Mainframe and PC Computing in American Cities: Myths and Realities
Norris, Donald F., Kraemer, Kenneth L., Public Administration Review
How much can PCs aid city management? This article is based on a 1993 survey that compares computing in cities that use only personal computers (PCs) with computing in cities that use central computer systems. The authors found that claims that PCs would speed up automation of governmental functions were not substantiated. Central system cities were more widely automated, had more widespread use among staff and were more likely to deploy leading-edge computer technologies than PC-only cities. Moreover, respondents in central cities were positive about computer impacts and satisfied with computing. PC-only cities had an edge over central-system cities in that they reported fewer problems with computers, but the test of statistical significance showed only a weak relationship. The authors argue that PC-only cities' reliance on ad hoc solutions, out-sourcing, or "computergurus, " results in a failure to develop ongoing support capabilities. In contrast, central-system cities have developed and enhanced these capabilities over time, thereby providing greater support for the computing function and a more stab1e technology platform.
Both elected officials and professional managers in local governments believe in the value of computers, especially personal computers (PCs), to their own work and the work of government. Various academic studies have demonstrated this belief over the years (e.g., Dutton and Kraemer, 1979; Perry and Kraemer, 1980; and Norris, 1989 and 1992). Yet, policy makers are continually confronted with claims about computing that they find difficult to assess and that occasionally defy rationality. For example, within recent memory it has been claimed that privatization or outsourcing would take the computing problem off the hands of local officials at less cost and that geographic information systems would enable officials to make Solomon-like judgments about such important matters as land-use planning (e.g., Richter, 1991; Loh and Venkatraman, 1992; Public Technology Inc., 1991). More recent claims are that client-server computing is the new low-cost way to governmental automation (Gagliardi, 1994) and that desktop computers are the means to increase employee productivity and to empower workers to deliver better services to citizens (Greisemer, 1983 and 1984).
One of the most persistent claims, which has at least a decade of history, is that the PC can effectively replace larger central computer systems in local governments (i.e., mainframes and minicomputers). For example, it is frequently asserted that, unlike mainframe computing, the introduction of PCs is an easy, low-cost solution to automation in government. It is believed that by adopting PCs, latecomers to computing can leap-frog the brain-dead mainframe and minicomputer technologies and still gain all the benefits of these earlier, cumbersome technologies--and then some. All that is needed is basic investment in the technology and the empowerment of workers to use the technology in their jobs. The need for Management Information Systems (MIS) departments will be only to help make the transition and to train users in the new technology (see, for example, Greisemer, 1983 and 1984; and Voss and Eikemeier, 1984).
PCs may be all that small local governments or even some small units within larger governments need to conduct their business. However, it is extremely unlikely that even the most powerful and sophisticated PCs on the market today can solve all of the automation needs of local government. Indeed, recent studies of the lifecycle cost of PCs, actual experience with PCs, and recent reports on PC-based client-server computing call several of these assertions into question. For example, while the initial cost of PC-based client-server computing has been shown to be lower than mainframe or minicomputer alternatives by 20 to 30 percent, the five-year costs of PCs were found to be two to three times as great per employee (Nolan, Norton, and Company, 1992; Miller, 1993; Ambrosio, 1993; and "Client/Server," 1994). …