Abstract
The author distinguishes between informants and sources of information, and argues that the concept of knowledge is tied to the former and not the latter. The distinction is then used to cast light on the necessity of the belief condition for knowledge and on comparativism, the view that a person might be said to know p in circumstances in which the alternative is q, but not to know p if the alternatives include r. Goldman's famous papier‐mâché barn thought experiment is also discussed. Craig concludes that where belief is lacking or knowledge appears comparativistic, the subject does not have genuine knowledge, for she is a mere potential source of information, not an informant.