Well, I’ve returned from PGCon 2015 in Canada, and after a couple days to decompress, it’s time to share. I wrote about the PGCon 2014 unconference after returning to Chicago last year, so I felt it was only fitting that I start there. I feel as strongly now as I did a year ago, that directly interacting with the PostgreSQL maintainers at this level helps the community thrive. Even though PGCon is generally a more developer-focused conference, being able to brainstorm with the bigwigs, even if nothing comes of it, means the ideas have been given a fair shake.
The format this year was quite a bit different than last year. I attribute this to Josh Berkus once again, in what I see as his attempt to formalize the process and give it more weight. Indeed, it’s hard to argue with his results. Just take a look at the 2015 Unconference Wiki page. It’s a deluge of information I wish we had about the 2014 talks, from attendees and talk summaries, to relevant external links and the all-important schedule. I’m a bit biased in that regard because I tend to produce and consume vast swaths of excessive information, but it’s an excellent reflection on how much the PostgreSQL community values documentation in general.
Unfortunately due to inclement weather, I missed the voting process and the first day of talks entirely. I desperately missed watching the talk selection process, though Josh said they did a lot of that electronically because several people would be late to the conference. I’m not sure how I missed that, so I’ll blame email; it deserves it anyway. Regardless, after witnessing the 2014 talk selection, I stand by my earlier assertion that it’s a sight to behold. It warms my crooked old heart to watch people so excited about technology and the amicable interaction they have with the friendly developers.
Despite the setbacks, I did attend several chats on Wednesday, which is another departure from last year’s format. In 2014, the selection process and every talk were constrained to one long Saturday and was very ad-hoc. I can’t tell whether or not distributing everything across two days is an improvement, but it certainly worked in the favor of anyone offset by rain delays this year. And the information available on Wednesday was certainly copious. Those at the same sessions I watched got a summary of semiconductor storage variants, PostgreSQL and Docker, the possible future of horizontal scaling, how pg_shard is progressing, and how partitioning might be better integrated into the core (my personal favorite). All told, it was a good assortment and most of them were fairly informative.
Through all of these audiences, something felt different from before, and it took me a while to figure out what it was: spontaneity and interactivity. Every session I attended, barring Josh’s own topic on Docker, had slides. It’s extremely difficult to have community discussion or collaboration when there’s a single speaker pushing his or her own agenda. From what I saw this year, the majority of the Unconference was plagued by this propensity to allow the speaker an open forum, as if they had won a bonus PGCon slot. I get the impression that was not the intention of the format, and I’ve heard slides might be disallowed next year. If that’s the case, that’s an impressively prompt response, and suggests I wasn’t the only one who voiced the concern.
I also suspect the unconference was harmed by holding it before the primary conference itself. I think part of what made everything work last year, was that the unconference was a response to the primary talks. Once we had a chance to learn about the current direction of PostgreSQL, potential upcoming features, extensions, modules, kitchen sinks, and so on, everyone could discuss inherent implications. We bootstrapped our brainstorming with all of the exciting information presented in the conference, and organically produced further discussions based on it. Without that, the unconference was just… more talks, and unnecessarily covered some overlapping information. It’s impossible to know everything a speaker will include in a topic, so we were operating blindly in most cases. I think that might have contributed to the passive audiences; there was no direction to work with.
And that really is the challenge, isn’t it? Is it possible to wrangle a group of developers and DBAs into a room and encourage semi-focused conversations without allowing it to devolve into anecdotes and speculation? Yes! But can it be done without injecting a ringmaster into each session to ensure there’s some kind of convergence on the stated topic? I definitely don’t want the unconference to become the equivalent of a moderated internet forum, because that cheapens its impact. This is in many respects a once-in-a-lifetime opportunity for many attendees: to be part of PostgreSQL’s future. I’d hate for anyone to miss it because they didn’t want to listen to a bunch of loose squabbling.
And that’s why I have really high hopes for next year. The trick with anything is to find balance. The first attempt revealed the potential of the unconference format, the second gave it some necessary structure but may have overreached slightly, and the third will combine everything into a cohesive amalgam everyone can love. These people are all geniuses, and I have every bit of faith they’ll obliterate my tiny, little mind with something I never even considered. That’s what they do, it’s who they are.
Me? I just try to pay attention so I don’t miss anything. Join one of the mailing lists so you can, too. Also, don’t be afraid to attend a PostgreSQL User Group in your area. It’s not uncommon to see a committer there depending on the location; you don’t always have to attend a conference to bask in the glow of these luminous god creatures!