Wikimedia Foundation Annual Plan/Quarterly check-ins/Reading Jan 2017

Notes from the Quarterly Review meeting with the Wikimedia Foundation's Reading and Community Tech teams, January 24, 2017, 9:30am -11:00 am PDT.

Please keep in mind that these minutes are mostly a rough paraphrase of what was said at the meeting, rather than a source of authoritative information. Consider referring to the presentation slides, blog posts, press releases and other official material

Present (in the office): Josh, Nirzar, Danny, Carolyn, Lisa Gruwell, Anthony, Zhou, Michelle, Heather, Juliet, Jaime, Monte, Tilman, Jon Katz, Toby; participating remotely: Adam B, Anne, Baha, bd808, Bernd Sitzmann, Brad, Corey Floyd, Dan Garry, Dmitry, Gergő, Jan, Joady, Joe Walsh, Jon, Josh, Katherine, Leon, Maggie Dennis, Michael, Moushira, Natalia Harateh, Niharika, Olga, Peter, Quim, rho, Ryan Kaldari, Sarah Malik, Stephen, Trevor Parscal, Wes, Zareen

Slide Deck edit

Presentation slides for the January 2017 Quarterly check-in for the WMF's Reading and Community Tech teams

Minutes edit

Slide 1 edit

<title page>


Slide 2 edit

Toby: Logos of our teams. Reading team, focused on our readers, Comm Tech, focused on technical needs of the community.

Slide 3 edit

Toby: Dec. 2015: Readership was declining. We went thru a strategic process and came up with a few goals to rectify this:

Slide 4 edit

Slide 5 edit

Slide 6 edit

Toby: A year later: Wikipedia is not dying!

Slide 7 edit

Toby: You can see, around October of this year, we started beating last year's page views. This has kept up for ~3 mos.

Slide 8 edit

Slide 9 edit

Slide 10 edit

Slide 11 edit

Toby: It's really cool how we came up with goals, set resources toward them, and accomplish them. It's difficult to talk about specific causality, but this is nonetheless something to celebrate.

  • For instance, U.S. election might have caused a bump; that said, the World Cup last year might have had the same effect
Also lazy loaded references could have made a difference.  Could be lots of different initiatives. 

Slide 12 edit

JK: (recapping our strategy) two-pronged high-level strategy:

  • help people learn more in parts of the world where we're well known
  • expand our audience and reduce barriers in places where we're not

...leading to 3 initiatives:

  • improving/modernizing the existing wikipedia experience
  • finding new readers and helping them
  • experimental features for new readers to engage and learn

Slide 13 edit

JK: Two pieces needed for all of these:

  • study to understand our readers better
  • Invest in services

Slide 14 edit

JK: This slide maps the above discussion to our annual plan.

Slide 15 edit

Slide 16 edit

Slide 17 edit

JK: (note on goal 1: not part of the strategic discussion above)

Slide 18 edit

Slide 19 edit

JK: A lot of our work revolves around this objective (improve the encyclo expeerience).


Slide 20 edit

Olga: (re: Improve the encyclopeida experience) The hovercards feature has been our main focus for the past Q. It's been our most-used beta feature. We're working to move it to stable.

We want to make links more useful to readers by providing context on what they might view. Because it's a big change we want to be thorough in reviewing our code and working with our communities.

The code in the past was v. difficult to work with, made us worry about maintainability, so we decided to rewrite it in order to improve maintainability and ease future development. It allowed us to go from a v. complicated structure to something more simple. (>1000 -> ~400 LOC)

We also did a lot of testing/analysis of both quantitative and qualitative performance of the feature. Want to be thorough in analysis and documentation. How is it being used? Does it affect fundraising? Etc.

It's been kind of a long process, but b/c we tested on so many WPs, we were able to go communities on which it had been tested and give them good data on usage. No one has said no. Hebrew Wikipedia asked for it.

Starting major rollouts next month, then rolling out in stages to finish with English around July next year.

Slide 21 edit

JK: we consolidated learnings toward end of deck

Slide 22 edit

Corey: (discussing role of services)

The mobile content service is used to surface (a) community-generated content like main page content, and (b) analytics content such as most-read article statistics.

Currently: We're integrating real-time trending articles into the products. Based on work by Jon Robson leveraging edit data.

In regards to new feature delivery, we're working on a push notification service so that we can deliver this kind of content in a timely manner. iOS will leverage this work first.

