Rafael Harth comments on Insights from the randomness/​ignorance model are genuine

[Rafael Harth]

But by number of bits, which is what you need to avoid repetition.