[Bigbang-dev] Gender diversity and draft productivity

[Sebastian Benthall]

Thanks for engaging on this, Juliana.


> I'm sorry if I don't understand very well here, but I think there is no
> operationalization difference, I'm just proposing to change the concept
> of "gender diversity" because, in this case, I don't see any possibility
> to include a more diverse spectrum. With just "gender" you can describe
> how women' active participation (in a traditionally male dominant
> space?) affect a wg production. Is that what you want to see? Or may you
> specify where the question comes from?
>

The question is coming from Article 19, who are sponsoring this phase of
development, as well as members of the BigBang community that have
questions along these lines.

At the last meeting, I recall Joey Salazar and Corinne Cath being most
enthusiastic about this topic. I invite them to clarify what they meant by
their question.

>From context and memory, I believe they are indeed interested in variations
in the role of women in IETF, where they are a minority.

>From the perspective of the toolkit, we can in principle expand the
functionality to address a wide range of questions beyond the original one.
But in order to address those questions, we need to be able to
operationalize <https://en.wikipedia.org/wiki/Operationalization> the
research question in terms of software operations. If it cannot be
operationalized like that, then I wouldn't recommend using BigBang to try
to address the research question, except perhaps as an auxiliary role in a
mixed methods approach.


> > 'non-binary' is, if such, a self determined gender option, not related
> >> to this "unknown" (.5) output
>
> > Is it? Why?
>
> People who doesn't feel comfortable with a male/female gender roll.
> Someone can decide by them self to use a non-binary name (i.e. Julien
> Marie) but the point with non-binary is that the person consciously
> define this in their name or preferred pronoun (and life, of course).
> With this tool, as you are using it, we have no possibility to know why
> the name is "unknow" in terms of gender. The most accurate answer, I
> think, is the western bias you mentioned, and also English language
> bias. For example, *Sol* (in Spanish), and *Inti* (in Quechua) -which
> translates *Sun*- are male, female and non-binary names in Andean South
> America.
>

Currently, the "unknown" option is getting used whenever the name
recognizer fails to identify a name as either masculine or feminine.

Looking into it... BigBang is currently using the `gender-detector` Python
package, whose creators are well aware of these issues.

https://pypi.org/project/gender-detector/
https://github.com/jeremybmerrill/beauvoir

Interestingly, it does appear to support internationalization, though it
supports only UK, US, Argentina, and Uraguay at the moment.

Most likely the best way to improve this methodology would be to make a
contribution to that project with a Gender/Name data set derived from a
different national population.

However, looking into it, the US name data
<https://raw.githubusercontent.com/malev/gender-detector/master/gender_detector/data/usprocessed.csv>
they have contains many more "non-Western" names than you might expect.
Here is a sample:

https://gist.github.com/sbenthall/7e32068066fe3352630beafa02c63f66

It appears that they have *Sol* labeled correctly as a gender-neutral name.
They say that it is "Unknown", meaning, I presume, that there is not enough
information to tell what gender, if any, the person has.

They have "Gurshan", "Gurshaan", and "Gurshawn" but not "Gurshabad" I'm
afraid.

These resources are made available by the Open Gender Tracking project.
I expect that to address this problem seriously, it would require some kind
of academic or non-profit investment in creating an equivalent dataset for,
say, India or China.
http://opengendertracking.github.io/


> >> 4. N/A
> >>
> >
> > This seems reasonable to me. "Unknown" or not enough evidence to tell is
> > also needed.
>
> That's what I referred with N/A
>

I believe N/A has a slightly different meaning: that the question was, in
the case of this particular answer, ill-formed.

As we are dealing with questions of statistical accuracy, I think it's
better to be specific when the data available does not bring the judgment
to sufficient confidence.


> Sorry again, maybe I'm in a wrong place. I am trying to use BigBang to
> make an analysis but I'm not a developer. The BigBang-user mailing list
> has no recent movement so I landed here.
>

No, you are in the right place. The users list is inactive.
Please feel free to elaborate on what you are trying to do with your
analysis.
We would be glad to help if we can.

That said, usage of BigBang is currently very hands on.
Very little works as a turn-key solution.
So you may find it worthwhile to change from "not a developer/data
scientists" to "a beginning developer/data scientist".
I won't presume to know your background. But I believe there is a world of
difference in that small change in subjective identity.


> I totally respect the purpose of this list and if you feel my comment as
> an abuse please let me know. But I also think that, as we don't need to
> be experts to talk about any topic, in the quest for inclusiveness it is
> necessary to contemplate a more interdisciplinary approach, and that is
> something we can obtain by discussing these kind of issues.
>

Your comments so far are very welcome.

If I'm signaling a limit on the scope of discussion, it's because I know of
other cases where discussions of these kinds of sensitive topics have
seriously derailed things.

Speaking for myself, I write and talk with sometimes more, sometimes less
expertise about many topics, interdisciplinarily, across a wide variety of
contexts. I would not consider it appropriate to bring up all of that other
work on this list.

However, I infer that you are considering using BigBang in an
interdisciplinary project of some kind. Would you like to elaborate on that?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ghserv.net/pipermail/bigbang-dev/attachments/20200708/d4810836/attachment-0001.html>