I'm not a mathematician, but I'd say that it's one of the simplest and most elegant instances of its kind, and that in general we don't seem to have good ways of dealing with iterative processes that can both lengthen and shorten (or increase and decrease) the size of the output in relation to the input. A generalization of the proof of this might be quite illuminating for a lot of hard problems in number or complexity theory.