An epidemic model is a statement of what follows from a set of assumptions, not a forecast of what will happen. The distinction was lost in public discussion during recent outbreaks, where projections generated under explicitly counterfactual conditions — no behavioural change, no intervention — were reported as predictions and then judged to have failed when the intervention they were designed to motivate was adopted. Modellers contributed to the confusion by presenting scenarios with the visual apparatus of forecasts. The underlying problem is structural: a model that changes behaviour invalidates its own projection, and there is no way of presenting such a result that is both honest about the mechanism and simple enough for a headline.