Parametricity and Reuse

[Brad Collins]

He doesn't talk about parametricity (or the Police's Synchronicity), but he does talk for a few minutes about reuse. And it sounds a lot like what we were talking about yesterday: The less you know about a function's input types, the fewer its the possible implementations.

Eric's point is the more focused a function is (i.e., the less knowledge you have to load the function up with), the more you can reuse it:

https://lispcast.com/how-does-fp-achieve-reuse/

Which flies in the face of what might be the first impulse: introducing a mechanism (e.g., switch/case, reflection) to handle all the varied specific cases. Turns out that when you do that, you make it less reusable.

I think it also speaks somewhat to Bubba's point yesterday about how difficult it can be to be forgiving about the format of the data you consume. By handling this case and that case and this other case ... oh, and that other case over there, you actually end up over-fitting to the domain, and the result is a function that is less reusable, not more.

–Brad