Sim racing scratches several itches at once. It is a machine to tinker with, a skill that rewards deliberate practice, and a surprisingly good teacher of the same lesson I keep getting from flying and from carpentry: going faster usually means doing less, earlier, more smoothly.
The thing I find genuinely interesting is the fidelity problem. Force feedback, tyre models, field of view, pedal load cells. None of it is the real thing, and all of it is trying to give your body enough honest information to make good decisions. That is a design problem I recognise from every other kind of interface I work on.