The Computer College

Every once in a while, somebody alarms about half the Masters with the suggestion that assigning Freshmen to House be the job of "some fellow with a computer." Such a scheme sounds like pure buck-passing, and in some ways it is just that, for the preparation of a complex problem for processing by any kind of automatic calculating device naturally takes a good many hours' work. And yet, as it happens, the computer alternative to the system that now overwhelms House staffs each spring is--if conceivably not desirable--at least feasible. It could almost certainly be made to be made more efficient than the present process.

According to Anthony G. Oettinger, director of the University Computation Center, Harvard would have to solve three principal problems if the Masters reversed themselves and resolved to use an automatic assignment system. The first, and possibly the most clusive, of these problems would be to decide exactly what personal qualities should be significant in determining where a Freshman is to be sent, and exactly how these qualities should be measured.

The Masters, in other words, would have to agree upon upon the precise criteria according to which the machine would distribute students. They would have a wide range of decision. Octtinger points out that the Masters could if they wished insure that the machine take into account a student's order of preference for particular Houses. Furthermore, a computer could be programmed to put slightly different groups of students in each of the eight Houses, and to preserve the present shades of difference between them. Agreement on what these differences are or ought to be, on the criteria for "weighted distribution," Oettinger suspects would be difficult to secure.

The second problem he forsees is getting the relevant information about each student recorded in a "machine-readable" form. Someone would have to cull the records of each freshman, punching the appropriate facts about him onto an IBM card, or typing them onto a magnetic tape. In Octtinger's view, if such a body of taped information had no other purpose than to facilitate House assignments, the effort spent to compile it would probably be unjustified. But he can imagine several other uses for it, all involved with the various sections of University records. The information office, for example, the Alumni Records Department and the Registrar's Office, could all code data about students into the same tapes used for House assignments--since the files of these offices in many respects duplicate each other. Everyone could use the same file, the memory section of a computer. Needless duplication might in addition vanish from the famous jungle of the registration process; registrants would fill out one card rather than eight.

The third problem, intimately connected with the first, is to decide the secondary "rules of the game": which factors should have priority, where to send students when the Houses they have applied for have no more vacancies, and so forth. Given this information and a solution to the first problem, Oettinger calculates, the actual writing of a program for the University's 7090 computer, or setting up the plug-board for a mechanical card sorter, could be done "by anyone." (He considers the choice between the sorter and the 7090 computer, like that between cards or tape for recording information, a relatively unimportant technicality. "Machine operating time in either case would be only 15-20 minutes.")

An automatic system has the large advantage that the program once written can be used, with only a little extra effort, year after year. One need only repeat the work of recording each successive class on magnetic tape.

Oettinger confesses that the prospect of making decisions about human beings on the basis of numerical criteria is scarcely heartening, but points out that the University is making many such decisions already, especially in the Admissions Office. Anyway, he estimates that in the huge majority of cases the machine would place students in exactly the same House as the Masters and the present system. And he adds encouragingly that the automatic system could doubtless inform the Masters of any extraordinary or borderline cases, and so allow ordinary human judgment to assert itself again.