[Printout | Contents | Search ]

Engines of Change: How computers have affected - and been affected by - the way Penn does business

By Randall Couch

Everybody has a bad day at work now and then. Imagine working in Alumni Records on the day this letter came in:

"...The Senior member of the family is the former Chairman of the Board of Trustees for the University.... "For years Penn has been sending my mail to my father in New Jersey. Every time you do so, I receive the material along with a sarcastic note from my father advising me to inform the University of the difference in addresses. This is the last time I intend to write. "The prompt resolution of this problem may earn you some money. Last year, I failed to donate to the Century Club due to the tardiness of the request in reaching me. Correct the problem, and I'll reconsider this year's donation "More importantly, the resolution of the problem will stop the flow of sarcastic notes-a point I am willing to consider when writing the check to Penn...."

In January 1979, when this letter was received, there were 160,000 active names and addresses in the alumni master record system. This computer file was the origin of most, but not all, mailings to former Penn associates. Penn's reply assured the writer that the alumni master record had been corrected. It could only express hope, however, that no other mailings would be misrouted. There was no way to find and cross- check other address files.

This small example reminds us how important the simplest business information can be, even for an academic institution. In fact, Penn's efforts over the years to collect, store, and make good use of administrative data parallel those of other large enterprises. This article draws from Penn's experience to illustrate trends in the handling of information. It is far from comprehensive; the complete history of administrative computing at Penn remains to be written.

Keeping Penn's accounts

Automated equipment was first installed in the Comptroller's Office in 1950-51. It consisted of electromechanical card-processing devices- keypunches, sorters, multipliers, and tabulators-known collectively as tabulating equipment. These devices were able to perform simple computations and sorting. Machine instructions and data were recorded on punched cards and sorted in stacks. Each file or stack of cards represented one group of variables or records, and each operation on a set of records required the entire stack to be processed. Exceptions, such as employees with nonstandard payroll deductions, had to be maintained on a handwritten list and calculated manually.

Crude as it now seems, this equipment cut the preparation time for the University's payroll from about a week to half a day. This fact is emblematic of what might be called the "age of automation." Efficiency was the business buzzword of the day, the time-and-motion study still reigned in management theory, and labor-saving devices such as tabulating equipment were in great demand.

The IBM Room

The captive nature of its market, however, was not lost on the U.S. Justice Department, which settled an antitrust suit in a 1956 Consent Decree that required IBM to offer its equipment for sale to its clients.

Penn took advantage of this offer almost immediately, purchasing the equipment in the tabulating section of the Comptroller's Office (known as "the IBM Room"), and saving several thousand dollars a year. The Comptroller's office shared its own developing expertise by performing tabulating services for other offices, and helped the Registrar's office to set up its own tabulating division in 1954.

Some positions were eliminated by automation, but a roughly equal number were added to operate the equipment. Those people, fortunately, could process an immensely greater volume of work. The postwar period saw a boom in government-sponsored research at Penn that added heavy new accounting and reporting demands. Penn's transition, beginning in the mid '60s, from a commuter school to a residential school also sharply increased the amount of data that had to be processed.

Still, the labor-saving devices kept just ahead of the growth in the tasks they took over. Some time was left to apply the machines to things that had not been done before. In l955, for the first time, a monthly comparison of actual expenditures to budget appropriations was prepared for each major University division. This "business snapshot" was the shape of things to come.

Managment Information

This computer, like the tabulating equipment it partly supplanted, was at first mainly used to automate existing tasks. Much early programming simply converted the old tabulator programs to the new computer languages, and many existing card stacks of data could be retained without modification. But electronic computers were not only faster than tabulating equipment, they were much more versatile. They also demonstrated a new and unique quality: They got faster and cheaper every year.

With each big drop in the cost per calculation, a new range of uses for computers became feasible. Electronic data processing (EDP) systems began to be designed to provide more ways to sort, view, and report information, to give decision makers a more timely, complete, and sophisticated picture of conditions and trends. Systems with this new emphasis were known in industry as management information systems (MIS). The three factors-an increase in the base number of transactions, the feasibility of performing ever more demanding calculations, and the desire for better management information-fueled a rapid growth in the demand for administrative computing capability at Penn that shows no sign of abating today.

Consolidation

But as more applications were written for the computer, and as the computers grew in power, the benefits of combining the processes into larger, integrated applications became apparent. For example, in 1970 the alumni master file mentioned earlier first combined biographical and demographic data in a single system. It replaced four separate card sets that had contained 42,000 cards each in 1967. Penn's separate personnel and payroll data files were combined in 1975 in a new system appropriately named "Integral."

Consolidation also had appeal for computing operations. In l964- 65, when the IBM 1401 was installed, the computing activities of the Office of the Comptroller and the Registrar were combined under the Office of University Data Processing and Research. Although some offices maintained separate programming and data-entry staff, work increasingly centered around the new organization. The Development Office was the second largest user of computer time at that time, and Admissions, Personnel, the Hospital, the Medical School, New Bolton Center, and the Faculty Club were also clients.

