Sunday, March 27, 2016
Remarks on Wikimedia Foundation recent events
If you pay attention to Wikipedia culture and the WMF, you may know that the Executive Director of the WMF, Lila Tretikov, has resigned amid some controversy.
It is an extraordinary story, especially since, given the nature of Wikipedia culture, so much information about events is publicly available. I'll point you to Molly White's "Wikimedia timeline of recent events" as an excellent synopsis of Ms. Tretikov's tenure as ED. The thing that strikes me most about that timeline is the number of people who left, and the long tenure of each person who departed. On the same subject, Terry Chay's note published on Quora also addresses this.
My own tenure at WMF was just over three years, from 2012 to 2015. In that time Željko Filipin and I built an exceptionally good browser test automation framework, which at the time I left WMF was in use in about twenty different WMF code repositories. My time at WMF was roughly evenly split between Ms. Tretikov as ED and under the previous ED, Sue Gardner.
There are two things about Wikipedia and WMF that I think are key to understanding the failures of communication and culture under Ms. Tretikov's leadership.
As background, understand that everyone in the Wikimedia movement, without exception, and sometimes to a degree approaching zealotry, is committed to the vision: "Imagine a world in which every single human being can freely share in the sum of all knowledge." I still am committed to this myself. My time at WMF absolutely shaped how I see the world.
Given that, what is important to understand is that Wikipedia is essentially a conservative culture. The status quo is supremely important, and attempting to change the status quo is *always* met with resistance. Always. There is good reason for that: Wikipedians see themselves as protecting the world's knowledge, and changes to the current status are naturally perceived as a threat to the quality or even the existence of that knowledge.
The other thing important to understand is that many of the staff at WMF come from the Wikipedia movement or the FOSS movement. Many (not all) of the technical staff began working with Wikimedia/FOSS software in college or even in high school, and ended up employed by WMF without ever experiencing how software is made and managed elsewhere. Likewise many (not all) of the management staff were (and are) important figures in the Wikipedia movement, without much experience in other milieux.
In practice, when attempting to make a change to Wikimedia software or Wikipedia culture, the default answer is always "no". No, you can't use that programming language, that library, that design approach, that framework. No, you can't introduce that feature or that methodology.
So a big part of the work for those working in this culture is persuasion. One is constantly justifying one's ideas and actions to both one's peers and to management, and to the community, in the face of constant skepticism. Wikipedians talk about "consensus culture" but in practice consensus is actually more along the lines of "grudging acceptance". Sue Gardner's most recent blog post explains this better than I ever could.
And because so many Wikipedians have such a dearth of experience of other tech culture, NIH (Not Invented Here) is rampant. It was difficult to introduce proven, reliable, well-known tools simply because they were *too* well-known; they aren't *Wikipedia* tools, they don't have *Wikipedia* support, there is limited knowledge of them within the culture.
The result of these forces is that significant feature releases tend to be fiascos, but each fiasco of a somewhat different character. When WMF released the Visual Editor, the software was not fit for use, everyone involved knew it was not fit for use (or should have, they were certainly told), and the community rejected it for good reason. On the other hand, the Media Viewer *was* fit for use when it was released, but it was such a new paradigm that the community rejected it even more decisively than they had the Visual Editor. We could even speculate that had Media Viewer been as unusable upon release as the Visual Editor was, it might have received a kinder reception from the Wikipedia community.
Some notable exceptions to the fiasco release pattern were the Mobile Web work; the Mobile Web team did a great job and demonstrably made Wikipedia better, even if often over the occasional objections of their peers on the technical staff. And the rollout of HHVM went well, as did the introduction of ElasticSearch, but none of these projects faced the Wikipedia old guard directly.
It also is notable that it took Željko and me three years to get our work accepted widely across all of WMF. Today I am building essentially the same system for Salesforce.org (the philanthropic entity attached to Salesforce.com) as Željko and I did for WMF. I expect to have my Salesforce.org project in the same position as the WMF project in one year, because I don't face the constant hurdle of having to persuade and persuade and persuade. Again, this is not necessarily a Bad Thing: the institutional skepticism and constant jockeying for acceptance of ideas, tools, and practices at WMF is a mechanism that protects the core mission of Wikipedia, even if it often makes the culture psychologically trying if not outright poisonous. You could argue that having to justify beforehand and evangelize afterward every step we took made the system that Željko and I built better than it would otherwise have been. If I seem to have a low opinion of the WMF understand that in my time at WMF I did some of the best work I've ever done, and I consider my time there to be the pinnacle of my career so far.
So it is perfectly understandable that Ms. Tretikov as Executive Director would want to launch an ambitious skunkworks project in secret. This is something CEOs do. CEOs have discretion over the budget, and they are responsible to shareholders for profits. But the Executive Director of the WMF cannot expect to hide a quarter-million dollar project engaging entities beyond Wikipedia without dire consequences, which is exactly what happened. Or was at least the final act in a long series of poorly executed maneuvers that alienated staff and community to the point of near-paralysis, and that caused a monumental loss of faith from the community as well as a huge loss of institutional knowledge as so many experienced staff abandoned the Foundation, or were abandoned by the Foundation.
I imagine that WMF and the Wikipedia movement will toddle on much as they always have. The Wikipedia vision of free knowledge for every human being remains compelling. And I hope that this troublesome period in the history of WMF can serve as a lesson not only to the Wikipedia community, but to the rest of us concerned with how best to make software work for our world.