So, why not just keep it as a potential energy term, and proceed the way physicists do, which is to use observations of outcomes to conjecture modifications and develop the theory? Doing so is also more amenable to accounting for sequencing, probabilistic outcomes, etc.