Insightful. Once you learn to see dependencies as implicitly revealing abstract data types you haven't yet teased out into existence, you start seeing all kinds of places where behavior that ought to be explicit and handled by an abstraction is smeared across multiple other concepts.
I've got a half-cooked blog post about this in the oven: "reify the state machine". I've seen plenty of code with lots of communicating parts, which is really a distributed implementation of a state machine. Rewriting to make the states explicit as data, and perhaps creating a central "state machine" type for that machine, can really help clarify the domain model and let you see what you're missing.
Insightful. Once you learn to see dependencies as implicitly revealing abstract data types you haven't yet teased out into existence, you start seeing all kinds of places where behavior that ought to be explicit and handled by an abstraction is smeared across multiple other concepts.