How we made the PGConf US 2016 Schedule

Welcome to our annual post on how we put together the schedule for the largest PostgreSQL conference in North America. It is long, thorough, and awesome, so buckle up and enjoy the journey!

For PGConf US 2016, we allowed people to submit for 6 training sessions as well as 44 regular speaking sessions. We had a total of 139 submissions, which meant we had to decline 2 out of every 3 submissions. We don't say this to brag: it's actually very painful to decline talks, especially when we have worked and hang out with some of the submitters for more than a decade.

The training submissions were competitive but were not nearly as daunting as picking the talks. Jim & I researched each training session and hammered out the lineup in under an hour. The rest of the talk submissions went to the full talk committee.

Like last year, we tried to answer the following questions when looking at the talks:

  • Does this proposal have real-world applications? Case-studies scored major points with everyone, and we tried to measure how broadly the case-studies applied across our attendee spectrum.
  • If the proposal revolved around a tool or utility, would the talk cover real-world deployments of the tool and what problems it helped to solve?
  • Did the talk offer a fresh perspective on a PostgreSQL topic?
  • If the topic was forward-looking, would it be appropriate for a user audience and provide enough information so users could take action now?
  • Did the proposal provide enough information about the talk? Generally proposals that were more detailed met with more favorable reviews. Also, the actually wording of the proposal counts; if you do not write your proposal properly, the odds are you will have trouble conveying your message while speaking.
  • It seemed that people read last year's blog post because the talk submissions seemed to echo the "real-world application" criteria. What made the final selection process very challenging was we had a lot of great examples of people and organizations running PostgreSQL in production that it made it challenging to pick which ones to include.

    It was also interesting to see what themes emerged in submissions over the course of the year. Last year, we had a lot of submissions around PostgreSQL deployment strategies and PostGIS. This year, the central theme seemed to be high-availability and distributed PostgreSQL. Personally, I am not quite sure what to make of this, but there is some sort of technology trend here and I at least want to note it!

    This was also the first year we included a "hackers" or "PostgreSQL internals" track on the agenda. To our pleasant surprise, we received a lot of high-quality submissions that covered topics on how PostgreSQL is developed and tested to its evolution over the years to what is coming up in the newer PostgreSQL releases!

    As you could imagine, as we went through the talks, there were some that were clear winners and clear disappointments, which allowed us to narrow down the list. However, we ended up at a point where we had 40 undecided proposals for 5 slots. This is both a testament to the quality of the talks submitted, but also when putting the schedule together is not as fun for the organizers. To add some levity, I thought at one point it was 40 undecided proposals for 10 slots, but I had an error in my spreadsheet; I should have used PostgreSQL (which ultimately all the data went into our database).

    We took a harder look at this 40 proposals and tried to ensure it satisfied the criteria we mentioned above and made our choices. We applied similar methodology to making our decisions as we did last year:

    • The talk was covered in some way by someone else who also received universal approval
    • We ultimately decided it did not provide enough use-cases based on the proposal
    • We just ran out of space :(

    We are very proud of the talks we selected, but we do wish we could have had a few more. We are looking into adding another day to PGConf US 2017, but it will take significantly more fundraising efforts to ensure we keep our ticket prices low. We did say last year we were going to look into adding another day for this year, and we did: however, to run the event the way we like to, it was cost prohibitive at this point, but we will work on it for last year. Your talks actually help us to recruit more sponsors who want to give back to the community, and hopefully this is the year where we have enough momentum to add another day for 2017!

    Last year, we offered to give feedback on why certain talks were declined. A few people wrote in for the feedback, which we gladly supplied, and some resubmitted this year to PGConf US 2016. The good news: all of them incorporated feedback from the selection committee and most, if not all, had their submissions accepted. Already several people who did not have their talk selected for PGConf US 2016 have reached out on how to improve their chances of making it next year, and we will do our best to provide constructive feedback.

    And now, the final part: making the schedule! As you can see, we decided to group the talks into four different tracks this year:

    • Use-Cases: How did an organization use PostgreSQL and in what way to achieve its goals? These talks should be accessible to the general audience
    • Operations: How do you configure, deploy, and manage PostgreSQL? These talks are geared towards system administrators / DBAs / devops folks, but have lessons for everyone
    • Development: How do you interface with PostgreSQL? What are some good strategies & tools for developing against PostgreSQL? These talks are geared towards developers, but have lessons for everyone
    • Internals: How does PostgreSQL work underneath the layers? How are features built for PostgreSQL? What's coming up with PostgreSQL? These talks are for a wide-ranging audience, from beginners and the general audience to C developers and an advanced audience.
    • We tried to match talks in time slots that would not have overlapping audiences, but I've learned through the years that doing that is more of an art than a science. It is cool to hear people say they want to go to all four talks during a session, but it also stinks. As with last year, we are recording the talks, but per certain restrictions on certain speakers we may not be able to release them all.

      We could not have done this without our selection committee, which comprised of Mehboob Alam, Joshua D. Drake, Jim Mlodgenski, and Jonathan S. Katz. They reviewed all 139 proposals in less than 5 days during a busy work week, so big kudos to them.

      As always, if you have any questions, comments, concerns, thoughts, kudos, etc. you can reach out to us at - we will try to answer all of your emails and please do not be dismayed if it takes some time.

      (Click here for last year's edition: