Discovery/Retrospective 2016-02-22

Review action items from previous retrospective

(Previous retro was 2016-01-25)

Dan: Follow up on the common terms query A/B test
- NOT DONE Relevant but not pressing Still worth doing (was targeted at reducing ZRR but certainly with a very limited impact).
Mikhail: Look into listing features that affected the results set for a query (sister project to 'query categorizer UDF')
- In progress. In very final stage of getting the query categorizer UDF into hive
Deb: work on road map for A/B/C testing
- Made progress. Created a doc. Need to get with the team to finalize
Deb: work on Quick Surveys
- Made progress. Working with reading team and ops. Blocked by external people.
Deb: talk with community (mxn and Edward Saperia) portal testing and how fast we can launch new things
- Talked to Edward. Tried to reach mxn. Will try again.
Moiz: Allocate UX for maps (based on priority input from Dan, which would require clearer definition of what is required)
- Nothing done officially. Julien has made a suggestion for a few minor improvements. Waiting for priority information from Dan.
- Julien helped with graphs. Suggested a fix for scrolling issue, but no work yet.
- Request for Yuri to create a phab ticket.
Tomasz: Clarify to the team the long term plan for maps
- Working w/Yuri to draft 2016-17 roadmap. Not finished yet, but that will be the vehicle. Still on Tomasz’s plate.
- Yuri interested in combining search/WDQS/maps rather than keeping them separate.
- Dan will schedule a conversation
Kevin: Consider splitting up retrospective next month (17 invitees)
- Discussions in progress
Kevin: Work w/teams to ensure that meetings (e.g. sprint planning) are as effective as possible.
- Portal planning has a new standing agenda; Search waiting until phab board has been changed

What has happened since the last retro?

(Previous retro was 2016-01-25)

Refocused on relevance lab / reviewing current cirrussearch config to meet Q3 goals
Guillaume joined the team
Updated Portal to have inline JavaScript be as a separate file and analysed the results
Progress (but not completed) the necessary pieces for rolling out completion suggester as prod feature. Mostly waiting on analytics parts now.
Integrated TextCat language detection into code
Budget planning for core and strategic goals FY16-17
Elasticsearch upgrade to 1.7.5 in progress (more is happening than has happened)
Progress defining new metrics around search satisfaction
Lots of scandals
Dashboard updates with cool features! :D +1

What went well?

Relevance Lab improvements went well. We added several new features and tools.
(Guillaume): Lot of support from David (right timezone), Erik, the whole team and a lot of the Ops team to find my way around
Max’s comment about the search engine/KE/Knight history is recognized by those who have seen it as the truest version of events. +1+1+1+1+1
Chris has had his first month at WMF a bit of a challenge - he’s done fabulously. :) +1+1
TextCat! (language detection library)
- Ported from perl to php. It’s ready for action. Went well
  - we should blog post and announce this
  - (some discussion about before/after we see it work, and whether appropriate for a general non-tech blog)
- https://www.mediawiki.org/wiki/User:TJones_(WMF)/Notes/Language_Detection_with_TextCat
Something about the javascript tests on the portal? Assuming they went well. :)
- The tests went well, have presented to team and community and will do another preso tomorrow with WMF Web Team. Big thumbs up to Mikhail!
- This was not an A/B test. Just a change to measure javascript availability/usage
(Chris): The conversation we’ve had about the “turmoil” have been constructive and instructive (for me). Despite what a lot of people seem to think, I think it is a great time to join the team.
- We have seen our commitment to the projects and movement
- This is a comment on me arriving in the team at this point, most probably much harder to get through for you who are into this for a long time....
Supporting our team by our team - we’re in this together :) +1
Dashboard updates now show more information than ever before - so useful for future work (what browsers visitors use, the percentage of countries, JS and image downloads on the portal. +1
Popularity score seems to be streaming into elasticsearch indices on a weekly basis. Would be nice to add page rank there too ;)
- we should blog post and announce this +1
  - It’s only currently used for completion suggester, would be good to get it at least on a testing basis into full text as well before announcing. I think.
I submitted my first bug to the team from the community! - C
- I made my first wiki page from scratch...and didn’t kill anything in the process ~Deb
Justin Ormont giving useful suggestions related to search metrics on mailing list and IRC
- He had been talking with Nik and James earlier
Getting more community engagement and feedback on WDQS, people start complaining about it not being stable enough (which means they care). Also GUI is improving a lot with WMDE help.
- Consider raising status above alpha?
  - It’s beta, not alpha! (Beta means betta’ than nothin’!)
- Should we publicize usage numbers via blog post or other?
- Lots of interest in SPARQL
Significant demand for “maps” becoming available for Wikipedia. It’s in beta-cluster now, should be there soon!
Huge uptake on graphs of pageviews - within days almost 50 usages in enwiki alone, and great interest in the political graphs (pageviews for each candidate)
- could be another blog post here
  - MORE WORK!? -C :) MOAR!

What could have gone better?

