On falling in love with old cars
A year-end essay on mechanical things that age well, written from the garage floor.
A short lede that sets up mechanical things that age well — why it mattered, what I set out to do, and what I learned by the end.
The idea
Every project starts as a Sunday-night frustration. This one was no different: a small itch that would not go away until I built something to scratch it.
I gave myself a single constraint and let everything else follow from it. Constraints are the cheapest design tool I know.
How it was built
I started in the stack I was most comfortable with, threw the first version away, and rebuilt a worse-looking one that I actually used.
The stack
Next.js on the front, a thin API in the middle, Postgres at the back. Nothing exotic — boring tools that let me move fast.
What broke
- Migrations, twice.
- My own attention, for about three weeks.
- A deploy at the worst possible hour.
Build the version you will use on a Tuesday morning, not the one you will demo on a Friday.
What I would do differently
- Ship the ugliest working version to myself in week one.
- Treat the calendar as the real budget.
- Default to a walk when I can’t tell if I’m tired or stuck.
Where it stands now
It is a small thing that a handful of people rely on, and I am prouder of the numbers I can feel than the ones on a chart. On to the next constraint.