Unlike timely dataflow, differential dataflow is meant to be something you can start using without understanding much about how it works. This is great news for us, because under the covers it is way more complicated than you might hope. We'll talk about that too, later, but for now let's just talk about how to create differential dataflow computations.