No he didn't, because "productive" != "(not?) difficult", they're mostly orthogonal. LISP is productive according to PG because it has a kind of unlimited amount of abstraction, I don't think he would say that LISP is difficult. LISP/smalltalk: PG-productive, not difficult Java: PG-unproductive, not difficult C++: PG-unproductive, difficult (Forth? Haskell?): PG-prodictive, difficult

So there are points in all 4 quadrants I think.