We Weren't Ready for the Flood

In the end I decided not to go to ICSE 2025 in Ottawa. I would have enjoyed catching up with friends and having coffee with people I’ve only ever chatted to online, but I couldn’t justify the time and expense. I also didn’t know what I could say that people wouldn’t have grown tired of hearing: as I wrote in 2020, I’ve had no impact on software engineering research or education despite years of trying, and I think people have grown weary of my jeremiads.

If I had gone, though, here’s what I would have said:

We have been studying software engineers and software engineering for over fifty years. Despite that, we still don’t have an agreed-upon way to measure developers’ productivity, which means we can’t answer the question, “Is AI helping or not?” And we still don’t teach empirical software engineering to undergraduates, which means that most programmers don’t know enough to know that most of the answers they’re seeing and hearing aren’t just wrong, they’re nonsensical.

The flood came and we weren’t ready; I wish I believed that we would be for the next one.

Update:

Why yes, I do have a suggestion about how to get ready: turn Rethinking Productivity in Software Engineering into an undergraduate course that teaches students the methods used to answer questions like this, how to use those methods themselves, and how to tell good science from splackulent crap like the “ghost engineers” screed that got so much attention in late 2024. I realize data science no longer makes university administrators salivate the way it did a few years ago, but if you call the course “Data Science for Software Engineers” you’re bound to get a few takers.