I probably should have written a blog post during the summer, displaying Angadh’s work . . . oh well.
SymPy sent me as one of 3 mentors to the GSoC Mentor Summit this year, at the Googleplex. I was very excited to go, meet some of the other SymPy developers (Matthew and Stefan), and meet others in the open source community. My overall experience was a little mixed though.
Matthew covered some feelings/conclusions I also share in his write-up on the event, but I have some others.
Meeting and interacting with people in the open source community was probably the highlight of the trip for me. It was interesting to talk to a few people who had taken projects that were originally academic/research code and translated them into more successful (open source) products. I’m not sure if it will be relevant to SymPy’s future, but it might be to mine. A common thread seemed to be: jumpstart a project with grants, keep improving it for long enough with that money, and then put it in a position where it can join an umbrella organization or can be used by professionals who will pay for support. I certainly don’t speak for any other SymPy developers, but I’ve never got the feeling that this was the intended trajectory of the project (at least the selling support part).
There was a talk on how to structure student/mentor/organization interactions for future GSoC projects. Some of the organizations have a much more defined structure than SymPy though, and benefit from things like daily group meetings. SymPy seems to have a more distributed organizational structure though – there are a lot of different modules, that all have some independence from each other. Despite this, SymPy’s code base is of very high quality, with credit going to the review-process/reviewers and the high standards that are enforced.
The talk on forming non-profit organizations was also interesting. One major takeaway was that the IRS would rather you put your project under an existing umbrella, rather than grant you 501(c) status, due to the potential for abuse. Also, a lot of work is involved in managing money properly once a certain amount of cash flowing through. Although there are hurdles, getting to form a board of directors sounded interesting. Scheduling board meetings could also be fun (with the money being spent responsibly, of course…).
On the less positive side of things: the unconference format. I think that it could have worked a lot better. Again, Matthew touched on this in his post, but all of the sessions (that were not just presentations) were very unfocused, with some more productive than others. The lack of moderation was a serious impediment to keeping the discussions on track. There were definitely a few times were people would hijack a session to try and talk about or show off their project. While I don’t have a problem with people showing off their work, there was limited time for each session.
The time limitations on the sessions was another issue. Each one was ~45 minutes, and almost always there was another session waiting for you to leave the room so they could start. I think in a few of the talks I went to, people would have been happy to stay in the room and continue to discuss the subject. Perhaps the expectations, of the session initiators (and myself), on what we would accomplish were too high. Perhaps the 45 minutes should have been spent networking with other people thinking about that topic, and then spawning a more detailed discussion in the future? Again, I think moderation would have helped this.
There were also some talks, with interesting sounding titles, that were just unproductive. There was too much recounting of what one organization did, with little generalization to what others could use. There was also little consideration of what decisions were made which led to important decisions; e.g. the fact that a project was split into parts A and B which were developed separately was recounted, but not what went into making that decision. Perhaps another example of expectations being too high…
I’m certainly not going to write off the unconference format – I think it could have led to some really cool things. But I don’t think I will attend another one that has sessions which are so short and under-moderated.