It seems like the author is studying “disorder” in an array by restricting themselves to looking only at local differences. They make up a lot of reducer functions, many of which relate to each other. But I don’t see the usefulness in playing with them — as none of the individual functions are ever useful. The author even says in their article about Amp that it doesn’t really work.

Basically all the functions only observe local behavior, but to measure “disorder” in a sensible way you need to take into account global relationships; the disorder of 2-3-4-5-6-5 is different than that of 2-3-4-5-6-1. There doesn’t seem to be a proposed way around this, but maybe I missed something.