Hi Everyone,
Thank you all for your efforts. However, my sense is that this attempt to organize elections isn't functioning well and is unlikely to succeed with the current direction. I think that it's time for NumFOCUS to try a different approach. I'll include some observations and recommendations for a next iteration below.
One Election
I would focus on running a single election, rather than on trying to form a committee. We lost a couple of weeks on organizational minutiae that (in my opinion) were not important to the critical path and burned momentum.
Besides I don't think we have enough knowledge today to know (or care) about how best to create a multi-year committee like this. There's a lot of learning ahead of NumFOCUS I think.
Execution vs Perspectives
The committee had a dual goal:
- Give external perspective on the election in order to avoid bias and improve legitimacy
- Execute the election
These are very different goals, and they probably need very different groups of people. In order to avoid bias / improve legitimacy I think that you want a broad set of people to observe and weigh in from lots of different backgrounds. There also isn't much work required or coordination here, so having a big group isn't harmful.
However, In order to execute an election you want a small group of people who have all committed to doing work, and who already work well together. They should all be in similar timezones and it's probably ok if they think pretty similarly (broad perspectives are useful here, but not as useful as above).
For the rest of this I'm going to focus mostly on execution, because that is what I personally was focused on
Committee selection
The process here was an open call "do you want to be on this committee?" and all were accepted. I recommend a different approach.
At the very least for execution I recommend that we at least set an expectation of work. I think that historically the SciPy conference asked "How many hours per week are you able to work on this project?" which was a good filter. I'd say that the election is likely to take ~5 hours per week for a few weeks. Some committees are about talking. Some committees are about doing. This is a "doing" committee and people need to opt-in to that early on.
Additionally, I could imagine other configurations here to finding a set of people to execute the election:
- Pay a group that does this professionally (do these exist?)
- Nominate a single Czar to do it (it's actually not that much work, the challenge is trying to coordinate many people to do that work together) and give them a lot of control (it would be very frustrating to do a bunch of work that is then judged by people who don't do work)
- Engage one of the various professional services groups in our space (Quansight? Quanstack?) to execute the election. For a professional group doing something like this is trivial.
- Have one of the projects / conference groups run things? They already know how to work well together.
- Have NumFOCUS staff run this (although this would make me nervous for bias reasons)
Communication thoughts
This is subjective and many things could work
In our particular configuration we had three places to talk about things:
- Slack
- Github
- Zoom meetings
Personally I was happy with our choice to use Github (even though it failed in the end). Probably we should have chosen one of Slack or GitHub though. Communication was diffuse.
In hindsight we should have had a regular zoom meeting (tangent: even though I dislike these). Doing a Doodle poll every meeting was bad. Also, trying to do this across all timezones was bad.
Recommendation
I'd ask Quansight to do it if they're game. Pitch it to them as a good marketing opportunity. They've been in the space for a long time, are pretty well trusted, and have a strong vested interest in NumFOCUS's success.
This doesn't solve the legitimacy problem, but that should maybe be solved separately from execution (perhaps invite project/meetup leaders to assess the work).
I'm out
Anyway, to make things explicit, I'm no longer planning to do work on this effort. If there's some official way to step down as chair then I'm doing that now. I've already communicated much of the above to @lsilen and @dutc .
If other people who were active in this effort have thoughts or constrictive critiques or suggestions I encourage them to voice them here and maybe whoever comes next will be able to learn from that experience.
Thanks all for your trust. I'm sorry that I wasn't able to deliver.
Cheers,
-matt