I spent more than fifteen years working as a carpenter. Framing, finish work, the whole range. It was not a stop on the way to something else. It was the job, for a long time, and I was good at it.
Nothing is ever square
The first thing the trade teaches you is that the drawing is a suggestion. The foundation is out by a quarter inch. The wall has a bow in it. The floor slopes toward the back door. You do not get to build the house in the drawing. You build the house that is actually there, and you make it look like it was always meant to be that way.
That turns out to be most of what I do now. The brief is a suggestion. The data is messier than the schema says. The user does something nobody anticipated. The work is the same work: look closely at what is actually in front of you, then make it good anyway.
Measure twice
Carpentry has a short feedback loop and expensive mistakes. Cut the board wrong and you can see it, immediately, and the board is gone. That builds a particular kind of care that is hard to learn from software, where the mistake is invisible for weeks and the material is free.
It also builds a tolerance for doing things in the right order even when the wrong order would be faster today. You do not hang drywall before the rough-in passes inspection. You do not ship the feature before you understand what it is for.
Finish work
The part of the trade I liked most was finish carpentry, which is almost entirely about the last two percent. Nobody notices a good reveal. Everybody notices a bad one. The same is true of an interface, a document, a product, a piece of music. The last two percent is where the thing either feels considered or it does not.
I still miss it on the days the code will not behave. There is something clarifying about a problem you can solve with a sharp blade and a steady hand.