|
Data processing was crucial to the success of the Open University where use of the computer has already evolved through three stages.
Introduction
By the time that the United Kingdom Open University (UKOU) was established in 1969, data processing was widely used in the administration of conventional universities. Such use, however, tended to be restricted to applications such as financial management, inventory control, staff and payroll systems, and to shared use of computers whose primary duties were academic. Experience within distance teaching was likewise insubstantial (Sorensen, 1969: 107).
The decision of UKOU's planners, then, to set the computer in the centre of administrative activities (Perry, 1972: 95) while not being revolutionary, was certainly bold, particularly given the academic's traditional mistrust of bureaucracy and the belief that computers, except in their role as research tools, represent bureaucracy. That decision had a profound influence on the nature of the institution that has grown up over the last decade. Was that influence benign or otherwise?
UKOU chose a highly centralized model for its administration. Although the regional offices have always enjoyed substantial autonomy, it has been at the centre where policies have been determined, strategies adopted and where the computer has remained custodian of the central resource of student data. Arguably, it has been this last factor that has contributed most in practice to centralization.
Data Processing Applications
To set the context for this discussion I will fall back on a diagram I used in an analysis of student administration in Kaye and Rumble (1981: 124). This diagram analyses student administration into three systems of admissions, student records and assessment and certification, and shows how the student record created during the admission process becomes the central resource for subsequent processes.
Such a diagram would have seemed very futuristic when in 1970 as my first task for my new employer I was asked to design an admission system. The system began by accumulating a file of applicants from which the first cohort of students would be selected once admission policy had been agreed. Here is a good example of how the computer made certain policies possible. With an admission period some six months in length it would have been possible, and certainly simpler, to have admitted students to the undergraduate program on a continuing basis. Instead, UKOU chose to allocate places all in one go. Each year the file of undergraduate applicants grows to some 40,000. When selection operates half of these receive direct offers of a place, some 5,000 are put on reserve, and 15,000 refused admission. Place offers are made on the permutations of course, area of residence, occupation and sex, a matrix of some 1800 cells. Considering only this one process, of setting selection parameters after repeated analysis of t
he file, of distributing 40,000 records in 1800 cells, the whole process iterated several times to optimize the outcome, and all within two or three days, then it is difficult to imagine that this could be accomplished without a high level of mechanization. This sort of selection gives much greater control of the overall entrant population than continuous admission.
Most of UKOU's characteristic components were in operation right from its first year, and the student record file was soon called upon to provide a central data resource for a wide variety of processes. It was the basis for the mailing of course materials and of home experiment kits to the student. Indeed, selecting a great range of mailing labels across a variety of courses and students became one of the chief occupations of the computer, which is at the centre of a continual series of communications from the institution, asking for fee payments, informing of allocation of summer school or exam, requesting return of the home experiment kit, prompting course choices for the next year, and so on. Not all mailings were premeditated. We were obliged to send out corrections, afterthoughts or apologies more frequently than we would care to admit. Stories abound of instances where simple errors created simultaneous bewilderment across the nation.
The Evolution of Data Processing at UKOU
Stage 1: 1970-1975
The evolution of data processing at UKOU appears to be falling into three stages. During stage 1, the years 1970 to 1975, we developed systems as we needed them—the normal response of any new organization. It was clear that to follow the admissions system, a system of student records was the next step. Then, as the institution developed its strategies, we found the need for systems to maintain records of the growing number of part-time staff becoming tutors and counsellors, and systems to process students' assignments. As is now generally known, the University adopted a dual assignment policy, using traditional tutor-marked assignments allied to objective testing marked by computer. This computer-marked assignment (CMA) system brought together a greater variety of objective testing techniques than, in all probability, had previously been combined into a single assessment system. It is almost inconceivable that such complexity would have been attempted had it not been for the computer and its document
reader, able to process two thousand assignments an hour.
Year-end procedures provide perhaps the most noteworthy example of using data processing to achieve results impossible by less sophisticated means. All undergraduate students start and finish their courses at the same time. The consequence of this audacious simplicity of academic planning is an administrative problem of outstanding magnitude. A nationwide campaign of examinations culminates in the recording by the computer of a hundred thousand examination scores, their analysis and standardization, and their combination in various ways with the results of over half a million assignments. This, again, must be accomplished in days rather than weeks, so that reallocation to the next year's courses can go ahead on schedule.
