TODO

- parallel calculation that is accurate to serial calculation, for speed
- formal description and paper start
- more extensive testing on hamming distance for 1 bit change ,and message length, for various digests, and same for 1 byte change, and also for half bits change.
- brief analysis of what system is
- try to prove for simple 1 block case:
  - digest uniqueness (i guess order invariant, is there a way we can fix that? add a slack order variable)
  - prove upper bound on hamming distance change for 1 bit change for 32 case
  - show that we expect 32 digest to be similar
  - even if can't prove state results of experiment and do brief analysis
- formalize paper