There's a whole set of problems where this is not what you have. You have data that's mostly non-contentious but you still have to share it and work on it. And yet you want to have a correct result. This is what STM is trying to address. A set of problems that can be solved by just splitting stuff into processes.