The problem is that it doesn't quite work like that. The computer doesn't know that the file has just been wav-mp3-wav'ed, it assumes any file is fresh and runs its algorithms accordingly.
Think of it like this :-
A writer writes a novel.
The novel is given to someone who produces an abridged edition - the same story, but less detail.
The abridged edition is given back to the original author and he is told to rewrite the novel from it - While the end product may be close to the original, it will have significant differences.
The new novel is abridged again - same overall story, but changes will start to creep in.
Keep this up and eventually the original novel will be beyond recognition.
An average size MP3 file only contains something like 12 percent of the information available on the original, so there's no way you could put back the other 88% exactly as it was to begin with. The errors start to creep in, then if you do it again, the program sees those errors as just another part of the track, so it processes those too, and so on.