Stage 2: 1975 to the present
During 1975 UKOU changed its batch processing computer using magnetic tape for a more advanced model using direct assess disks, and with a database and terminal network. This machine could be seen as the apotheosis of the central data resource philosophy. With the earlier machine, the physical expression of this philosophy was a very large serial file on six reels of tape; an efficient file if you were transferring to it fifty thousand course results, but very clumsy if you wanted to immediately update it with student Bloggs' latest address. The order of the new computer, incidentally, resulted from a strategy review as early as 1972 (Perry, 1976: 7). Initially we simply converted our stock of systems to the new computer—that is if you can describe as simple a task that consumed 30 man-years of work! Then, in 1975, began the work of redesigning systems to take advantage of the new facilities. If I now say that the first of this second generation of student systems came into operation last year (1980)
you will get some idea of the gestation period of large-scale computer systems. However, that gives a rather false impression unless qualified by adding that this period was also occupied by simultaneously setting up the new database and telecommunications environment.
Lessons from experience
Back in 1969, as well as planners who declared the whole venture impossible without data processing, there were those who argued that to give prominence to the computer would unacceptably constrain academic policy. Now, more than a decade on, we may ask who was right. As always, of course, they were both right. First, though, one should say that data processing is expensive. The administrative computer has regularly cost UKOU between five and six per cent of its recurrent expenditure, a proportion that may well cover the entire administrative overhead of many a less complex teaching institution. You may say that to do what is being done using an army of clerks instead would be even more expensive, but then one would not attempt to land on the moon without a moon rocket. The very power of the computer to create systems of great complexity is, I feel, the most difficult of gifts to use as well.
There were, and indeed still are, a class of problems arising from the interaction of academic traditions with the demands of operational data processing. Academics are not good at meeting deadlines. Apart from the headaches this gives to editors and printers of course materials, the failure to deliver assignment marking parameters on time means delay to operational schedules. More seriously, parameters are sometimes wrong. The computer will contentedly wrongly score thousands of assignments and send off thousands of letters before anyone notices the error. With highly integrated automated systems, small errors can be greatly magnified.
Then there is the problem of deciding what data to maintain. Coming from a commercial computing background, I tended to resist the attempts of academic colleagues to burden, as I saw it, the admissions system with a range of social and educational data without operational relevance. Clearly it is difficult to decide which burdens will eventually yield meaningful results, and which will remain simply burdens, but in retrospect I was perhaps wrong to scorn intuitive arguments. With manual systems this is a non-problem—manual systems rarely indulge in keeping data other than that essential to their operation.
As well as making many policies strategically possible, the administrative computer is accused of constraining innovation and restricting academic freedom. It takes several years from conception to implementation of a major computer system. Given the need to set and honor deadlines, there are absolute constraints over second thoughts and late inspirations. Several times in those early years, we scrapped a system representing some man-years of work after a very short operational life. At times this was because of bad design, but more usually it was because system development had run too far ahead of policy development. The computer did at times win a battle, but I think that policy always won the wars.
UKOU is still a highly innovatory institution, although our general direction of travel is naturally clearer now than it was ten years ago. However, the problem of reconciling academic innovation with the detailed long-term planning demanded by the computer remains far from solution. If anything, the growing sophistication of the data processing system has increased gestation periods, so the problem of asking policy makers and strategy implementers what sort of systems they would like in four or five years time remains. They may not be absolutely certain what sort of system they ought to be operating right now, and, besides, the inheritors of their prognostications in years to come may see the whole thing quite differently.
Stage 3: where to from here?
Earlier I mentioned a three-stage evolution. Our second generation of administrative data processing systems will be complete by 1984. By that time we shall be some way into stage three, which I see as the incorporation of management information systems into our stock. Not that our systems do not at present produce great quantities of information, much of it directed towards the higher management function. However, a true management information system can result only from a prior identification of management requirements and a recognition of the need to integrate data derived from both operations and research (Erdos, 1975: 11). This takes a lot of experience.
There are those who see in the computer a means to advance directly to the level of management information systems without the intermediate frustrations of learning by experience. They are sadly mistaken. The computer has no inherent ability to solve our major problems for us. It simply enables us to use our experience to greater effect.
References
Erdos, R (1975) The system of distance education in terms of subsystems and characteristic functions. Papers to the 10th ICCE International Conference. Malmo: ICCE.
Kaye, A. and Rumble, G. (1981) Distance Teaching for Higher and Adult Education. London: Croom Helm.
Perry, W. (1972) The Early Development of the Open University. (Report of the Vice-Chancellor January 1969–December 1970). Milton Keynes: Open University Press.
Perry, W. (1976) Report of the Vice-Chancellor 1975. Milton Keynes: Open University Press.
Sorensen, L. (1969) Electronic data processing in the administration of correspondence education. Proceedings of the Eighth International Conference of the ICCE. Paris: ICCE.
|
