composition.al

Celebrating five years of this blog, and a change of pace

I started this blog five years ago, in January 2013. At the time, I lived in Bloomington, Indiana and was in the middle of my fifth year of grad school. I had started working on a new project with a new advisor in my fourth year — not exactly a highly recommended approach to finishing a Ph.D. — and I wanted a place to write about what I was working on. I hadn’t managed to get any papers about my project accepted yet, but they couldn’t stop me from blogging, dammit.

At first, I wrote lots of posts about my dissertation work. I eventually managed to publish, graduate, and get a job, and since then, I’ve continued to use this blog to write about whatever I’m thinking or learning about: distributed computing, domain-specific languages, verification, machine learning, and lots of other things. I’ve also advertised conferences and workshops that I’ve been involved with; told debugging stories; and dispensed advice. Inevitably, the personal and the professional intersect.

This blog doesn’t have anywhere near the huge following of folks like Dan Luu or Julia Evans, but I’ve been fortunate to have picked up a few readers who enjoy it and who I hear from often. I think that part of the blog’s success comes from the fact that I make myself post regularly. I committed to writing two posts a month when I started the blog, and although that means that some of the posts end up being “filler”, I think the benefits outweigh the drawbacks. Having the end of the month looming is often the nudge that I need to finally finish and post things that otherwise would have lingered unpublished for a long time.

At least, that’s how it works sometimes. Too often, though, I let a month go by without posting and then have to backdate posts to the previous month in order to keep up. This has been a problem for some time, but after my daughter was born last July, it got much worse. My posts “What do people mean when they say ‘transpiler’?” and “My first fifteen compilers” were both dated July, but they were actually published in August and September, and I’m sure I wouldn’t have even managed that if I hadn’t already had drafts of both of them started before Sylvia was born.

My two August-dated posts went up in September. When I went back to work full-time in mid-October, the schedule slipped even further, and my September-dated posts didn’t go up until November. By the time December came around, I had six posts to finish for the year; with effort, I managed to crank them all out by January 10. And now it’s the end of the month again, and, predictably, I’m behind.

I’m proud to have made it five years at the twice-a-month pace, but it’s become clear that it’s not a sustainable pace for me anymore. So in 2018, I’m going to try changing my pace of posting to once a month, and seeing how that goes. Once a month seems like something that I can manage. I’m already feeling relieved about this decision. Instead of dreading having to backdate posts for January, I’m excited to write about my colleagues’ and my SysML submission for February. That’s the way that blogging should feel — like fun, not a burden!

Now that I’ve made this decision, I’m realizing that my old posting schedule was causing me to make bad decisions and engage in false economies: at one point, I even remember thinking, “I should agree to serve on the so-and-so program committee, because that means I can write a post about its call for papers, and that’ll take care of one of the posts I need to write this month!” (I said no to that PC invitation, but the fact that I was even tempted to say yes just to get a blog post out of it meant that there was something wrong with how much I was expecting myself to blog.)

Why not do away with this quota system altogether and just blog if and when I feel like it? Looking back on the posts that I’m happy with from over the last five years, I’m certain that a lot of them would never have been published if it hadn’t been for deadline pressure. An example is “Using the simplex algorithm for SMT solving”, which was one of those posts I cranked out in December when I was under the gun to finish a bunch of posts by the end of the year. It was a draft I’d had half-finished for a long time; I was writing about something that I had found interesting, but I wasn’t sure if I was really explaining things properly or if it would be interesting to anyone else. But the deadline meant that I couldn’t afford not to finish up and post it, and when I did, the reaction I got was quite positive! So, I really do think a bit of deadline pressure is good for me. The trick is to have a sustainable amount of deadline pressure, and not so much that I can’t keep up.

Comments