Data Org Working Groups: retrospective (2020)

Project

time

1 month

impact

established cross organization groups as a tool for grouping people

Problem statement

Data Org architects, builds, and maintains a data ingestion system and the ecosystem of pieces around it. It covers a swath of engineering and data science disciplines and problem domains. Many of us are generalists and have expertise and interests in multiple areas. Many projects cut across disciplines, problem domains, and organizational structures. Some projects, disciplines, and problem domains benefit from participation of other stakeholders who aren't in Data Org.

In order to succeed in tackling the projects of tomorrow, we need to formalize creating, maintaining, and disbanding groups composed of interested stakeholders focusing on specific missions. Further, we need a set of best practices to help make these groups successful.

Prelude

Data Org had just had a re-org and a bunch of people were now in new teams. There was a concern that as generalists in new teams, it would be difficult to do some of the kinds of work we used to do especially where it's not part of the mandate for our new teams.

This re-org meant that I was solo-ish again. I was mulling over how I can create a good enough approximation of being on a team without a ton of extra work.

Further, I work on crash ingestion which is the other side of crash reporting and this particular system involves teams and individuals spread out across Mozilla. I was trying to figure out how we could coordinate better. Mozilla doesn't have anything I'm aware of to support that.

Anyhow, so I was talking with Mark about that in a 1:1 and he mentioned how he was trying to figure out how to group people in ways that aren't organizational structure. I offered to do a Google doc and collect research and links on working groups and like structures.

That evolved into a proposal for Data Org Working Groups which was accepted and turned into a thing at the end of 2020.

What I found

There are a lot of links out there to blog posts and introspectives on different ways to structure groups that cross organizational boundaries. Many of them are structured in a specific way to solve specific needs and it seemed like they had a lot of bookkeeping and management to keep them going.

Here's my rough notes on research--they're rough and possibly wrong in places especially where the blog post doesn't match reality.

  • Python special interest groups

    https://www.python.org/community/sigs/guidelines/

    • Every SIG must have a clear mission with a well defined conclusion and end date.

    • SIGs shouldn't overlap with other SIGs

    • Every SIG has a coordinator who is responsible for reporting on the SIG activity and for shepherding.

    • SIG activity occurs on mailing lists that are managed automatically.

    • To create a new SIG, you have to request it on the meta-SIG.

  • Mozilla module owners

    https://wiki.mozilla.org/Modules

    • Modules are defined by a significant chunk of code or work

    • Modules have "owners" and "peers" who are responsible for the module, decisions, and governance of the module

    • Module owners can change the owners and peers for a module over time.

    • There is a Firefox Technical Leadership Module which oversees all modules.

  • ACM SIGs

    https://www.acm.org/special-interest-groups

    https://www.acm.org/special-interest-groups/about-special-interest-groups

    • SIGs focus on specific topics and improving the skills of their members.

    • SIGs offer opportunities to network and stay connected with peers.

    • SIGs publish newsletters and magazines, have recognition programs, and run conferences and trainings.

    • It might be interesting to have research-focused working groups.

  • Spotify guilds

    https://www.atlassian.com/agile/agile-at-scale/spotify

    • Guilds are a grouping of people across tribes.

    • Anyone can join a guild. They consist of a coordinator, active members, and passive members.

    • Coordinators don't "own" a guild--they coordinate the guild's activities.

    • Guilds can focus on anything from engineering disciplines (e.g. web frontends) to shared interests (e.g. making coffee)

  • Pods

    https://blog.walkme.com/pod-structure-in-rnd-teams/

    • Pods are autonomous groups of people with complementary skills.

    • Pods are organized around a shared purpose. In the article, it was a product feature. The pod owned all the tasks for involved in their tasked feature.

    • Pods have minimal dependency on external groups--they have representatives for all the things they need in the pod.

How I thought about it

Minimal system

I wanted to create a "working group system" that had enough structure to be tangible.

I was also painfully aware that as a group, we have a lot of work to do and couldn't spare a person to be managing this working group system full time. So whatever system we ended up with needed to be either very light-weight maintenance-wise or self-managing.

The system should provide enough infrastructure to make running a group straight-forward.

Ability to evolve

If we build the smallest possible thing and it's successful, it'll outgrow what we build. It's important that it can evolve as needs arise.

The idea that this system evolves over time should be baked into the system.

This also allows us to push off a lot of decisions to future iterations.

Different kinds of groups

I wanted to support different kinds of groups:

  • Core working groups that have KRs they're responsible for

    For example, "Revenue data working group" or "Firefox health working group".

  • Problem domain working groups

    For example, platform infrastructure working group that owns platform infrastructure for Data Org.

  • Discipline working groups

    For example, Staff Engineer working group where staff engineers can discuss career goals or Python Working Group for discussing Python things.

  • Topic working groups

    For example, Coffee Working Group for discussing coffee making.

All of these have shared system needs and I think we can create a system that covers them all.

Participation

I wanted to create a system that supported groups that included people from wherever regardless of whether they were in Data Org or even a Mozilla employee.

Support all stages of group lifecycle

I wanted to make sure our system supported all stages of group lifecycle:

  • Creating groups

    Have a template for figuring out:

    • Why does this group exist and what does it cover?

    • What kind of group is this?

    • Who should be in this group?

    • What are the deliverables/artifacts (if any)?

    • How does it communicate?

    Have a set of conventions for mailing list/Slack channel/Matrix channel naming and links to instructions on how to create them.

    Have a process for how to go from an idea for a group to an actual group.

  • Running groups

    Have "recipes" for running groups.

    • How to run a group that's primarily public. How to run a group that's primarly Mozilla-only.

    • How to run asynchronous groups that rely on mailing lists and other asynchronous methods.

    • How to run synchronous groups that have meeting cadences or Slack/Matrix channels.

  • Ending groups

    We don't want to be saddled with groups that are dormant or have long outlived their usefulness. We need a light-weight way to figure out when a group has run its course and needs to be disbanded.

Conventions that lead to safe, respectful spaces

We want people to be great. Unsafe, disrespectful spaces destroy that.

This system should codify conventions and practices that lead to safe and respectful spaces by default.

NDA and security-sensitive materials

Data Org often is under NDA for various things. Further, we manage data and other security-sensitive things.

This system should codify conventions and practices that lead to reducing mistakes around releasing NDA and security-sensitive material.

Findable

Managers should know which working groups exist so they can help their team join working groups that are useful.

People should be able to find working groups they're interested in joining or are related to work they're doing.

Artifacts

We're constantly losing historical knowledge--let's work to reduce that here.

This system should codify conventions and practices for capturing artifacts of group decisions. How to use issue trackers, how to keep meeting notes, etc.

What we ended up with

I put together a proposal and shopped it around for a couple of months incorporating feedback as I got it.

The end result is the Data Org Working Group 1.

1

This is a page in Mozilla's Mana which is for Mozilla employees only.

That Mana page covers:

  1. how to define and create a working group

  2. how to run groups

  3. how to disband groups

It includes a bunch of prompts for things to think about when setting up a group and what shape it should have. Is it predominantly synchronous (bad for timezone diversity) or asynchronous (sometimes things take longer)? Is it public (everyone can participate) or private (can handle NDA and security-related material)? Does it have deliverables? Does it own KRs? Does it have stakeholders? How are they notified and communicated with? Where do group artifacts end up? What would cause this group to wind down?

It includes a small list of best practices for setting up and running groups. Conventions for naming things, where we suggest keeping notes, etc.

It includes an index of existing groups with links to their respective Mana and wiki.mozilla.org pages. This makes it possible to find groups you're looking for.

It includes an owner for the working group system.

We have a few groups already that we could convert to Data Org Working Groups.

Next steps

Creating working groups involves creating a proposal and shopping it around. I plan to keep an eye on proposals and see what gaps in the system they highlight.

I plan to revisit and do a Data Org Working Group v1.1 pass in March 2021. I'd like to improve some of the conventions. For example, what if someone has questions about working groups--where do they go to ask?

I think I also want to check in with groups every 3 months or something like that so as to hone the Data Org Working Groups scaffolding and conventions and also to help groups stay healthy. "How're things going? Are you hitting any issues?"

Conclusion

I talked to a few people about working groups, but I wish I had more input from others. I should have set up 1:1s with certain people to discuss working groups. It was the end of 2020 and everything was hard, but I wish I had found the time/energy.

I put a lot of thought into some aspects that I decided to push off to a future iteration of Data Org Working Groups. I don't think that thought was wasted, but I'll feel happier when it gets integrated.

I like that it has the ability to evolve over time. I think that'll give it a good chance of staying relevant, so that's cool.

Want to comment? Send an email to willkg at bluesock dot org. Include the url for the blog entry in your comment so I have some context as to what you're talking about.