Slide 23 edit

Slide 24 edit

Josh: On iOS, we're exploring push notifications. We had a goal for last Q (spilled over to this Q) to deliver our first push notification. Traditionally, notifications on WP have been for editors. This will be the first to be targeted at readers.

The app product teams (Android and iOS) tend to trade off developing new features. We're working now on updating the Nearby feature to incorporate features like keyword search. Will be rebranded to Places. Will leverage the mobile content service (or a similar service).

Slide 25 edit

Slide 26 edit

(illustration of the future of Nearby/Places)

Q from Katherine: Is the map search using work done on OSM integration, or?

Josh: No, not using OSM, will integrate with Apple Maps which is quite robust. Did an evaluation, OSM would need working with a third-party library which was large and buggy.

??: Why apple maps vs. google maps?

Josh: ... also, Apple did work on privacy re maps, a unique identifer is sent per each tile request rather than over the entire session

Wes: is Android going to be able to benefit from this?

Josh: from some other aspects, work done earlier with/by(?) maps team

Wes: Is that going to have the potential to be seen on mobile web?

Josh/Toby: Probably out of scope here, probably technically possible

Michelle: can readers opt out of notifications?

Josh: it's actually opt-in. On English WP they're fairly low volume, maybe 3/week vs dailyon frwiki. The frequency varies by project (based on how frequently that community decides to update the main page content).

Michelle: can opt out back later?

Josh: yes

Slide 27 edit

Dmitry: Our goal is to make broad improvements to our offline functionality.

Last Q we ran a volunteer study with design research to gauge our usability. Found quite a few issues around our reading lists feature, as well as general offline behavior. We task-ized a bunch of them and are working on them this Q.

It's especially important because we intend for offline to be one of our core use cases.

Slide 28 edit

Slide 29 edit

Jon: There's also an objective about improving rendering on mobile.

Olga: Much of our work this Q was about reaching out to communities about features for which dev work was completed in previous quarters. (WD descriptions, related pages) Now they are live on just about every Wikipedia (WD descriptions since last Fri on En) Related pages -- recent pause came from us b/c we wanted to test something on the stable English channel. Wanted to test performance of the feature after go-live, in a way we haven't been able to do before. We'll be looking at how it performs and affects page views.

We created and tested a new header for the mobile web to reach out to new readers and increase retention.

Q: What does FOUC mean: A: Flash Of Unstyled Content

Adam: (re: Reading Infrastructure)

We're working on getting things looking good on mobile. Currently MobileFrontend does a lot of this. We want to move toward working with the editor community to help us ensure that things render well on mobile. Working on the TemplateStyles extension. Working with Coren. Starting outreach to community members.

We're probably looking realistically at next Q before editors start using this tech. We want to get it relatively stable this Q.

Slide 30 edit

Slide 31 edit

Nirzar: When we did the new reader research it was a great opportunity for design to identify our problems and try to solve them.

One problem: Our brand is not widely known even though a lot of people use us.

Slide 32 edit

Nirzar: We plan to add (localized) Wikipedia branding to the website.

Slide 33 edit

Nirzar: What's so special about this: The process: Started with generative research -- so we know this is a valuable problem to solve Design explorations with prototyping and testing Quantitative and qualitative analysis

One of the things we wanted to make sure doesn't happen is a decrease in searches due to the more prominent branding drawing attn away. Early testing showed this didn't actually happen.

This slide doesn't cover a community outreach aspect which we also did.

We'd like to follow this moving forward.

Toby: This is the SOP for the Reading team. We're going to build the products users want and they're going to be great.

Nirzar: There was a lack of ? with in the team; something to work on next time.

Wes: Can the results here be carried across all mobile platforms?

Nirzar: We've tried -- for example, people thought the explore feed on iOS was not branded or confusingly branded. We applied this lesson to Android.

Slide 34 edit

JK: New readers project. Our goal is to launch one experience for new readers.

Slide 35 edit

Anne: At the end of 1Q last year, we got our three focus areas (discussed above).

Affordability: Haven't started working on this yet.

Other two areas (awareness, offline): We're working on structuring our high-level thinking in these areas. Zach is running some stuff around campaigns in the spring We're working with the Grants team to incorporate awareness into their rapid grant cycles

Offline: We're working on mapping the whole ecosystem

  • Advising a grant for Kiwix
  • Working to understand other use cases

Toby: I want to congratulate you and the new readers team much further than we first envisioned. The collaboration with movement partners is awesome.

Slide 36 edit

Anne: In a tactical sense, the objective is to launch one new experience aimed at new readers.

