Hayami and Willis Peterson present a practical application of Alfred Marshall's social welfare concepts.1 Their approach, in the tradition of public goods analysis, seeks to measure the social returns to improvements in information about U.S. farm commodities. Their strong theoretical argument is jeopardized by their empirical evidence leading to exaggerated and unrealistic conclusions. We have two basic criticisms of their analysis: 1) they use highly inelastic demand elasticities for grains which exclude export and animal feed uses; and 2) their conclusions are dominated by the marginal social returns for onlv two of the seventeen commodities studied. Their distribution of returns raises specific questions about sampling techniques for the two commodities which may be more important than the broader issue of social returns to public information services.