I’m Jeff Derrick — a platform and data architect at MIT Sloan, where for twenty years I’ve built and run the systems the school depends on: the data platform underneath, the API layer that guards it, and the front-end framework on top. Small team, long horizons, no framework churn. Software that has to still be working in ten years.
This is where I write about the part of that work nobody can see.
Most of what makes software good is invisible by construction — the outage that didn’t happen, the feature someone talked the room out of, the abstraction that never needed to exist. The best engineering I know leaves an absence rather than an artifact, and absences don’t show up on a dashboard.
AI made that sharper. Generation is cheap now, and nearly everything built around it points toward more — more code, more features, more visible output. The subtractive work was always the most valuable, and it just got harder to do and no easier to see.
I don’t have a tidy answer to that, and I’m suspicious of anyone who says they do. Most pieces here end on a question rather than a verdict.
Every other week or so. No paywall, nothing to sell.
Older pieces — including the 2019 one the first post here is built on — are on Medium.
