@EricMH, the paper by marks does not include any reference to “observed ASC” versus “ASC”. Can you elaborate the different you are seeing? Where does Marks make that difference. As I understand it, the claim is that ASC is greater than or equal to the function I implemented. With a better compression algorithm, the ASC estimate might increase.
Perhaps I should clarify that my ASC function implementation is a bound on the true value. It is a lower limit on the true ASC, which will be greater than or equal to the number computed here. Is that the distinction you are getting at here?
Would you like to change the algorithm so it exactly fits the definition?