The same sort of thing that's pseudo about pseudo random numbers...usually it is indistinguishable, sometimes in the case of Clojure one is reminded that it is probably wise to keep the underlying mechanics of Java's type system in mind if one wishes to maintain an accurate mental model. In particular, it is useful to keep in mind the way in which class and interface types structure their child types within a system of static typing.
This isn't a bad thing. Being built on Java provides tremendous utility. But the way in which Java makes lumps under the Clojure language layer which blankets it gives Clojure a different flavor than say Common Lisp...and note I said "different" not better or worse.
Anyway, that's the idea I was trying to capture with the term. I wasn't disparaging Clojure, just suggesting its relation to the JVM comes at the price against Haskell style greenfield implementation.
That has nothing to do with static or dynamic typing. -1/3 is a clojure.lang.Ratio. java.lang.Math/abs is an overloaded method that knows nothing about it. If you want that to work, wrap it in a float or use abs from clojure.math.numeric-tower.
