Wikimedia Discovery/Retrospective 2015-06-16

Review of action items from last retro (May 6)

Suggestion: Mobile team had a monthly showcase (a very positive/happy meeting)
- DONE: We just had our first monthly showcase, and more are scheduled monthly
Hopefully we can make our team page more visual
- DONE: Team page now has photos
Not sure what the point of some of the meetings is
- DONE: Every meeting has a purpose ( https://www.mediawiki.org/wiki/Discovery/Process#Recurring_Meetings )
Need to get better with mailing list, process, etc.
- IMPROVED: Process is defined and documented. No specific actions taken on "getting better w/mailing list").
Lack of process is preventing us from getting some things done that we should (but is heading in the right direction)
- IMPROVED: Process is defined and documented ( https://www.mediawiki.org/wiki/Discovery/Process )
Getting work that the product owner wants doing actually picked up by engineers
- IMPROVED: Hopefully the new process has improved this
Establish prioritization -- What are our goals? Are we on track to achieve them?
- IMPROVED: Hopefully Dan has made the goals and priorities clear
Unclear how ongoing work (Wikidata Query Service, Maps & Geo) relate to our departmental objectives
- IMPROVED: Hopefully this has been communicated by Dan and Wes
We should all get together somewhere/sometime (ideally not at SF office)
- IN PROGRESS: Wes is still working on this
Need some kind of survey thing where people can say when they'll be available for offsite
- DONE: This was done (google doc, google sheet)
Still no schwag, patches, or t-shirts
- IN PROGRESS: On the way, expected in early July

What went well

Wes gives thumbs up to Dan's talk, other presentations, and team attending them
Feels like we have a better handle on metrics now
Feels good that someone (Oliver) is working on the relevance measurements (at least starting to have a plan is great)
James is doing fabulous summrizations of external presentations
Dan did an awesome presentation!
Prioritising tasks has been working really smoothly by insisting they go through the PdM consistently
State of the Map conference was hugely useful (conference in NYC)
Habemus Scio! (hopefully we have found our data analyst)
Hiring has been going well (aside from when Wes+Tomasz were out simultaneously)
Meetings have been rpoductive, and often quick
Good use of IRC channel (helps make us more of a team)
- Using buffered clients is helpful so people get it even if they're not there (suggestions for Linux/Mac/Win?)
Doing well accommodating remoties
Nice onboarding template by Nik
- +1. I'm using it for the Analyst job as a template.
Feeling positive about recruiter transition (was worried, but seems OK)
Deflected non-Discovery requests away from Oliver with some success
- Oliver really appreciates support from Wes/Dan/Tomasz and others

What could have gone better?

Clarify the association between prioritized tasks and team goals/milestones
- Process for pulling tasks into sprints has not been communicated clearly
Data Analysis is bushed; ongoing tasks suck up too much bandwidth to allow for long-term planningx
- We are working on some short-term work at the cost of not doing longer-term beneficial work
- Non-Discovery work didn't help, but hopefully improving
Data Analysis is only half-scrum; we don't do points and Oliver's hairline would like us to switch (to scrum)
- Another case of process not being communicated clearly
Tuesdays feel like a write-off with all the planning (~4 hrs of meetings)
- Wouldn't save time to spread it out over multiple days
- Nik/Stas prefer clumping them in a single day
- I feel better with one "meeting day" than lots of interrupting meetings on different days
Clear blockers more quickly/frequently -- e.g. designate a scrummaster, or hold more frequent standups
- Kevin is the "scrum master"
- Example of blocker is portal EventLogging
  - Any code review is slow in WMF...requires frequent poking
  - If something gets blocked a long time and nothing breaks, maybe it is lower priority?
We are working on a lot of things slowly, rather than a few things quickly
- process based on code review is a forced pause, so some multitasking is unavoidable
Confusion due to having engineering work in data workboard
Foundation doesn't have standard for IRC buffers) (bouncers)
- It's called email ;)
  - Google Wave?
- If you email techsupport@wikimedia.org they can set you up with an irccloud account. On the downside you have to use their client, but it does the buffering and whatnot.
- If you ask mukunda (irc: twentyafterfour) he has setup something on digital ocean for which he is willing to give out accounts as well
- Max would rather not feel obligated to read logs from multiple channels
Some information droppage (is Bob Flagg doing work for us? )
- Bob is joining us as search relevance eng. Should have been announced.
- When people are out, some stuff might not get delegated
Didn't get David scheduled ito the retrospective
Protecting Oliver and really everyones time from randomization tasks (is improving but ongoing challenge)
Few of us sitting close to each other
Two search outages recently
- Nik had to bring cluster back while tired because others aren't trained up
- Erik volunteers. Max woldn't mind but feels overloaded with maps stuff. Nik proposes David.
- WMF contact list and/or Eng contact list should be kept updated
We should have no single points of failure anywhere
People being sick
Budget has been a moving target
Guiseppe has represented us in ops, but we might need more

What confuses us

Are we hiring frontend engineers? Yes, we are. Thats our one open rec.

Retrospective of retro

Went well. Important.
Should it be half an hour longer?
- Shorter encourages discipline
Yuri needs a Mac :trollface:

Actionable Concerns (pulled from above)

Kevin

Schedule our next retrospective
- DONE. Retrospectives are now recurring meetings.
Clear blockers more quickly/frequently -- e.g. designate a scrummaster, or hold more frequent standups
Process for pulling tasks into sprints has not been communicated clearly
Kanban (not Scrum) process has not been communicated clearly
Everyone should make sure their contact info is up-to-date (https://office.wikimedia.org/wiki/Contact_list)
- Everyone has info now...I can't tell whether or not it is up-to-date
Didn't get David scheduled into the retrospective
- I have made a few changes to my process which will hopefully avoid this in the future
Stacking meetings on one day works well for some people, but not for others
- More people seem to propose stacking them, so will continue with existing schedule for now
Lack of IRC norms regarding buffers/bouncers, and whether people read scrollbacks

Dan

Confusion due to having engineering work in data workboard
- FIXED (engineering work has been removed from analysis board)
We are working on a lot of things slowly, rather than a few things quickly
Data Analysis is bushed; ongoing tasks suck up too much bandwidth to allow for long-term planningx
Protecting Oliver and really everyones time from randomization tasks

Wes/Tomasz

Guiseppe has represented us in ops, but we might need more
- IN PROGRESS: Wes is meeting with ops, and expects ops elastic support person for Q1
Some information droppage (notably when Wes/Tomasz are out simultaneously, or Dan by himself)
- Noted, and will remain vigilant in the future
Few of us sitting close to each other
- IN PROGRESS: Tomasz has proposed a new seating plan
Budget has been a moving target
- Can't go back in time to improve the past, but can improve moving forward
- Wes is approaching grant funding and as issues arise will petition for additional funds as needed.

Nik

Two search outages recently
- Ops wrote up incident reports
- Root causes are being tracked in phabricator
Nik is a single point of failure