A technology exists largely in the minds of its labor force and in the future will be distributed between those minds and the memories of computers. If programs, stored in one or the other of these forms, constitute the core of a technology, then important consequences are likely to follow from the fact that automation greatly decreases the cost of making copies of such programs. One of the obvious consequences of cheaper copying is that there will be under investment in program improvement unless steps are taken to reward inventors of programs or to subsidize invention. A second consequence is that the comparative advantage of automation will tend to be particularly great in situations where frequent arid rapid program change is called for, and will tend to be relatively less in areas where only a few copies of a program can be used. Since human programs are at least modestly capable of on-the-job learning and adaptation to specific situations, the range of feasible automation will depend heavily on the extent to which similar learning and adaptive features can be incorporated in automated programs. The concept of technology as consisting of stored programs gives a somewhat novel framework for theory about the rate of technological progress and the rate of diffusion of new technology.