Scientific computing is actually a use-case I see this working well in; IMO, it's perfectly fine to have terse code that exactly matches up to equations in a longer paper that justifies correctness, etc. (as long as the paper is easy to find from the code!)