A family of computers
When IBM announced the System/360 in April 1964, it was making an unusual bet. Until then, the company sold several unrelated lines of computers, and a customer who outgrew a small machine usually had to rewrite programmes for a bigger one. The 360 was presented as a single family of machines, from small to large, that ran the same software and used the same peripheral equipment. The name was a reference to the full circle: the range was meant to cover every purpose, scientific as well as commercial.
The project was enormously expensive. IBM reportedly invested billions of dollars in developing the machines and the factories to build them, and the company’s leaders knew that failure could have threatened its future. By 1967 the gamble had clearly paid off. Orders had been large, deliveries were under way, and IBM’s revenues in that year were on the order of $5 billion, well ahead of any rival. Competitors, often grouped together as the “seven dwarfs” next to IBM, included Burroughs, Univac, NCR, Control Data, General Electric, RCA and Honeywell. Most of them trailed IBM in installed machines by a wide margin.
What a 1967 computer centre looked like
A mainframe was not something an office could simply plug in. It was installed in a dedicated room, often with raised floors for cables, extra air conditioning and a staff of operators, programmers and technicians. The central processor was a set of tall cabinets. Beside it stood tape drives with spinning reels, disk drives that held a few million characters on removable packs, a printer that produced fan-fold paper at high speed, and a card reader and punch.
Punched cards were still central. Data and programmes were typically prepared on keypunch machines, each card holding 80 characters, and fed in as stacks. Output often came back as printed reports, and a large organisation might produce huge volumes of paper each month. Programmers submitted “jobs” to be run in batches, and might wait hours or overnight for results. Interactive terminals existed, but the typical business user did not touch the computer directly.
What companies did with them
The main uses were routine paperwork at scale. Banks used computers for accounts and cheque processing, insurers for premiums and policy records, and manufacturers for payroll, inventory control and production scheduling. Retailers and utilities used them for billing. Government agencies, including the Census Bureau and the Social Security Administration, ran some of the largest installations. Universities and research laboratories also used 360s, as did companies working on engineering design.
Some of the most visible applications were online systems. Airline reservation networks, of which American Airlines’ SABRE system was an early and well-known example, linked terminals in booking offices to a central computer so that seats could be sold in real time. Banks began to connect branches to central records, a step that eventually led to the cash machine. These systems were costly and complex, and in 1967 they were the leading edge rather than the norm.
Software was the hard part
The hardware attracted attention, but the programmes were a bigger problem. The operating system for the larger 360 machines, known as OS/360, was a very large software project and reportedly ran well behind schedule, with shortcomings that took years to resolve. The experience later became a standard reference point for how difficult it is to manage big software projects. Programmers worked mostly in assembly language, COBOL for business data and FORTRAN for scientific work. At this time, software was typically supplied with the machine rather than sold separately; IBM’s decision to price software and services apart from hardware came later, at the end of the decade.
Skilled staff were also in short supply. The occupation of “data processing manager” was growing, and companies competed for people who understood both the machines and the business. Many organisations discovered that installing a computer changed how work was divided, and that records which had previously been scattered in filing cabinets now needed to be gathered and cleaned before a computer could use them.
Compatibility and its consequences
The idea of a compatible family proved influential. A company could start with a small model and move up without starting over, which gave IBM a strong hold over its customers. Other manufacturers responded either by building machines that could run 360 software or by targeting specific niches. At the smaller end, minicomputers such as those from Digital Equipment Corporation, which cost far less, were starting to find buyers in laboratories and factories. Time-sharing, in which many users share one machine through terminals, was being explored at universities and a few commercial firms.
The size of IBM’s lead also began to attract the attention of regulators, and the company was later the subject of a long antitrust case that began at the end of the decade. That belongs to a later story.
Why it matters
Computers in 1967 were mostly a tool of large organisations. Few individuals had direct contact with one, but many encountered the results in the form of computerised bills, bank statements and payslips. Concern about automation and about the “computer age” was already a subject of public debate, and some people worried that machines would replace clerical workers or reduce individuals to numbers on a card. The same machines also helped manage the work of the space race and large government programmes. In the wider economy, the System/360 helped make data processing a standard part of running a big business, laying the groundwork for the more decentralised computing that followed in later decades. The figures given above are approximate, and sources differ on some details of IBM’s costs and sales.