Re: Debian, lists and discrimination
On Sun, Aug 08, 2004 at 12:16:37AM -0400, Jaldhar H. Vyas wrote:
> On Sat, 7 Aug 2004, Matthew Palmer wrote:
> > As to the barriers to involvement in Debian by women, it's pretty obvious
> > that our gender participation ratio is decidedly different to that of the IT
> > industry in general, let alone the general population. I believe (although
> > I'd find it harder to back up with real numbers) that our female / male
> > participation ratio is also lower than participation in the wider OSS world.
> "The IT industry" is awfully vague. I question if you can make useful
> conclusions based upon it. I recently taught a perl class for working
> professionals in which 60% of the students were female. What does that
> say about the IT industry? The Perl community? Nothing that's what.
It shows that there were quite a number of women interested in a technical
topic. That counts for something.
You've got to have something to compare against, obviously. "The IT
industry" is possibly a poor choice, depending on what you want to classify
as "IT". But there are certainly fairly accurate industry figures for
gender participation -- I know, because I've seen them for Australia,
grouped in a variety of ways.
Comparison to OSS contributions is even better, because it excludes the
possibility that women don't contribute to OSS. You've done this for some
projects, which is good.
> A more apt comparison would be with other free operating system projects.
> Let's see: FreeBSD has approximately 330 people with CVS commit
> privileges.  I counted 5 identifiably female names. That's 1.5 %.
> Gentoo has approximately 250 developers  I counted 4 identifiably
> female names. That's 1.6%. In contrast, the last time this subject came
> up, I looked and Debian developers comprised approximately 1%. (But
> because, we are a bigger project, that's more actual women, 8 or 9 iirc.)
> So we are slightly behind but by a statistically neglible amount.
>From the 2004.04.05 version of the keyring, I count 927 unique people or
unknowns (there are several "might be people, might be 'bots'"). Wandering
through the name list, I count 122 "might not be male" and 4 "almost
certainly female". That equates to an actual female contingent of somewhere
between 0.4% and 13.2%. Note that the "might not be male" is very, very
broad -- basically if I don't know of a male with the same name, or I know
of a female with the same or a similar name, it's on the "might not" list.
Suddenly I have this urge to add a "gender" flag to db.debian.org to make
these sorts of things easier, but then we have "race", and before you know
it DHS is doing racial profiling on our keyring to find out if we're a
All in all, it is a fairly small difference statistically, I will grant you.
I'd like to measure the larger contributor pool, though, if possible,
because that's mostly what I'm interested in at present. Boosting the
keyring with feminine names is a worthy, but secondary goal of my