2012.12.15 03:13:07 (279771037078528000) from Daniel J. Bernstein, replying to "Martin Boßlet (@_emboss_)" (279768438954659840):
@_emboss_ Communicating n inputs 0, 00, 000, ... costs O(n^2). Storing the same inputs in a crit-bit tree costs O(n^2). What's the question?
2012.12.15 02:55:56 (279766710247112704) from Daniel J. Bernstein, replying to "Martin Boßlet (@_emboss_)":
@_emboss_ Another nitpick: crit-bit trees guarantee performance linear in the input size; worst case is a few dozen cycles per byte.
2012.12.15 03:02:48 (279768438954659840) from "Martin Boßlet (@_emboss_)":
@hashbreaker But if input size is unbounded, ever longer inputs perform poorly or don't they? Maybe we can discuss this in Hamburg :)