We're working on a feature involving saving articles for later. Going to hand off to Olga in 2-3 wks.

Katherine: Just to second that - we talked about this work throughout WikiIndaba, and clearly of major community interest

Slide 37 edit

Slide 38 edit

Olga: The first thing we're going to focus on this Q will be our print styles. The ones in place now are outdated in terms of style/typography and not optimized for mobile reading.

We want to update both our desktop and mobile styles. Currently, mobile users also get the desktop version which isn't very helpful.

Slide 39 edit

(illustration of updated desktop and mobile versions)

JK: Print out an article with the existing functionality to see what an improvment this is

Katherine: How will section collapsing work? (Generally, cost of access/printing is an issue, (here, reducing cost by selecting content) are you providing for this?)

Olga: We haven't thought too much about that issue, testing of ? didn't get such a good response so we'll probably go toward testing the full offline prototype

Anne: We were talking about this when brainstorming about concepts, but didn't choose to spend time on it before knowing whether the general area was useful for people.

Wes: Looks nice!

Slide 40 edit

Olga: Next thing tied to modernizing mobile web is defining what an offline-capable mobile web would look like. We'll continue to research how we'd need to change the mobile site to make it more offline-capable.

We want to be careful b/c a lot of these things will require larger planning/technical decisions. We're taking the time to research first; we'll produce a written report with our data/findings and then define a way forward.

Toby: Ops has asked us to update Offline Content Generation (OCG) so that will play into this as well.

Slide 41 edit

JK: Community of Readers: Working on building interactive features for readers that will also help editors. Mobile-friendly contribution mechanisms that don't require a lot of context.

Slide 42 edit

Dmitry: Last quarter on Android our goal was to ship a feature to allow users to edit Wikidata descriptions for articles. Description editing stood out as a viable candidate to start with. The feature is pretty much complete and should ship in a couple of weeks, after some design finalization and testing.

Slide 43 edit

Slide 44 edit

(Illustration of the description editing feature)

Dmitry: It's pretty addictive after you've done a few!

Slide 45 edit

JK: Maintaining core readership functions: We also spend a lot of our time making sure things keep working.

Slide 46 edit

Slide 47 edit

Gadgets and bot operators generally access our content via the Action API.

Recent initiatives:

  • exposing page view data (Will roll out over the next month)
  • exposing ORES scores

Q3: we'll continue on the ORES side of things. Looking at the ORES platform with Aaron and Amir to identify gaps and to ensure we're prepared for the future.

Toby/Katherine: Do we have usage info for these services/APIs?

Adam: Generally or the new stuff specifically?

Toby : new

Adam: Haven't run a report specifically on the new stuff. We haven't done a lot of promotion of these endpoints yet, that's something we should do later this Q

Slide 48 edit

JK: OCG is failing. It's been in a long decline. We've been working with services/ops/WMDE to identify a suitable replacement. This will have implications for PDF rendering on all projects. We'll be doing a community consultation to make sure folks are OK with the implications of what we're working on.

Slide 49 edit

Slide 50 edit

Josh: In the apps, before, we didn't have a good way to reach our readers. We've developed an announements service that allows us to insert artibrary cards in the feeds.

Couple of uses already:

  • We've used it to do a quick survey about why people were using the apps.
  • We used it for a December fundraiser.

Slide 51 edit

(announcements illustration)

Slide 52 edit

Slide 53 edit

Josh: We've also been working on usability. Consulting with a blind iOS developer. Making sure the app is accessible to visually impaired people.

Slide 54 edit

Slide 55 edit

JK: This slide summarizes learnings across the teams.

-Great support from the Services team

-The apps and services teams hit a cross team deadline on getting announcements into the feed for fundraiser time

-Hovercards rewrite: It was a skunkworks project for a long time but we think the rewrite was the right decision. Really accelerated development.

Toby: "rewrite" can be a daunting word for managers, but a lot of the background work didn't need to be thrown away. Not like throwing away 75% of our work. Good learning for managers.

JK: Things we'd do differently:

  • We'd budget more time for community consultation if we could do it over. It's hard to predict time, effort, response rate.
  • Analytics: We want to be data-driven but there are challenges around privacy, infrastructure.
  • Recruitment for testing and evaluation is a challenge.
  • We did a reorg of our engineering team at the same time as we were figuring out how to build services within that team.

Michelle: you mentioned challenge of multiple communicaitons channels, community suggestions falling off the radar because of that?

