I think we should document the complexities of the protocols, at least
where it isn't obvious (the comparison protocols).
The complexities include the amount of local computation, the number
of bits transferred, and the number of rounds.
Even though VIFF ignores the traditional notion of "rounds", the
number of rounds needed tell us the number of network exchanges needed
due to inherent dependencies in the formulas. VIFF cannot do anything
about that, so I think this information is still valuable when
comparing protocols.
|