Computers: The Supply Equals the Demand, But the Money Might Be Hard to Come By

The SDS 940, core of Harvard's time-sharing computer network, has an unlisted phone number. Dialing that number rewards the caller with a unique, minor-key whistle, or else a busy signal, if 32 of some 125 SDS teletype units distributed around the University are already in conversation with the central computer.

Most of the time fewer than the maximum 32 are in use, and the answering signal precedes a minute or two of silence during which the caller turns from telephone to teletype and begins to make conversation. If no teletype contact is established during that period, the computer gas impatient and cuts off the caller.

Bell Telephone installs the teletypes, or "consoles," for $75 a month plus line charges, and, according to David D. Dix, associate director of the Harvard Computer Center, is the only institution to clear a profit off time-sharing. For the Computer Center, and for Harvard, the attraction of the SDS 940 lies in its ability to solve comparatively small problems instantly, and to serve users all over the University.

The competition consists of four other computers ranging from the IBM 1401, which Dix describes as "old and uninteresting" and which is now used principally for printing output; to the IBM 360-50, a more modern affair which, among its other duties, analyzes data from the Cambridge Electron Accelerator; to the two IBM 7094's, giant computers which solve complex problems in organized batches, hence the title "batch-processing."

Time-sharing and batch-processing are opposite methods of computer work. With the first, the users are "on-line," which means the computer functions precisely when it is asked to by the user; with the second, efficient use of the computer requires htat the user assemble problems of just the right size, and feed them in at just the right rate. For the student or Faculty member submitting a question downstairs at the Computer Center, a wait of two or three hours is usual.

Software

When the Computer Center was first built, in 1946-47, there was considerable interest in the design of computers. Now, says Dix, "we're only interested in using them." Software--specialized programs--has replaced hardware--computers--as the Center's main area of innovation.

Simultaneously time-sharing has started to expand faster than batch-processing. The Medical School, and medicine in general, have started buying large chunks of time on the SDS 940. One new project responsible for this upsurge is a heart-care unit to monitor continuous heart-patient problems, and to detect subtler signs of danger than the cardiologist can ordinarily hope to notice. Consoles also aid interviews with patients suspected of having genetic problems, by rapidly accumulating genetic history.

At the undergraduate level, an eight-man committee appointed by Dean Ford has recommended that consoles be installed in the Harvard and Radcliffe Houses, and selected freshman dorms, for unsponsored use. Frederick Mosteller, professor of Mathematical Statistics and the committee's chairman, points out "if you start a thing like this, there'll probably soon be more of a demand for it." To date, there has been little demand from students for access to computers, but the departments have independently been requesting funds from the Faculty for hardware, software, or both.

Mosteller and others frown on this trend. They see money being wasted on new hardware while the existing computers continue at less than capacity. And if more hardware should be required, Mosteller's committee would like to see it purchased by the Computer Center rather than by individual departments.

Instructional Use

The other major recommendations of the committee are directed toward increased instructional use of computers. At present, Dix says, "I would guess that about 80 per cent of the money spent here is for research, and most of the remaining 20 per cent for classroom use, probably most of that by the Business School." The committee asks that "the instructional use of computers should have a priority equivalent to that of the research uses."

Though the Center does virtually no direct work for the government, it receives a good deal of Federal Funds. Enough, Dix says, that it must abide by the government's edict against idle time being used free of cost by students. The Center must collect for every job done on its computers, and for what technical assistance it give those who use its facilities.

"One of our big concerns has been how to make more students use possible in the face of this requirement," Dix says.

4 Hours Nightly

The solution outlined by Mosteller's group is two-fold. First, the Houses would buy time on their own consoles nightly from 8 p.m. to midnight for relatively unrestricted undergraduate use. Second, the Faculty would allocate a separate fund, to be administered through the departments, for course work involving greater input and therefore requiring the 7094's.

The consoles are inadequate, says Mosteller, because "it is possible for a man to think up a legitimate problem that would tie up the computer." The SDS 940's on-line operation involves a relatively small amount of total input, even when all 32 possible teletypes are in contact.

As a rule, computer work done by math, statistics and physics students is of the sort the SDS can accommodate: small in input, even when the solution process is complex. The Humanities and Social Sciences, and in some cases the Natural Sciences, often require work with a much larger input, sometimes too large for the time-sharing system.

The inefficiencies suggested by departmental requests for computer hardware were only part of the reason for creating Mosteller's committee. It was also prompted by the dramatic leaps which another school, Dartmouth, has recently taken in the instructional use of computers.

Dartmouth has its own time-sharing system, somewhat different from Harvard's, but more extensive and more extensively used over the last few years. Many of the computer committee's suggestions were based directly or indirectly on the comparative success of Dartmouth; "Harvard," Mosteller says, didn't seem to us "to be leading or perhaps even keeping up."

The committee did not recommend any new computer purchases, concluding that the existing facilities were inadequately utilized now. But if Harvard has enough computers, its computers and their operators seem not to have enough space. Already the Computer Center is splintered off into several separate locations in Cambridge and one in Boston proper.

Dix sees another obstacle to expanded computer use at Harvard--the "have" school's difficulty getting Federal funds. Mosteller attended a computer conference this weekend at the University of Maryland, where the main concern was the setting up of conferences to help specifically small schools gain access to computer networks.

Dartmouth, Dix says, has "considerable private endowments" in the computer area. Harvard, to undertake the kind of expanded computer activity recommended by Mosteller and his colleagues, will have to eke out funds from the Faculty, which this year ran an unusual financial deficit.

The prospectus, despite the computer committee's 72 pages of explication, is not all rosy, which makes it the more remarkable that Mosteller can declare, "We'd like to be able to say, 'The equipment is available, go and use it.'"