I read the unofficial Google internal memo that has been the subject of some controversy, and upon reading it my Wikipedian-trained instincts were to wonder where the citations were that should, if they were available, have supported numerous assertions that were made in that memo. I'm not an expert in diversity -- and I suspect that the author of that memo isn't, either. In the absence of verifiable and reliable sources, I'm skeptical of numerous assertions that were made in that document.
This leads me a question that I've had in mind for awhile. How can we increase the diversity of Wikimedia technical contributors and staff? I'm referring both to gender diversity and racial diversity (people of African descent appear to be significantly under-represented).
My unscientific hunch is that what would help is increasing people at young ages to consider a career in a science, technology, engineering, or math ("STEM") field, and then continuing to support their interest from elementary school through college.
(Personal story: I was a poor performer at math in middle school and at one point I emotionally gave up on the subject, yet I did significantly better when I reached college and (a) had instructors whose styles were more compatible with how I learn and (b) had classroom environments that were more supportive of learning.)
I don't know to what extent Wikimedia should be involved in encouraging people at early ages to become interested and stay involved with STEM, and I think that we should ask ourselves if perhaps this is an area in which we should make some financial and time investments, with the goal of facilitating development of diverse candidates into engineering and technical roles for the community as well as organizations like WMDE and WMF. We probably shouldn't be steering people at young ages to make long-term commitments to STEM or the Wikimedia ecosystem, but perhaps we could take some actions that would at least encourage them if they seem to be interested in STEM to continue their academic growth in those domains. I don't know if there is data that explains how gender and racial disparities develop and how to address them, but my hunch is that the earlier that the issues are addressed, the better.
I don't know what other options to suggest; perhaps people here will have some ideas. I'd particularly like to invite Victoria to the conversation; perhaps she can comment sometime in the next several days (probably not for several hours, since this is still Sunday evening on the US west coast).