The problem (in my mind) with using P is that only a small % of problems are formalizable as communicating state machines. In your greenfield big data system,

The problem (in my mind) with using P is that only a small % of problems are formalizable as communicating state machines. In your greenfield big data system, what were the state machines? What states could they be in?