What are the assumptions in our code vs in the Alto standard? How do we test whether or not these assumptions are valid with real-world data?