JK: We've been maintaining a wiki to pull information from various sources onto one centralized hub

Toby: normal communication rules apply here, need to meet people where they are . What Jon said about pulling info from disparate locations into one central location applies.

How do you reach people outside the wiki? Mel Kramer's work applies here, as does the new readers work.

Heather: Thanks for the work on awareness. How have you mean mesuring efforts?

Nirzar: I can talk about the protocol for the usability tests. We were mainly testing short term memory stuff. JonK: working with Zach on it. let's take question offline


Slide 56 edit

Danny: Community tech's work runs on the calendar year because of how the survey works. Q2 is the end of our year. It's this time we do the survey that tees up our work for Q3.

Slide 57 edit

Danny: We finished most of our top 10 wishlist before the Q started, just a few things left over. Shipped plagiarism detection before the Q started. Created for en in Q1 and then internationalized it. Live on fr, cz, pt. In feb we'll be offering it to a bunch of others.

Cross-wiki watchlist is the other thing, it's gone longer than we expected, we'll still be working on it going forward. Because it's a huge project. But people (in the community) really want and expect it, so working on it is in our team's mandate.

Slide 58 edit

Danny: The real struggle is that X-wiki watchlists and some other things rely on DB changes to really big tables. Working closely with Jaime Crespo (DBA) (i.e., bugging him a lot). Looking at 2017, there are several things that are also going to require heavy db work. Going to need to meet with ArchCom, Tech Ops to define what's doable in a year.

Ryan: A lot of the features in the survey were features that had been asked for a long time. Changing the size of the edit summary field, for example. The reason they haven't been done is that they've been blocked on DB changes. We're oging to have to work closely with the DBAs to figure out how to prioritize this work, and how to ensure the work is done without negatively impacting the projects.

Slide 59 edit

We also conducted the 2016 wishlist survey, it went great. Now we're looking at the top 10 tasks for this year, figuring out a general order. We'll hopefully finish that up in Jan, dig in in Feb. Listed here are things that will probably be worked on first.

  • global prefs
  • Notification of unsuccesful login attempts (security feature)
  • Fixing Xtools (an important volunteer-developed project for which the volunteer is no longer active)

Q: What does Xtools do?

Ryan: It's a swiss army knife for admins and power users. Mostly around analyzing editors -- used heavily when trying to evaluate admin candidates. People use Xtools to analyze their editing history -- have they been consistent? what namespaces do they edit in? etc. Also tools for analyzing articles in various ways.

Slide 60 edit

<wishlist survey results (thru 61)>

Slide 61 edit

Slide 62 edit

Danny: We had some extra time/bandwidth to work with smaller editor groups. (i.e., not just English WP.) Dont' want to neglect small groups of editors/admins with nonetheless important needs.

Slide 63 edit

Ryan: Examples of such things we've worked on this Q: shipped improvments for admins, campaign organizers, new page patrollers, wikiprojects.

Maggie: Wooo! Blocking tools!

Toby: There's also been work on pulling product into reviewing grants. Building up the muscles of collaboration between these teams. Something we've not done particularly well in the past.

Maggie: I'm very grateful. This has been v. beneficial.

Danny: Admins/stewards/so on - we'll be working with them on new tools

Also working with communities on tools for evaluating programs, gathering statistics

We're still struggling with a way to make the process of what we're choosing more transparent. Up to now, people told us about a bunch of needs and they seemed important. We need to be able to better explain why we choose one thing vs another.

Slide 64 edit

Slide 65 edit

Bryan: This Q I worked a lot with the Tool Labs community on-wiki to develop a takeover policy for abandoned tools, and a right-to-fork policy. And to establish a community-based committee to oversee those policies and understand how to comply with them and take admin actions that come out of them.

Basically a two-year process that wrapped up yesterday.

Also did 2nd annual end-user survey. Got good quantitative and qualitative feedback that we can compare with the previous survey results from 2015. In upcoming quarters we'll focus on documentation, which was mediocre and still is.

Slide 66 edit

Slide 67 edit

re Striker, work will continue on improving workflow re: administrivia on creating an account, uploading ssh keys -- basically a wizard system for the onboarding workflow

next quarter i'll continue with the same project for tool account creation itself. One of the main focuses will be around the right to fork policy. Each tool can declare a default license in a way that will allow us to share it.

Slide 68 edit

Slide 69 edit

We're working on creating evaluation criteria for a new Platform as a Service (Paas) layer -- a workflow wrapper over the Kubernetes backend system, to establish the patterns and make it easier for people to publish their tools.

