Wikimedia Discovery/Meetings/Search retrospective 2017-04-18

The Retrospective Prime Directive: “Regardless of what we discover, we understand and truly believe that everyone did the best job they could, given what they knew at the time, their skills and abilities, the resources available, and the situation at hand.” — Norm Kerth

Action items from previous retroEdit

  • DG: Ask android team about gradle [DONE]
  • DG: Consider dependencies between quarterly goals (e.g., language analyzers needed to wait for ES 5 update) [NOTED, and mentioned during the Discovery quarterly review]
  • GL: Set up gradle (maven) meeting within search [DONE]
  • KS: Start email thread about having separate frontend/backend meetings [NOT DONE]
  • This action item, which I didn't get around to before the tuneup was announced, seems brutally ironic now
  • KS: Include a "tiny things" section in the next retro, but switch the format [DONE]
  • KS: Send email about items that got votes but weren't discussed: [DONE]
  • [2] **How do we feel about quarterly planning process? Current process is ad-hoc and light, and works.
  • KS: Facilitator should take an extra 5 minutes during retro prep to dedupe the "what has happened" list :) [DONE]

What has happened since the last retro (2017-03-28)Edit

(Pasted in from )

  • Jan presented the Wiktionary widget during the CREDIT showcase
  • Posted about the upcoming sister projects snippets in search results on various Village Pumps (task T162064#3161941) and to a few email lists (task T162064#3161951)
  • A blog post about Discovery's Search team was released, along with a Twitter and Facebook post, detailing what we've done over the last several months and what we have coming up next.
  • Product/tech "tune-up" (tuna)

What has been going well?Edit

  • Maven discussion was interesting
  • Elasticsearch upgrade to v5 (is completely done)
  • The Wiktionary widget is awesome
  • DataCenter switch seems to be a non event for Cirrus (famous last words, the actual switch is tomorrow)
  • Community communication has been good - mostly all good feedback for search (blog, fb, twitter)
  • Turns out that Discovery is a well formed and cross functional team that is highly performing - we just needed to tell people.
  • Named something mjolnir (meow meow)
  • Interactions with o19s (opensource connections elasticsearch learn2rank plugin) seems to be going well, they’ve accepted a PR (pull request)
  • Mr G, Mikhail and Chelsy had a couple good talks on Puppet
  • I [Dan] have felt very few changes as a result of going remote. Cool!

What has not been going so well?Edit

  • 2 [dt][DC] It turns out that Discovery is allergic to tuna. (who knew!?) [NOTE: "Tuna" was an internal nickname for the "tune-up" that was announced in June 2017]
  • 4 [EBj][TJ][dt][EBe] RfC for sister project snippets has raised a bunch of old issues (rather than sticking on topic)
  • 4 [TJ][EBj][DC][EBe] Learning python / pyspark deployment/testing/etc. taking more time than expected. But moving along well enough. Not sure how to address bus-factor (lottery factor) here.

What else is worth noting?Edit

  • It snowed in Warsaw today
  • 1 [Ebj] My mail delivery dude is a bit of a jerk
  • 1 [JD] Solar systems work surprisingly well in Dublin
  • It keeps raining in San Francisco. What the hell, supposed to be sunny in April.
  • 4 [TJ][DC][EBe][JD] Quarterly check-in on Monday went well

Is there any tiny thing that we could experiment with? (People, Process, Tools, etc.)Edit

  • 1 [dt] There are large things that we could will experiment with, like splitting the team in half (pieces).


  • 4 [EBj][TJ][dt][EBe] RfC for sister project snippets has raised a bunch of old issues (rather than sticking on topic) (9:21)
    • RFC was created by a community member, in response to our posts
    • Concerns from community members about images (non-free, relevant, offensive), and the question came up during yesterday’s Q review too. So we’re looking at not showing images for now (will display just a link to results or maybe nothing at all for multimedia results other than what is already there)
    • Could deploy on other wikis first, en later, but would prefer to go all at once
    • Possibly switch to title-only search for wikivoyage
  • 4 [TJ][EBj][DC][EBe] Learning python / pyspark deployment/testing/etc. taking more time than expected. But moving along well enough. Not sure how to address bus-factor (lottery factor) here.
    • I share the concerns. Maybe should spend more time on python?
    • Learning as I go.
    • Suggestion: Do code review together? Maybe in the weekly search tech meeting
    • Nobody really feels comfortable +2ing the code right now
    • Would it make sense to pair program in the future? [Maybe]
    • A lot of the code is SQL concepts but in imperative form
  • 3 [TJ][DC][EBe] Quarterly check-in on Monday went well
    • Do people know what we’re doing?
    • This quarter was similar to past: “Looks cool! Sounds great!”
    • It’s obvious when a quarterly checkin goes badly; harder to tell when it goes well
    • Previously, lots of questions was a bad sign, but doesn’t seem that way now; seems like people were paying attention
    • The questions were clarifying, not challenging. Sat in on an earlier checkin by another team, and there were no questions (just a positive comment).
    • Katherine did acknowledge the challenges of not having a decision yet on the maps team future
  • 2 [dt][DC] It turns out that Discovery is allergic to tuna. (who knew!?)
    • Sounded like that meeting was a lot more clear, which is good, even if the news wasn’t great
    • Should we start talking about how we will work together in the future? Looks like the decision is pretty much done, so doesn’t seem the time to challenge it.
    • Before the meeting, we asked for boundaries about what they were flexible on. We don’t know whether more work will be coming in from new managers. Are other managers supposed to be hands-off with new people? We think so, so Discovery members can continue working on what they have been. Toby is out this week, so the conversation about analyst placement has stalled.
    • I get the sense that they’re not super-focused on details of how we get work done; they’re leaving that up to us. They are focused on where everyone will be when they push the button. We still have the offsite, and we were already going to talk about the future. We’ll talk about the workstream and how to coordinate.
    • We’re trying to get people like Jon Katz involved in the offsite (by hangout or in person). The intent is that they would be in listening mode.
    • Any concerns about not having a product manager for search backend? [Yes]
      • Idea has been floated that Erika would take care of that. Ultimately hoping to get a PM.
      • If you don’t have a PM, there will be problems with people understanding how quickly work can be done. Getting a handle on capacity. Someone has to do that.
      • Prioritization becomes a big issue.
    • Don’t want to lose community goodwill that Discovery has built up. Asked whether we could keep the name Discovery. He (Toby) felt the name has a good reputation. If a team isn’t called Discovery, maybe we can at least keep it associated with projects for the next year.
    • Unless something changes, as of May 8, neither front nor back end will have direct TPG support
      • Final TPG decisions haven’t been made; hope to have a better idea tomorrow
    • Any news on Chris?
      • He expects to continue to work with us (as of last week)
    • Offsite planning?
      • By design, it was controlled up-in-the-air. Not practical to plan specific topics months in advance. So not that difficult to change the main topic.
      • Once the tuna is publicized, will people view the offsite as a boondoggle? Backlash?
      • I feel like it’s more important than ever, especially since mgmt has been clear that the parts will continue to work together. It’s fortunate we have it planned.

Action ItemsEdit

  • KS: Send email about voted items that didn’t get discussed
  • Follow up on lottery factor related to pyspark etc.
  • KS: Let KH/EB (and others) know as TPG decisions become firmer