Lots of research is mostly programming, e.g. my applied maths PhD. The way to try out software-based ideas is with programming.

I think the point was more than a lot of programming is not actually programming. Working in the industry, most somewhat complex systems require mostly work on paper, planning, research, reading documentation etc. and in the end some writing of code. Too often though, that is dismissed because it's "less agile" and a few years down the road the technical debt is huge.

There's no point building a hypothetical system. You have no idea if it works until you try it. And lol, documentation? For a system that doesn't exist?

This type of research happens in teams