Slide 70 edit

Slide 71 edit

Slide 72 edit

Slide 73 edit

Slide 74 edit

Slide 75 edit

Slide 76 edit

Toby: We created a services team in the Reading Infra team. The team had asked for more resources for services. We talked to the various teams to gather feedback. Be deliberate: take your time but don't be slow. Don't surprise anyone.

=== Slide 77 === ->

Ran out of time. Metrics will be discussed in a separate meeting.

Chat from BlueJeans: 3 poke-e-man!

Wes W Good morning


michael M should i be hearing audio right now?

Brad? B Hmm, annoyingly touchy "leave" button

Wes W yes

I hear them

michael M thanks, i'll try reloading

michael M better!

Olga O test? I can't hear anyone anymore

Quim Q There is plenty of noise from the room

If you cannot hear it, your audio is kaputt.

michael M yeah, you have to select an audio source (not just "view screen share only"), activating your own microphone is a different option

that's how i messed it up the first time anyway

Wes W Great logos

Trending Api is exciting to see

Maggie Dennis MD Meatspace. <3

Katherine K Is the map search using work done on OSM integration, or?

Wes W my first push notification received was for the election results

Katherine K typo in top articles near(y) you

Wes W curious on licenising issues with that

Corey Floyd CF Possibly address current notificatons on iOS vs push notifications service?

Wes W Sorry Corey - was refering to Lic in using different tile service

Maps UX look great btw

Josh J ^props to Rita and Carolyn

Dan Garry DG Not necessarily just on slow connections... it happened on my phone even on LTE

Glad to hear that's being worked on

Katherine K I've experienced it too - didn't know that was wht it was called.

Wes W Very obvious

Like that it will change per language

michael M you're cutting out anne

Katherine K Just to second thst - we talked about this work throughout WikiIndaba, and clearly of major community interest

Anne A +1, really exciting to work with community members here. i'm stoked about the future


looks nice

Assume this all intersects with the offline content generator (OCG) work?

can answer offline

Anne A yes, wes. it's using electron afaik

Wes W Excellent just confirming - we should probably update the community around that - but sure that is already part of the plan

Dan Garry DG It's not just you Dmitry ;-)

Wes W Really great to see this

Katherine K Very cool!

Dan Garry DG It's also aligned with "the encyclopedia anyone can edit"

For the longest time, that description was the only piece of content shown in the app that wasn't editable

Katherine K I impressed some folks at WikiIndaba with in-app editing. They didn't realize it was so simple as to tap the pencil.

Dan Garry DG Fits with the mission!

Katherine K That's super cool

Is there a fire alarm going off?

Josh J yes, outside

Katherine K Ah, okay. As long as not in the office!

Maggie Dennis MD It's making it a tad bit difficult to follow. And I hope the office would evacuate

Dedication can go too far.


Max M that hovercards rewrite is being held up as an example of a success as part of the tech debt convos with Victoria et al

Anne A motorcycle revving, fire alarms, and car alarms are much-loved by the microphones

Maggie Dennis MD Thank you, Max.

Wes W remote realities

Katherine K New office on the way

Wes W

+1 Katherine

Max M Maybe 1 Montgomery will have better insulation...

Dan Garry DG FWIW I can still mostly hear you, although it is very annoying

Quim Q "It's getting hot in here"

Dan Garry DG The alarm that is, not what you're saying :-p

Quim Q Did Jon mention that Zareen is in the call? Hi! It was a pleasure to meet you at the Summit! Good work!

Zareen Z Hi Quim! great meeting you as well

michael M I can't really see who's talking or hear what's being said

Katherine K Michelle, asking about internal communictions challenges

Quim Q (Last side comment: was a great read. Back to the meeting)


Zareen Z thanks!

Anne A the community members at Indaba are much more connecting off-wiki as well

Quim Q Technical Collaboration Guidance is addressing these expectations on consultations, fwiw.

Anne A so it's even more fragmented

Quim Q But a lot more work needs to be done, agreed.

Anne A thanks josh

thanks team! need to go to airport

Ryan Kaldari RK Well, not longer than I expected

Quim Q


Katherine K

you guys didn't realize you were the architecture advance team... s Wes W Really like the category break outs

Maggie Dennis MD Wooo! Blocking tools!

Quim Q Being transparent about the bribes you receive?

Katherine K Is the takeover process now live?

Maggie Dennis MD Thank you all!