It can be applied recursively, but with diminishing returns and an increasingly alien (to a software developer) set of rules as you go down. If learning one thing made you 10% better at the thing above it, you would also become 1% better at the thing above that.