Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Suppose the optimizing consumer only has local information, then what? Obviously, there will be a random element of consumption. The link between the frequency distribution of consumption by person and preference surface is worked out. Under convex preferences there will be a unimodal distribution and conversely. A local optimizing theory of addictive behavior is developed from nonconvex preferences. The resulting bimodal distribution of consumption implies a nonrobust outcome for median voter politics. Such a local optimizing model of addiction is compared with globally optimizing models of addiction. Copyright 1994 by Kluwer Academic Publishers