Did not initially make much intuitive sense. We believe that learning to reason about data requires students to grapple with the same sorts of interpretation issues; in the process, they need to develop conceptual (not necessarily mathematical) models of data that can guide their explorations. The idea of data as signal and noise, physically embodied in the workings of the Galton Board (see Biehler, 1994), is perhaps the most fundamental conceptual model for reasoning statistically. Future research should help us learn how the idea develops and how we can foster that development in our students.