Relevant information known to be available may go unused in research and development because of information overload or because its use is excluded by deliberate policy. Exclusion by policy shows that R&D is not, and does not aim at always being, efficient in the sense of fully reflecting all available relevant information. It may still be efficient relative to chosen strategies of information use and nonuse. Overload may be a sign of strategic error, or may be accepted as routine and normal.
introduction Progress in research and development depends on adequate communication, and those whose job it is to support work in R&D naturally want to know whether communication is as efficient as it might be, and, if it is not, whether there is something that can be done to make it more efficient. This is certainly true for information science, for which the support of R&D is a central concern. Efficiency might be understood in different ways, but one especially interesting way is the economists' concept of efficiency reflected in the e&ient market hypothesis (Wilson, 1993a).' According to this hypothesis, which is generally well supported by empirical evidence, capital (stock) markets are in fact efficient in the sense that they fully and correctly reflect all relevant information in determining security prices. This sense of efficiency is especially interesting, in that it matches a widely recognized requirement on rationality itself: rationality requires utilization of all available relevant information in arriving at theoretical and practical conclusions. For a philosopher "It is a truism that inductive conclusions must be assessed with respect to our total evidence" (Harman, 1973, p. 165); a social scientist speaks of estimates that are "rational, in the sense of taking account of all available information" (Elster [ 1989(Elster [ p. 1091, other references , other references ' This article is a continuation of Wilson (1993a), and relies on exposition and arguments given there.