I stumbled across this interesting idea for a pledge to boycott all-male panels at technical and scientific conferences:
There's an underlying assumption behind the pledge that I know for a fact isn't universally true: that an all male panel can never be representative of the available experts on the topic of the panel, thus suggesting unconscious selection bias whenever it happens.
I've only been involved in 2 conference panels in my life - one at PyConAU 2011 on Python 3, and one coming up at PyCon US 2012 on the current state of Python packaging and software distribution.
The first panel consisted entirely of CPython core developers. The lack of gender diversity on that panel had nothing to do with the conference organisers, it was an accurate reflection of a diversity problem that exists elsewhere: the number of female core developers for CPython still stands at a grand total of zero, and the number of female patch contributors over the life of the projects is incredibly low as well (as indicated by the CPython ACKS file.
Other significant Python communities (including the PSF itself) are better in this regard than CPython core development, but a policy (personal or otherwise) that would make panels on CPython verboten is clearly either premature or lacking in nuance.
The prospects for a diverse panel about Python packaging aren't any better - the project leaders we're considering inviting cover pretty much all the major projects, but are once again all guys (and I'm actually hoping enough of them won't be at PyCon that I can avoid having to cull the list of potential panelists myself)
(Given that many people consider having anything to do with attempting to improve Python packaging to be a soul destroying venture that is doomed to leave you feeling drained and disappointed, I certainly wouldn't be pointing newcomers to the community in that direction unless they indicated a prior interest. Even they were already interested, I'd at least warn them that the distributed nature of the problem and some of the complicated history makes it even more political in nature than most design and development activities).
So, there's a genuine diversity problem, but the panel make-ups are merely reflecting the supply-side diversity problem, rather than a selection problem.
Reading this article did make me ponder that question further, but these topics are sufficiently niche and the relevant communities sufficiently small that I'm still happy that the lack of diversity on the planned panel is due to a real lack of diversity.
I'm now the BDFL delegate for the in-work PEPs related to packaging, so if I'm not aware of someone working on something related to packaging, they're far enough away from getting anything integrated back into the standard library to make what they're working on off-topic for the upcoming panel.
Getting back to the original article though, I think the nuance I find to be missing from it is that it doesn't scale down to niche topics from communities genuinely lacking in diversity, both of which will often apply at open source community conferences - the community as a whole has diversity issues, and panels may be put together on quite arcane topics.