Like most college kids, I had a college job. For almost two years, I was providing IT support in the School of Social Work (“A brown guy working in IT support? You don’t say”).
A big part of my job was to talk to the staff and faculty of this liberal arts school. When I looked left and right, I saw most of them were women. Most students in the classes were women.
As much as I liked working with them, I could not relate to them. I didn’t feel like I belonged.
A controversy has been brewing in the software development world. ElectronConf, a conference hosted by Github, announced it was cancelling the conference because the chosen speaker panel was not diverse enough. Still, the optics on this are bad.
That’s because there’s two missing pieces of information here. The trivial one is that they did that only after someone on Twitter pointed out that the announced speakers were all men. The significant piece is that the speaker review process was blind.
From the event website (emphasis mine):
Submissions will be initially blind reviewed by a panel of GitHub employees from a range of departments and backgrounds.
What that means is that the speaker information was irrelevant for the purposes of talk proposal review. This includes speaker name, gender and race. It is fair to assume that the intent of this blind review was to select talks based on their technical merit (or rather, the ability of the proposer to show their merit).
As with any controversy, the Internet has been abuzz with comments from both sides of the argument. What we know is that the blind review process did not work. In fact, it completely broke down.
What I want to see Github do now is:
Refine their marketing channels
Github is a product for developers. 95% of its users are self-identified males. That means 95% of the people they reached with their blog were males. If Github wants to reach to a more diverse audience, perhaps they should do active outreach to groups like Girls Who Code, National Society of Black Engineers, and do personal outreach to industry leaders.
Check for bias in their review panel
Maybe the panel itself was biased. 78% of Github’s technical staff are men; it’s within the realm of possibilities that the review panel was even less diverse. If they don’t already, perhaps having a review panel that’s representative of the speakers Github wants will be helpful.
Publish a post-mortem
It is likely this will happen again. Github should publish a post-mortem specifically addressing the concerns with this blind review process, their learnings and changes they will be implementing.
Support minority groups
In a randomized, blind sampling, a subset of a uniform distribution is also uniform. For tech, this means that a randomized sample of a male-dominated industry would also be male-dominant. Diversity issues in tech are occasionally attributed to “the pipeline problem”. Github should start to materially invest in supporting minority groups by sponsoring scholarships, and inviting members of minority groups to the conference.
Tech with its viral factor, can cause a product to reach millions of people. But to be universally useful and accessible, the people writing the code and making the decisions need to be representative of the users. As research as shown, working with people that look and act like you is comfortable, but often counter-productive.
This controversy highlights the work that needs to be done. Github, as one the leaders in software development, should identify the debate beyond a PR mess, address it with passion, and not make rash decisions.