The main problem with that approach is it makes rigging the 'elections' trivial when probabilistically every result is equally as plausible as another.