Actually, I did a small test and found that after gzip, the file size only expands by about 10x. Running both input and output through bz2, the obfuscated file only comes out 3x larger. If you were very protective of your code, and you had enough of it to justify loading up a bz2 decoder on the client side, you could actually make that economical bandwidth-wise.

That said, this was a very small test; the original file was a random snippet of JS code less than 500 bytes, and that itself took a considerable amount for hieroglyphy to chew on, so I can't really do a proper test of a larger input file.