Sat Apr 06, 2013 8:54 pm
It's an information theoretic thing. You can demonstrate mathematically that it's impossible to come up with a compression scheme that makes *everything* smaller or equal-sized; some things get smaller, some get bigger. The trick is to make sure that the algorithm you use means that the things that get bigger only get a bit bigger, and that things that get smaller get much smaller.
Things that are already compressed look like random data; effectively, they have no patterns for a compression algorithm to exploit. So they're in the class of things that get bigger. What you're trying to compress is, sadly, in this class.
Director of Communications, Raspberry Pi