Penn's academic computing center, which had moved to 3401 Market Street in the late l960s, joined with those of Drexel and other schools to form a non-profit consortium known as UNICOLL. Originally conceived as a research support facility, UNICOLL evolved into a commercial provider of computing services.

Meanwhile, the programming team from the Registrar's Office merged in 1972 with the Office of University Data Processing to form UMIS- University Management Information Services. UMIS, now effectively Penn's central computing organization, became a client of UNICOLL-which was looking for commercial revenue-the next year. UNICOLL provided facilities management for Penn's mainframe, and UMIS developed Penn's applications and maintained them and the operating systems. All these changes were part of the general trend toward consolidation, with its advantages of efficiency and consistency.

"One University"

Other offices as well found that the grand unified approach to systems sometimes made them less flexible. Limitations in the security structure of applications such as the personnel system made it impossible to offer department-level access to the data. Instead, departments were provided with data "extracts" containing the information they were authorized to view, which took longer to prepare.

Ironically, local access to information became much more important shortly after the formation of UMIS and partly as the result of a successful UMIS application-a new general ledger system. In 1973, Penn adopted responsibility-center budgeting. The new general ledger system had made it possible to credit each School or center with the bulk of its own revenue, and charge it a proportionate share of the administrative costs of running the University. The new budget structure included algorithms for subvention, from central funds, of Schools whose academic mission was vital to the University but which, by their nature, did not generate great revenues from research and other activities. While affirming the concept of "One University," the new structure gave each unit much greater budgetary autonomy.

Now the responsiveness of a computer system to local needs often became a more visible measure of its success than University-wide consistency or cost-effectiveness. Schools or departments that felt they needed more focused, flexible computing than a central group could provide were able to buy their own machines. As the '80s dawned and powerful, affordable minicomputers came to market, several Schools and departments did just that.

The network and the desktop

PCs changed many things on campus. Perhaps most important, they made computing familiar to people who had previously left it to the "techies." Increasingly, non-computing staff understood the advantages computers could provide and began to employ them in every area of work. The quality of Penn's computing became generally recognized as a factor in its academic and administrative competitiveness. While personal computers in conjunction with PennNet would make possible improved access to administrative data, PCs also greatly increased the pressure on central systems to be more responsive to local business needs. Users wanted the same control and flexibility from central systems they had become accustomed to on their desktop machines.

The adoption in the mid-l980s of the ADABAS database management system for Penn's administrative mainframe brought that flexibility one step closer. Replacing various indexed file structures, a true database management system established a foundation on which much more powerful and sophisticated applications could be built. New development and annual giving, student aid, payroll/personnel, student records, and gifts systems all made use of database technology. Tools such as FOCUS allowed departments to draw their own customized reports from the database files.

The installation of PennNet in l985-87 accelerated these trends. Terminals, and later PCs emulating terminals, proliferated in offices. The growth of this data communications structure meant that more administrative systems could be, and had to be, designed for use by departmental business administrators and other widely distributed staff. SALINC, a central system for processing annual salary increments, was one of the first to provide users with online updating capability (rather than requiring them to enter updates for later batch processing). Its security structure also permitted departmental users to view appropriate data online. It eliminated the handling of some 15,000 four-part carbon forms annually. This system and others like it put Penn in the forefront among universities for distributed online entry of administrative data. Penn took a further step in making central systems more accessible to users with the advent of PARIS, a voice-response input system that allows students to register for courses by telephone.

Cooperative development

Market forces are providing new tools. For instance, institutions are increasingly unwilling to invest heavily in proprietary technology. This behavior is pushing vendors, however reluctantly, towards "open systems" based on common, often nonproprietary, standards. Client/server architectures allow networked computers of different types to divide work between local and central sites.

Equally important are new approaches to application development being used in the Office of Information Systems and Computing as well as in the Schools. Staff users, with their increased sophistication about computing and their intimate knowledge of business processes, will be fuller partners than ever before with information technology professionals in the design of new systems. Techniques such as early prototyping, rapid application development (RAD), and, on campus, Program Management at Penn (PMAP) provide structures for users to participate in key decisions-decisions about where in a business process computers should be employed and how those systems should be configured. This represents a very different model than that in use ten years ago.

Still, much thoughtful and cooperative work remains to be done before Penn can reap the full benefit of these new tools. The most ambitious such effort to date has begun under the name "Cornerstone." The results of this project will help Penn's decision makers recast the University's business processes, its computing architecture, and its financial applications in a new mold-one that will serve Penn and its constituent units equally well through the '90s and beyond. The first in a series of articles on the Cornerstone project will appear in Penn Printout next month.