There is a long tradition of invoking ambiguity avoidance as a functional factor in explaining the rise of differential argument marking. But more recently, some authors have contrasted anti-ambiguity as a motivating factor with “predictability-based marking” or “expectation sensitivity”. Here I revisit the debate, starting out from the observation that ambiguity or indeterminacy is rampant in language use and language structures. It could not be otherwise because there is no way to specify every aspect of meaning that might conceivably be interesting. On the empirical side, I extend the discussion of functional motivations and pathways in differential argument marking to other kinds of differential coding, especially “differential discourse function” marking. I argue that in all these systematic differential-coding situations, expectation sensitivity provides a good explanation of the typological patterns and their diachronic motivations, while ambiguity avoidance is often irrelevant.