The whole “knowledge engine” / Knight Foundation kerfuffle +9000
- kill the KE? :)
  - The term or the initiative?
  - Max: If we continue to use this term, we need to come up with a concise no-BS definition
  - Tomasz: Labels probably aren’t helpful in this case. KE is a deprecated name.
- We answer with a lot of “No”s. Not enough “Yes”s in my opinion ~JG
  - We dismiss a lot of past-ambitions. What’s left for our long term strategy? ★★★
Clear communication about the inception of Search and Discovery by executives
Our communication outside - should have weekly’s and maybe more announcements. If we have something to brag about, let us brag. +++++ (wikidata style) ★★★★
- Office hours, anyone?
  - Needs a different forum? IRC got basically no participation
    - with all the attention we’re getting now it may be different
      - Hmmm…. -C
      - Let’s ask Rachel how much useful it is for other teams ATM +1
- See my note about status updates below in new action items
We still haven’t got any of the portal improvements actually out.
- We updated the stats - will do those again but need a better long term process
- Deb is keeping the portal improvements page updated and hoping for community feedback
Also have not shipped, to full production usage, anything for Search Q3 goals. But much progress is being made.
All those markers in Google Docs make it hard to read this retro +1
- Etherpad? +1 -1
- Coggle (https://coggle.it/) 2D makes for interesting discussions
- Kevin: Consider other note mechanisms (but maybe after we redo retros?)
We have not published our projects and how many people are working on each ★
- For general transparency, we could say how many people are working on which sub-projects
- here? https://www.mediawiki.org/wiki/Wikimedia_Discovery
- Consider adding the information to the page, but need a maintenance plan

What didn't fit into either of those buckets?

Personal health during turbulent organizational times
(Guillaume): There’s a lot of info to digest, from network, encryption, nginx, varnish, continuous integration, integration testing, PHP, Ruby, Puppet, Jenkins, IPSec, IRC, RCStream, MediaWiki, Logstash, … It’s fun to learn, but still not sure which part I should forget and which part I should not.
(Guillaume): Lot of ideas from any new pair of eyes should be written down and centralized somewhere. On my side, I wrote a first email, but there is a lot more content that could be written. Probably Deb and Chris have a similar position.
- Should we add more “document” tasks to onboarding docs?
- probably just a pointer to where to document what

How many projects should we pursue as a team ★★★★★★
- Oliver: On the question "what are we doing about maps?" the answer involved maps, graphs, search, portal, and WDQS. we should be narrowing down. That's what we should be doing. if the answer to "maps?" is "maps and WDQS and graphs!" that's not acceptable.
  - Just a quick reminder, the general impression of volunteers and quite often staff too is that we have too much resources allocated, not that we’re a small team
- Stas and Yuri proposed that we should continue pursuing multiple projects because we have a very large team, and see which project get traction in the community, and increase their resourcing proportionally.
Nik (manybubbles) wrote a blog post about how to utilize our CirrusSearch dumps - https://www.elastic.co/blog/loading-wikipedia
- This should be mentioned on the public mailing list - by someone familiar with it

Bluejeans destroys my laptop battery when not plugged in
- There is a bluejeans app, which has a “low bandwidth” mode
Discovered this help page: https://en.wikipedia.org/wiki/Help:Searching/Draft
What do we think about combined showcases/lightning talks/etc?

Discussion

K
- We need to own communications about our own work
- We were already communicating well on our day-to-day work. We had been told to let the execs own the big timeline communications, which with hindsight was a mistake.
How many projects should we pursue as a team
- Problem is that we started off a year ago with 5 different projects. We put 2 or 3 on the back burner, but we are still working on 5. They are still part of the conversations. Worry that we are doing what the foundation has always done: We don’t like killing things, or saying we’re not going to work on something actively. We continue to work on things a bit. Make sure that we are not split 5 ways. If we don’t want to spend resources on something, should we [notetaking fail]
- We have been bad at killing things that don’t work. It would be less optimal to kill things that do work.
- We should stick to our Q goals. Stay focused
- There are several types of projects: Those that are successful with budget, successful without budget, and not successful(?). Budget is controlled by management. What about a project that is hugely supported by community but don’t get the budget.
- Are we taking on so many projects that we don’t have resources to support them. If so, then we are not a sustainable team. If they are on the back burner and don’t need active resources, maybe that’s ok. Probably not a solvable question in the 2 minutes we have left. Good topic as we look at upcoming annual and quarterly goals.
- If we have enough people, but not enough hardware, shouldn’t we re-balance that somehow. The two should be in alignment with each other

New Action Items

Chris? Follow up to Yuri’s idea on status updates for Discovery (ala Wikidata - https://www.wikidata.org/wiki/Wikidata:Status_updates)
Kevin? Mail list conversation about “How many projects should we pursue as a team”
Yuri: Request for Yuri to create a phab ticket about maps UI needs
- https://phabricator.wikimedia.org/T127844
- https://phabricator.wikimedia.org/tag/kartographer/ - VE column
Dan will schedule a conversation about long-term maps plans
Chris: we should blog post and announce this (textcat)
Chris: we should blog post and announce this +1 (popularity score)
Chris: Should we publicize usage numbers via blog post or other? (WDQS)
Chris: could be another blog post here (graphing pageviews)
Chris: Let’s ask Rachel how much useful it is for other teams ATM +1 (office hours)
Kevin: Consider other note mechanisms (but maybe after we redo retros?)
Chris? Consider adding the information to the page, but need a maintenance plan (who works on each project)
Tomasz? Should we add more “document” tasks to onboarding docs?
Chris: This should be mentioned on the public mailing list (Nik’s post)
Kevin: ask for feedback on this retro