I'm not talking about "reading a proof to see what I can copy", it's about being able to rely on what is written there. Also, there are many papers that serve as references that I want to be able to look up quickly. Rederiving every result any time I want to use it is not the way to go.
