I began this article with the thesis that the director of a university computer center is in a double bind. He is under increasing pressure because of competition with networks and minicomputers at the same time that his funding base is weakening. The breadth of demand for computer services, and the cost of developing new services, are increasing dramatically. The director is pressed by budget officers and internal economics to run more efficiently, but if in so doing he fails to meet new needs or downgrades effectiveness for some existing users he runs the risks of losing demand to the competition and hence worsening his immediate financial problems. The impact of networks on this state of affairs might be, briefly, as follows: 1) The centrally planned computer utility would take these pressures off the individual campus computer center and lodge them in a state, regional, or perhaps even a national network organization. While this might be desirable in some cases (depending on the scale of operations), I believe that economies of scale would tend to be more than offset by diseconomies in planning, management, and control; by a reduction of responsiveness to users' needs; and by a slowing of the rate of innovation in computing. 2) The distributive network substitutes a "market economy" for a centrally planned one. Subject to a certain amount of planning and regulation, which might be undertaken by colleges and universities themselves, individual researchers can tap larger markets for services, and participating institutions can obtain at least part of their computing needs on a variable cost basis at prices determined by competition. 3) Membership in a distributive network with sufficient breadth and depth of resources can emancipate the director of the computer center by widening options and allowing him to serve more effectively the steadily broadening range of legitimate academic and research computing needs without his having to stretch his internal resources too thinly. In other words, he can solve the problem of simultaneously improving the breadth of service and increasing operating efficiency. 4) Involvement in distributive networking will raise a new kind of question for the senior officers of colleges and universities. This is the decision concerning the development of computer services for export to users at other institutions. The effect on the university's own academic program (in the sense of its becoming a "center of excellence" in a particular computerrelated discipline), the risks involved in trying to attract outside users on the network, and the consequent responsibility for providing continuity of service at the peril of suffering in national academic reputation will be key considerations. The worth of, and probably the demand for, such services will be a function of the excellence of the development work, and this in turn will depend on its involvement with the university's academic resources. The "computer services export" question is fundamentally academic, as are decisions on the expansion or contraction of teaching and research programs, and it must be dealt with in the same terms. The next few years will be crucial ones for colleges and universities generally, and for their computing resources in particular. The advent of computer networking raises a host of academic, economic, technological, and organizational problems. In spite of these problems, I believe that distributive networking will have a significant and positive effect on campus computing services.