IBM System/360

IBM PC, early 1980s

Today's youth (speaking with no intent to impugn their intelligence) can look as far back as the 1980s and say that they understand the history of that era. This was the decade of the PC's mass acceptance, which is heralded as the beginning of the computer age as it affected most Americans.

But this is somewhat shortsighted. Computers were long in use before they became affordable to middle-class households. The first business computers were unveiled in the early 1950s. IBM was actually a latecomer with its 701 in 1952; Remington Rand's UNIVAC division had delivered the UNIVAC I to the US Census Bureau in 1951.  

The IBM 701 could be considered the first "mass produced" computer, in that each computer was assembled with the intent of precise replication for future models of the exact same type. Still, it could hardly be described as high-volume; Thomas Watson, Jr., IBM's CEO, unveiled the computer at a business meeting of which he noted: "as a result of our trip, on which we expected to get orders for five machines, we came home with orders for 18." A total of 19 were installed, which pales in comparison to 46 orders for the UNIVAC I. In either case, there was no general-purpose computer that sold in great numbers. Very large businesses could afford a system designed just for them, a process which financed the state of the art, but did not trickle down to medium-sized and smaller businesses, which have always held the decentralized but largest part of the economic clout in the American economy.

IBM 701 console

The problem facing widespread acceptance of computers in business and government is that one size does not fit all. The needs of American Airlines (for them, IBM developed the groundbreaking "Sabre" reservation system in 1960, a basic system still in use today) were not the same as a hypothetical Columbia Steel Pressings, Ltd. CSP, let us assume, is an energetic little company. Their foundry could use a computer that would quickly calculate coal-air mixtures that would produce the hottest flames at the lowest cost, but that was about it. With only 400 employees they didn't need a computer to handle hours or payroll or any other mundane tasks which were still possible by adding machine and ledger.

Before System/360, CSP didn't have a truly desirable option. They might do some research and realize that they just needed to get a small-scale computer for their limited needs at the moment, but if the company suddenly found great success, they would not be able to allocate much in the way of new tasks to the small computer they had bought, which certainly cost more than $100,000. If the demand for computer tasks escalated, they would need to spring for a computer of greater capability, which could cost in the millions. The initial investment was not only in the machinery but in the programming, which was much less portable and more system-specific. This labor would be lost altogether if the programs originally used were not compatible with the new system, which meant that virtually all companies would prefer to avoid costly experiments in the computer field, in some cases making them wary of even entering the field in the first place.

 Single 360 CPU at use at a Volkswagen office.

Massive 360 (Model 91) used by NASA

System/360, unveiled in 1964, was a truly innovative scheme. We know of computers now as being very mutable and modifiable machines, with easy-to-change modules and components. But prior to 1964, this concept wasn't yet actually realized. However, it was realized in the 360. When a small business bought a basic System/360 with a slow CPU and tiny onboard memory, it was always possible to migrate upwards with more peripherals, memory modules or even with a new system altogether. If they bought a new system in a few years, the old programs they had used were intact and fully compatible. (In fact, some programs written for the System/360 architecture are still compatible with present-day IBM System z mainframes.) The promise of no duplication of effort and cost was very alluring to many customers. Furthermore, IBM made the System/360 capable of emulating its earlier computers, so that existing customers would be able to buy the new system for its technical capabilities and still execute some of the same programs they previously used.

While the extremely good execution of IBM's idea is noteworthy, I am interested more in the idea itself.

The S/360 actually created the concept of a modular machine before experts even perceived that a demand for this market existed. For that reason, 360 was a spectacular gamble and was labeled as such in the press. At a time when IBM was one of the world's largest companies, it was literally betting the company's public image, fortunes, and marketability simply on the success of one product. It was so far ahead of its time, while simultaneously being so successful, that I cannot imagine a machine that so quickly and painlessly brought humankind further in technological development. 

The scale of Thomas Watson, Jr.'s gamble cannot be overstated. The project took three years and over $5 billion to develop. $5 billion in early 1960s dollars corresponds to nearly $38 billion today. In other words, the development of one computer system, by a single private company, was nearly one-quarter the cost of the entire Apollo program which placed the first humans on the Moon, which had the benefit of enormous public funding and literally hundreds of contracts for designs. The mind of the average person in the 1960s could no sooner grasp $1 billion than people of today can grasp $1 trillion.


