Abstract
Discussion of the cognitive division of labor has usually made very little contact with relevant materials from other disciplines, including theoretical biology, management science, and design theory. This article draws on these materials to consider some unavoidable conundrums faced by any attempt to present a particular way of dividing tasks among a labor team as the uniquely rational way of doing this, given the interdependence of the underlying evaluative standards by which the products of a system of division of labor will be judged. Divisions of labor will typically cut across these interdependencies in ways which leave the outcomes of a process of labor hostage to path dependencies and suboptimalities. Some attempts to avoid these results are shown to be unsuccessful. All these difficulties are compounded by the fact that, in many cases, the division of labor has to be constructed over a ground of values that is itself being constructed simultaneously with the products which they are invoked to assess. Key Words: risk spreading interdependency NK fitness landscapes complexity epistasis Kuhn modularity path dependency.