I argue that there is no viable development of the instrumentalist inference within a model research program. I further argue that both Friston and Pearl blankets are not the right sort of tool to settle debates on philosophical internalism and externalism. For these reasons, the inference within a model program is far less promising than the target article suggests.