Wikimedia Technical Conference/2018/Session notes/Fireside chat with Victoria Coleman, Toby Negrin, Erik Moeller, Franziska Heine

From mediawiki.org

Rough Transcription:

  • Introductory remarks
  • Victoria’s remarks:
    • Prompt from Deb: What organization have platforms similar to ours?
    • Shoutout to Erik and Katie
    • Affirming how hard this job is, of taking this platform since it has been static for so many years with barnacles on it and patches, and growing it
    • At Yahoo!, this was something they were grappling with this as well; the membership stack had been around as long as the company, which is basically as long as the internet, and little had been done to it
    • Very fundamental since it was the “door” to all of yahoo’s properties, integral to their success
    • Operating this platform on such a scale, handled 5 billion logins a day, which rendered it untouchable and “too big to fail”
    • After 15/20 years of working this way, Victoria joins and her colleague was telling her about how exhausted their engineers were maintaining this. This is her vision of what happens to static platforms - engineers falling on the floor from exhaustion
    • It isn’t just the WMF that faces this, but also major corporations, so take heart.
    • They had 800 libraries, and maybe they knew what 10 or 20 did. Every change was an unpredictable event. This creations a culture of conservatism.
    • They were serving a login view on mobile that was [?]. It took 4 weeks to push out a change to it, because there were so many libraries that were breaking the system in unpredictable ways.
    • Starting to rebuild on the side of the house where the volume was low but the logic was more complex, and made the decision to use Java, which lead to headhunting to source the right people
    • How to lead to a different stack without hitting these problems.
    • Let’s avoid arguments about which stacks are better (e.g. PHP vs. JS) - but the use-cases are the important thing
    • We will find out a few years from now!
  • Erik
    • Previously worked at WMF. Starting editing in 2001. Involved in some of the early development. Joined as deputy Director of WMF. Started building up . Now work for Freedom of Press foundation. Working on SecureDrop - allows whistleblowers to provide info anonymously.
    • For Wikimedia, what platforms are comparable? Nothing. Competitors? None. WM is in a unique landscape.
    • Some similarities with OSM, but also totally different, very decentralized, more Apache style, open network of participation. But parallels with needing to create policies, with new contributor engagement.
    • Also Debian, Mozilla, etc.
    • Technical debt in any corp with complex tech stack. Especially for orgs that reach the 5-6 year age.
  • Franziska - heads software development dept at WMDE
    • Want to give more insight in berlin and wikidata and jump off Erik’s remarks
    • My job is kind of a combo of Toby and Victoria -- both of those sides are brought together in WMDE / with Wikidata
    • Most of you also know Birgit and others from the team
    • What did we do this year up to now?  Quite a lot, actually.
    • On one hand, focused on building out product features:
      • lexemes v2,
      • census,
      • constraint checks
      • term box (esp. For mobile editing)
      • ElasticSearch (with Stas from WMF)
      • Docker stuff for Wikibase (enabling customization for Docker image for Wikibase)
    • While doing that, we were also reorganizing the way our engineering teams work and organize themselves, we gave the whole dept a structure; our product people are becoming more of a team
    • This is actually a lot of work on top of actually creating the products
    • We also held three workshops that were the work of [?] .  These were to work with users of third party Wikibase installations and to learn how to better support them.  We gave them environments to play with. UK Research Council was the biggest one, also some people from Canada.  It’s an emerging community and we are trying to be good partners.
    • So many other projects I haven't mentioned!
      • Training program for Wikidata so that users can train others
      • Project (and another conference) with German National Library
    • We want to set up the org to grow even more in coming years
  • Toby - CTO
    • both technology and product all have their yahoo scars
    • It was a booming and successful company for a long time, esp in early 2000s
    • learned from the yahoo experience about how a beloved site can lose relevance, which really motivates him now to keep us relevant
    • Part of the user-generated content wave. Different lessons on the product side, on how quickly a site can lose relevance
    • Yahoo used to give out blackberries; used to say, we don’t think there’s a lot happening in this mobile web-space
    • Toby started using the web on an iphone and was just amazed
    • Zuckerberg has said about twitter: it’s a clowncar that fell into a gold mine, but that’s true of facebook as well
    • but poor yahoo’s portal, did not work on a mobile device
    • at wikimedia, contribution is one of the key things here that keeps us relevant
    • we need this contribution to stay fresh, and the ability to contribute on a phone is INCREDIBLY import to keep us relevant and in people’s minds and the future
    • This mobile work is the focus of the product side of things
    • the non-editing features such as washlist [sp/] that we are bringing in
    • this is a “target rich environment,” and its hard to try not to do it all, but that’s our goal at the moment
    • how difficult change can be - its difficult to grow, and it’s hard to grow things that people are used to, but it's necessary
    • shipping page previews, one of the most recent desktop changes. this was launched without much of a peep and lead to a 20% growth
    • we can do hard things and keep changing!
  • Erik:
    • Wikipedia was very successful even before there was a Foundation, which is a pattern I haven't seen elsewhere. Newspapers were writing articles about us and Jimmy was making public announcements about software changes that were going to come, but there was only a single employed developer making those changes! As organizations grow they usually do so rapidly, but for us it wasn't until the move to SF. Mismatch between organizations desires and capabilities. Aaron Schulz was an early contractor and made some early contributions on flagged revisions [that make a reviewed article visible], and for a while that was the big idea everyone was talking about, as the most important thing anyone could do, with big announcements at WIkimania about it coming in 6 months, but being built by 1 developer at 15 hours week.  
    • This pattern of mismatch between ambitions and capabilities didn’t really go away even after the Foundation moved to SF and started growing around 2008.
    • It would take another 3 years before the foundation had design resources on staff, and another 5 years before Analytics resources on staff.
    • When discussing things like visual editor with the board, were asking to just “plop” other things in - which won’t work. More complicated, seven person team when they had maybe two or three.
    • The early years were characterized by taking on very ambitious projects with insufficient resources. This led to a lot of learning and improvements over time.
    • Was this a mistake? Arguable, but if we hadn't tried then we wouldn't have learned many of those lessons.
    • We should have said "no" a lot more, earlier; but on the other hand that's just a reality of very tiny organizations, that there are more needs than capabilities.
    • We are getting better at balancing our demands with our resources, and making sure the teams are working better - ops getting involved when it needs to, for example.  Better managing our own resources, in addition to having more resources.
    • Having the resources to implement things properly is such a wonderful position to be in.

