[EuroPython] Lack of diversity within selected talks

[holger krekel]

Hi Lynn,

i am just one member of the programme committee of EP2014 but would like
to share some information and my own perspective on the discussion and
issues.  We hopefully can arrange a committee meeting soon and can
conclude on the path forward, also taking your suggestions into account.
So this mail is no official answer or so but hope it contributes to
clearing things up a bit.

On Tue, Apr 15, 2014 at 02:28 -0700, Lynn Root wrote:
> To the EuroPython organizers, talk reviewers, and community at large,
> 
> For those of you who do not know me, I am a board member of the Python
> Software Foundation, the founder and leader of PyLadies San Francisco,
> and an engineer at Spotify.  I have been a speaker at the last two
> EuroPythons, with 3 talks last year, and a keynote the year before.

I enjoyed your talks, also the DNS one a few days ago :)

> I see that the list of preliminarily talks are publicly available. Side
> stepping my issue with lack of communication to proposers of talks at
> large, I am writing to bring light to the lack of diversity of the
> current list of talks, and propose some action items.
> 
> There is how I understand things as they are. Please correct me if I am
> wrong.
> 
> - talk selection was/is being done blindly, as in no identifying
> information about the speaker is revealed

Yes, about 250 reviewers went through the 300 submissions and we
got 3-5 reviews on most.  The reviewers could not see other reviewers
comments/evaluations in the first round.  And they could not see
authors or genders or countries.  Which doesn't mean that gender-bias
doesn't take place otherwise, of course.

> - there are very little women on that preliminary talk list slated to
> speak

yes but there are still 50 free slots awaiting allocation.  The
currently published talks include only those which received "+1"s but no
negative ("-0" or "-1").  The others we call the "conflicted talks" and
there are many that got several positive but one negative vote. They
were not included in the first published list.

> - there are multiple selected speakers slated to give multiple selected
> talks

I am not sure how many "multi-talkers" we actually have.  We should be able
to find out shortly.  FWIW i submitted three talks myself and one got accepted
in the first round.  Of the two not-accepted ones there was one that made it 
to Pycon 2014 surviving harsher competition.  On the other hand, one
talk i gave as a well-received keynote at EP2013 was not accepted at all
for Pycon 2014.  So judging from my personal experience, I believe we 
have quite some randomness in talk selection at python conferences.
But that's more of a general sidenote.

> If you do not find a problem with item #2 and #3, please read this
> article [1] about importance of diversity in a technical field.

I also do see a problem particularly in #2.

> Here are my suggestions to rectify this issue:
> - limit speakers to only give one talk. Yes this means going back on the
> original acceptance.

I am certainly fine with asking multi-talkers to cut down.  If the programme
committee revokes already accepted talks is a different question.  I'd like
to see how many cases we actually have.

Do you know, btw, if Pycon 2014 employed a "one talk only" policy?
It seemed to me that some people gave 2 or maybe three talks at Pycon.
And personally, i wouldn't totally rule out the possibility of allowing
more than one talk but such cases should be considered explicitely
and carefully.

> - reopen CfP and reach out to PyLadies globally to help get the word
> out. As one of the main leaders of the global organization, I know this
> did not happen originally.

Working with PyLadies sounds like a great idea and it's bad we missed
out on this so far.  I'd like to discuss the idea of doing an additional
specialized CFP in the context of the still remaining question of the 50
to-be-allocated slots.  Obviously a full re-open of the CFP would be
quite a challenge, also given that eastern and vacations are quickly
approaching.

> - re-review the talks. Give preference or help for those who would be
> first time speakers. First time speakers may need far more help writing
> a proposal tailored to the EuroPython audience. As reviewers, you have
> an understanding of the EP community and should help pull up new
> speakers.

For the 50 remaining slots i'd like to increase focus on "does this talk
help to grow the community" with a particular focus on diversity.  As a
secondary measure, looking through the already accepted talks might be
feasible but i hope we don't need to consider reverting acceptance
against a proposers will.

> - related to #2, and #3, have open office hours  or create general
> availability during the time that the CfP is reopened to help those who
> want it craft a good proposal.
> - select talks for the remainder of the program with the context of the
> preliminarily talks in mind.

right, that's what I'd like to aim for.

> I understand that the blind selection process was meant in good faith to
> remove bias.

I am personally skeptical of the "double blind" selection process.
It probably better fits for answering "is this a valuable study/paper
contributing to the advance of science" rather than the question i see
Python conferences having: "Does this talk interest a part of the
community and does it help to grow our communities? Do we have a
diverse range of topics, countries, personal backgrounds including gender
properly reflected in the schedule?"

> However, the result is troubling, and needs to be looked at
> in context. If this preliminary list has any influence on the actual
> program, the conference will suffer in terms of overall diversity in
> attendance. I'm not writing to discuss the merit of diversity at a tech
> conference, because I have faith that the reviewers and organizers
> already grasp its importance. But this email is to address what I feel
> needs to change.

Thanks a lot Lynn.  I hope we can all work this out and move to a better
direction now.  Hopefully i can say a bit more after we get some more
involved people together, part of which are currently travelling 
or otherwise busy etc.

best and hope you also recover well from the Pycon blast :)
holger


> Thank you,
> 
> Lynn Root
> 
> [1]
> http://www.nytimes.com/2014/04/06/technology/technologys-man-problem.html?_r=0
> _______________________________________________
> EuroPython 2014  Berlin, 21th27th July
> EuroPython mailing list
> EuroPython at python.org
> https://mail.python.org/mailman/listinfo/europython
>