Another way to look at this is something analogous to a Fourier transform. Given a function, try a frequency and see what you get. So maybe some of the things that conceptually make that nicer might also have an analogue here?

This paper really is a breath of fresh air, for mine