Compression of random binary data

[danceswithnumbers at gmail.com]

So if the theoretical min compression limit (log2(n)*(x)) has a 3% margin but your transform has a less than 3% inflate rate at most then there is room for the transform to compress below the theoretical min. With every transform the entropy changes, the potential for greater compression also changes, with each pass you can compress untill the entropy is so random it can no longer be comressed.