Yesterday on Slashdot, [1] they had a blurb [2] about Zeosync's [3] claim of a compression scheme that supersedes Claude Shannon's work.
I only mention it here because an associate of mine (who will remain nameless unless otherwise told) not only knows the company, but the company my associate works for has been contracted to actually implement the algorithm, which I've been told is so convoluted that even the scientists at Zeosync are struggling to comprehend. I find that rather amusing (not so my associate, who, being the head programmer, has to most likely write the code).
I've also been told it's not a constant 100:1 compression ratio, but that's the best case you can expect to see.