> Proving that you can treat "processing one event" as globally atomic was the funnest thing in grad school.

Uh? I guess it depends on what 'atomic' means but in the real world that isn't the case.