Subject: Re: [boost] GNOME outreachy
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2018-08-24 08:44:32
> Brief web search should explain the issues.
> Alternatively, this story should
I wouldn't necessarily draw from LLVM an opinion for Boost, as indeed
some recently received private email pointed out (let's call it
"passionately worded", it's weird how boost-dev is my sole source of
such kind of email, despite me participating in over a dozen mailing
lists and forums).
Setting aside the moral or legal issues with discrimination based on
sex, race, colour, language, religion, political or other opinions,
national or social origin, association with a national minority,
property, or birth (the UN and ECHR definition I believe), I also don't
think there is any such problem in the C++ community, except in one
regard which I'll come back to.
Speaking as a former GSoC admin, many of our students - perhaps even
most - came particularly from the poorer parts of the world where the
GSoC stipend is worth a lot more. I don't think there is any racial
diversity problem. We definitely showed excessive selection bias towards
those who wrote good English, and I did a lot of work to try and debias
that towards people who could code instead. But even then, in the
statistical analyses of our ranking patterns, I never discerned any bias
As anyone who has attended a C++ conference or standards meeting knows,
there is a wide fluidity of folk, some of the more fluid of which are
amongst the highest esteemed engineers in C++. Nobody sees anything once
people start discussing C++. I don't think there is any LGBT diversity
problem, if anything in terms of percentages I think we're ahead of
general society, possibly precisely because we don't see anything but
expertise in C++.
In all this, C++ has a very similar community to Physics, where a very
similar culture prevails. In fact, the staff at C++ Now in Aspen often
comment on the similarities, because the Physics conference is very
close date wise to C++ Now, and apart from us being a bit plumper on
average, and wearing more t-shirts, it is apparently hard to tell us apart.
Where I do think we have a diversity problem is in female participation,
and I think that has similar reasons for a similar lack of participation
as in Physics (the percentage of women is almost identical). In my five
years at GSoC I believe we only ever had one woman as a student, and
that was despite my best efforts to do better. I'd need to go dig out
the spreadsheets, but the cause was lack of female applications. I don't
think we ever received more than five out of many hundreds of
applications. And incidentally, the very worst of those applications was
vastly better than the average we receive.
Now *that* diversity problem I'd very much like to do something about as
that's a real problem in C++. I know me and Jon Kalb and lots of others
have discussed this problem in depth, and on multiple occasions, and the
conference organisers have adopted codes of conduct and other welcoming
measures, plus some of the folk at ACCU have worked on increasing
bringing more kids into C++ via coding camps and so on. So it's not like
nothing is being done about this.
But equally, like in Physics, it's very tough to do better if the C++
employment culture is uncompetitive for women. What few women start out
in a career in C++ see a marked exodus about five years in, they move
into management or into a different technology, and I find that
unwelcome. But also illustrative - C++ has a very traditional employment
culture compared to say web development, or Python, or even Rust.
Employers expect you there onsite eight hours a day, there are a lot of
legacy code bases full of arcane bespoke knowledge, and experience is
often prioritised over capability in big multinationals. None of this
makes C++ competitive to alternative choices for women whose work
history may need to be more flexible, less full time, less buried in
Anyway, that's my ha'pennies worth on this topic: better to concentrate
on problems specific to C++ rather than import culture wars from the US.
And if you're reading this and you are in a position to enact workplace
cultural changes more tolerant of part time work and other female
participation friendly changes, please do make it happen. We get there
incrementally only through individual change, and C++ needs to compete
better against other programming language ecosystems and cultures than
it has been doing.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk