
>     6) Volunteers must self-identify to qualify for selection under
>     2.  This self-identification is not subject to challenge.
>
>
>     I don't actually know if that will effect the output of the
>     Nomcom, but hope springs eternal.
>
>     Mike
>
>>
>>     On Sat, Nov 25, 2023 at 12:42 PM Michael StJohns
>>     <[e-mail]> wrote:
>>
>>         I'm going to reset this slightly and offer a heretical comment.
>>
>>             The lack of a particular community participating on a
>>             given nomcom is NOT a problem, well-known or otherwise. 
>>             It's one fact from a fair random selection.
>>
>>         Instead, "The Nomcom candidate selections under-represent a
>>         given community over many Nomcoms."  would be a problem if
>>         the statistics showed that.  Is that the actual problem we
>>         should be addressing?
>>
>>         Is the following a more accurate statement of the perceived
>>         problem?
>>
>>         If Mallory's document said "We have this problem of
>>         under-representation in the IETF leadership.
>>         Under-representation in the proposed candidates is measurably
>>         increased in the years where the voting members of the Nomcom
>>         do not include a member from that community and here's the
>>         statistics to show that case.  If we imposed a requirement to
>>         require at least one of those community members on each
>>         Nomcom statistics show this would help to resolve the
>>         under-representation of that community within the output
>>         selections."   I would probably be more accepting of
>>         Mallory's proposed solution in that case.
>>
>>         Unfortunately the current document doesn't say that, and I
>>         can't actually figure out what the problem being addressed
>>         actually is.  It can't be that losing a lottery is the actual
>>         problem.
>>
>>         I might still argue against the simplistic approach proposed
>>         to solving the under-representation problem.  I would suggest
>>         that instead the liaison/advisor approach coupled with
>>         greater oversight by the confirming bodies and a willingness
>>         by them to require the Nomcom to reconsider selections as a
>>         means of achieving appropriate levels of representation might
>>         address an under-representation problem.
>>
>>         We address the input conditions to affect the output
>>         conditions.  If adjusting the input condition results in no
>>         difference to the output conditions, why are we twiddling? 
>>         If adjusting the input conditions does affect the output -
>>         show me the math.
>>
>>         A few other questions before I end - related to the proposal
>>         on the table and that will need to be addressed without hand
>>         waving:
>>
>>         1) Does the mandatory requirement apply if the volunteer pool
>>         participation for a community falls below 10%?  5%?  3%?  Or
>>         goes above 20%? 30%?
>>
>>         2) What is the criteria for selecting other communities for
>>         similar treatment?
>>
>>         3) Are claims of community membership subject to challenge
>>         (similar to claims of organizational association)?
>>
>>         4) What happens if the sole selected community member needs
>>         to be replaced?  Consider both the before the Nomcom convenes
>>         and after?
>>
>>         5) Is there a sunset clause?  If so, how is it triggered?  If
>>         not, why not?
>>
>>         Lastly - the more general algorithm that would probably work
>>         without collapse is to select the special community members
>>         at the beginning of the process applying all of the rest of
>>         the rules for the remaining batch of selections.
>>
>>         And to capture the possibilities:
>>
>>         Possible changes to the Nomcom process (zero or more of these):
>>
>>         1) Impose a requirement to select at least one member from
>>         the non-male community to serve on the Nomcom.
>>
>>         1a) .... at least 2 members...  [[generally to address a past
>>         issue which has resulted in knock on present issues, you over
>>         represent the affected community for a period of time]]
>>
>>         2) Increase the number of voting members from 10 to 15 or
>>         16.  At 16, with 10% of the pool, a community has a 4 in 5
>>         chance of at least 1 member, a 1 in 2 chance of at least 2.
>>
>>         3) Provide a non-male liaison/advisor either for every Nomcom
>>         or for any Nomcom without a non-male voting member.
>>
>>         4) Have a mandatory 1/2 briefing for the Nomcom on diversity
>>         issues prior to their first deliberation?
>>
>>         Later, Mike
>>
>>
>>         -- 
>>         Eligibility-discuss mailing list
>>         [e-mail]
>>         https://www.ietf.org/mailman/listinfo/eligibility-discuss
>>
>