Victoria’s Remarks re: where we came from and where we’re going

    • We are here all together to figure this out. Once it’s as easy to edit on mobile as on desktop, we’re there. Once it's as easy to find an article in Swahili as it is in English, we're there.
    • As a community, we need to come together and identify priorities.
    • Attended the translation meeting early, we need both a product and a technology strategy after attending that meeting
    • We need an AI strategy. We don't even know what that would look like if it were to be successful
    • Figuring out where we want to be in all these arcs is key
    • Question to ask: Seventeen years on: Are we a software company? Are we a production [?] Are we an NGO? What are we?
    • To provide a juxtaposition: used to work at Nokia, which for the longest time was in denial and considered itself a supply chain based company. Now they do browsing and became complex software items, which they did not adapt to
    • We were talking earlier today about the calamity of having two parsers, and it is a calamity for an org this small, but consider that Nokia was killed by the Symbian platform -- there was a different version for every phone the company built.
    • We think we’re in a bad way but it’s good to compare us to these fallen giants
    • @Greg how long does it take to do a build? Greg: ~minutes.
    • How long did it take to do a Symbian build…. Two weeks! 48 hours to compile!
    • What is the velocity of this product if that is the delay? If we are as dependant on our software platform as nokia was for Symbian, then the future is bleak.  We need to support our community in many ways BUT fundamentally we are a software organization and we should put this in the center. If we fail there, the projects will fail.
    • We are not too big to fail.

Franziska:

    • +1 to Victoria. Nobody is too big to fail. Good reasons to continue to ask ourselves the hard questions.
    • Applies not only to MediaWiki but also Wikibase.
    • Next year we'll be continuing along the same lines as this year, focus on core aspects of Wikidata, look into quality staying high, lowering barriers to entry to edit and manage the data there,
    • We’ll continue to work on constraint checks, shared expressions, automated quality scores, but also closer integration between Wikidata and the Wikimedia sister projects
    • E.g. how can we edit wikidata items from within the Wikipedias, or [?]
    • At the same time as we continue to focus on these core elements, we’ll continue to treat the Wikibase ecosystem as another important stream of our work
    • We’re talking to users we want to continue building for, doing add’l conferences
    • Next year will include Federation, ease of installation, but also:
    • How do we approach partnerships - what kind of partnerships are there, what requirements, do they need training, what does it take to work with them, what documentation do they need to work on their own?

Toby:

  • In answer to Erik’s question: it is never enough
  • Katherine likes to say that we’ve grown as an organization, but as you grow you start to see the gaps even more
  • As a C-team, they wrestle with this.  There’s still definitely a ways to go.
  • Won’t talk too much more about it but he’s excited to figure this out with V and everyone
  • Wants to talk about what they will do BETTER in the future, since we can’t predict the new avenues
  • I’d really like to be able to understand users and their motivations, building the affordances they need to participate in free knowledge
  • Lead with understanding of users. We’ve come a long way but there is more to go.
  • Feeling a gap with our communities- so much depends on being able to work with them, instead of doing things to them.
  • We live and die with the communities so we can understand both in a quantitative sense (how many people does it take to create a community, etc)
  • But also don't have perfect ideas on how to help established communities to grow, and face the challenges their scale has.
  • It will take the whole foundation, as well as the product and tech teams, in addition to the community to make these things work. Overall, wants to get closer to the users and communities.

Q&A Section:

  • Question: [Derk-Jan Hartman (TheDJ)] Where do you guys see the best opportunities for non-WMF people within the movement?  Particularly third-party or volunteer devs.
    • Erik: generally, there’s an unlimited number of opportunities for volunteers to show the foundation where it needs to go. We have Magnus [Manske] as an example of someone in the community with a history of prototyping ideas of where we need to go, such as Resonator, which we really need to build from a prototype to a tool. The Foundation is actually in a terrible position to institute new projects, because when we make noise, its immediately tainted as “the bureaucracy wants this” and is seen as top-down in a negative way
    • Many ideas I'd like to see more completely realized. Wikidata has short snippets, but we don't have anything ideal for short excerpts that could be shown inline, for example in episode guides, anything that requires short blurbs.
    • There are many examples like this where you can kind of see the gaps in our software, but even for things like the New Projects page, people in the community aren’t pushing them forward the way they were, for example, 10 years ago.  E.g., library of 3D objects. You’re not going to send a message to wikimedia-l and get excitement and volunteer commitments.  Need to work to build community interest, build a prototype, make some noise again, be persistent. But I think there’s still plenty of room to do that kind of thing now.  
    • Volunteers, and chapter affiliates, are often in a better position to do this.
    • Toby: Wikilove could use work.  You should be the ones taking chances, trying new things. You should be the next Magnus.  We’re actually now trying to be much more open with things that we’re putting resources toward and that our teams are building.