Yoav Ravid comments on K-complexity is silly; use cross-entropy instead

[Yoav Ravid]

Seems good to edit the correction in the post, so readers know that in some cases it’s not constant.