I think that's why Lisp is pleasing in a mathematical sort of way. You start with your basic axioms, which are simple to understand. The real test, however, comes from your ability to build up from those axioms an elegant, abstract framework which gets closer and closer to describing your problem set. This is why smug Lisp weenies say that in other languages, you wrap your problem around the language whereas with Lisp, you wrap your language around the problem.

There's definitely a mindset behind building Lisp programs that makes it worth using, but sadly those benefits require lots of work to arrive at and are often buried in pages of elitist, manic-depressive drivel on certain newsgroups when trying to describe why someone should use Lisp.