Pi & friends are all well and good when all you're doing is filling out a few initial bitfield states in a pipeline, but when you're designing a highly complicated process by which to deterministically produce a safe but "random" curve, the process of designing that process itself is going to have some unjustified choices in it. The meta-algorithm, if you like, could have been designed to produce a specific result given the input pi.
Publishing & rigidly defining the method before choosing the seed would seem to alleviate all possibility of foul play.
Pi & friends are all well and good when all you're doing is filling out a few initial bitfield states in a pipeline, but when you're designing a highly complicated process by which to deterministically produce a safe but "random" curve, the process of designing that process itself is going to have some unjustified choices in it. The meta-algorithm, if you like, could have been designed to produce a specific result given the input pi.
Publishing & rigidly defining the method before choosing the seed would seem to alleviate all possibility of foul play.