[Haskell-cafe] Free theorem - what do we need it for

Some time ago I get knew about free theorem that states (omitting undefined & co.) that there is 1 haskell function forall a. a -> a. The property seems to be workaround by Typeable but it stills not work for all types so it cannot be used in, for example, Monad or Functor. I am afraid that I fail to see practical benefits (I mean that I don't know all implication not that I think there is no benefits). What are the benefits of free theorem except large chance that data will obey he functor/monad laws? I.e. what advantage it gives over the 'reflection' from other languages? Regards