There are really 3 ways to prevent spam reliably in a "global" system, ie a commons:
  1. algorithmically, e.g. bayesian spam filtering, ... a constant arms race
  2. [transitive] reputation, e.g. WoT
  3. making it expensive
The more of (1) and (2) you can use, the less of (3) you need. But you probably want all of them at scale.
There has to be a fourth option
reply