This reminds me of the "pi encoding" - if pi is, in fact normal, you can encode any file as a tuple (`start`, `length`) encoding a substring of the expansion of

This reminds me of the "pi encoding" - if pi is, in fact normal, you can encode any file as a tuple (`start`, `length`) encoding a substring of the expansion of pi. It's often phrased as a compression paradox of a sort - surprisingly many people overlook the fact that `start` will tend to take a lot of bits to encode.