This gamble paid off extremely handsomely. IBM's introduction of the machine was met with more than a thousand orders in the first 30 days. Between 1960 and 1968 (with the S/360 coming out in the midpoint of those years) the market for mainframes exploded from $600 million per year to $7 billion per year. That is compounded growth of 36% annually! IBM's market share ballooned to over 70% of computer sales globally. 


Seldom has a company ever been in a more enviable position. 


If we want to draw comparisons, General Motors's highest market share in the American market stood at almost 60% at various points in the 1970s. The Ford Model T outclassed everything else for almost two decades, but when it was retired in 1927, only about half the cars in the world were Model T's, and Chevy outsold them that very same year. Standard Oil might have achieved 90% control of the US oil market using somewhat extra-legal means in the 1890s, but it was not as dominant in other world markets. IBM's dominance was global, since no regional players in Europe, Japan, the UK, or the USSR had anything like what they offered. Protectionism in the computer industry, save for the Cold War alliances (which rendered profitable some rather backward computer firms in the Soviet bloc), has not been a major handicap for giants like IBM, and later Intel and Microsoft.

True innovation is so hard to come by. It's usually only present in the minds of those who haven't yet found great success. For a successful, large corporation to reinvent the business that it has previously dominated, and rely just on the strength of its new product, is absolutely unprecedented. This is what IBM did; it wasn't just looking for good year-on-year profits and a good 90-day sales report. What Watson and his company did was significantly advance the state of the art and show IBM's creativity and engineering competence. These strengths would carry them forward for decades.


My unscientific opinion is that this spirit of creativity and engineering know-how is all that America has left as we careen into the 21st century with a battered economy. First Japan, and then Korea, and now China have steadily taken over manufacturing of technologically-advanced parts and machines. However, the ideas, designs, and testing of the world's most popular software and hardware continue to come from American companies who employ many domestic engineers and programmers.  Intel might outsource every single fabrication process, but unlike with baby toys, socket wrenches, and DIY furniture, the most expensive part of building chips is the design and testing, which is referred to as NRE (non-recurring engineering) cost. And as far as American tech companies are concerned, most of this is still spent in American offices, labs and universities. That's happy news for me, since I hope to be joining the rank of electrical engineers in a year or so. But still I do worry about job security if we can't stay competitive.


Back on topic: Far be it from me to claim that the capabilities of these old dinosaurs are in any way competitive with modern machines. If you take a step back 40 years, you will have missed out on all those years of Moore's Law, and end up with very primitive machines indeed.  Still, they were extremely reliable and they didn't really adhere to the concept of "planned obsolescence." That's why I want an IBM Model M keyboard; they're built like nothing else.

It is worth noting that the System/360, and even mainframe computers in general, were maligned as "archaic" and in a state of great decline by the 1980s. Some businesses attempted to use parallel implementation of many cheaper PCs to achieve their computing needs. As computers became more powerful, computing demands did not necessarily keep pace, enabling firms to replace their "big iron" with smaller, cheaper machines. But if stability is the most important factor, mainframes are still essential to business. Sometimes the daunting purchase price would be worth it to receive excellent service and complete customer care that only IBM provided to its elite customers. It's a far cry from today's machines, which come with perhaps two pages of instructions in 43 languages and with a rough English translation.


IBM's reputation in mainframes continued to garner some support for their smaller machines; even as the minicomputers and PCs were seen as increasingly overpriced and less innovative later in the 1980s. It was still occasionally remarked that "Nobody ever got fired for buying an IBM." 


This is cold comfort for IBM, which has been hemorrhaging employees and profits since the early 1990s. Despite extremely good advertising and public image, IBM is as small and weak as it has been for most of the 20th century. As with IBM's glorious past, little physical evidence exists of the System/360. Few remain, as it is quite expensive to store and maintain such large computers, and there is significant scrap value in the precious metals inside the machine.

Still, in my mind, when I compare the sheer influence, innovation, and success of the S/360, I would place its greatness above any of the other pioneering mainframes, any Apple products, or even the original IBM PC. My opinion is that the IBM System/360 is the greatest computer of the 20th century, which must make it pretty close to the greatest technological achievement of all time.