Science: IBM's age of miracles is over: Two revolutionary lines each reigned for 30 years. It won't happen a third time, says Martin Campbell-Kelly

Thirty years ago this week IBM astounded the computer world by announcing System/360 - a range of computers that would replace its entire highly successful product line. In a telecast to its sales offices around the world - a noteworthy innovation in itself - IBM's youthful, Kennedyesque president, Thomas Watson Jr, announced a range of six mainframe computers and 40 peripherals - printers, magnetic-tape storage units, disk drives, and so on. The name System/360 'betokening all points of the compass' was chosen to suggest the universal applicability of the new machines.

While a major IBM announcement had been expected, the breadth of the System/360 launch came as a bolt from the blue. Computer pundits were taken aback by its scale. For example, a journalist from the business magazine Fortune, who coined the phrase 'IBM's dollars 5bn gamble', wrote: 'It was roughly as though General Motors had decided to scrap its existing makes and models and offer in their place one new line of cars, covering the entire spectrum of demand, with a radically redesigned engine and an exotic fuel.' It was believed that System/360 was the largest civil R & D project ever undertaken, and 'not even the Manhattan Project, which produced the atomic bomb in the Second World War, cost so much.' Although the development of System/ 360 was a high-risk strategy for IBM, it was a path that it was forced to take to deal with the looming 'software crisis'. By the early Sixties, software was proving more expensive for computer manufacturers to develop than hardware. Yet whenever a new computer model was introduced, most of the existing software had to be thrown away. When users replaced their first-generation computers with transistorised, second-generation models, all the software had to be rewritten. This sometimes cost more than the new computer, and often resulted in horrendous organisational disruption during the change-over.

System/360 - which IBM called the third generation - was designed to be software-compatible throughout the range, from the smallest machine to the largest. For IBM this meant that it would have to write only one set of software for the complete product line; for IBM's customers it meant that their investment in software would be protected when they upgraded. It also meant that once customers had bought IBM they were 'locked in', as they could only move to another manufacturer if they were prepared to scrap their software. Of course, IBM was trapped, too - it would be forced to maintain software compatibility with System/360 for 30 years and more.

System/360 has been described as 'the computer that IBM made that made IBM'. It was an audacious and dazzlingly successful product strategy. IBM already had three- quarters of the world market for mainframe computers, and System/ 360 enabled it to sustain that position for another 20 years. There was, however, an Achilles heel in the strategy. IBM had created an industry standard which left it vulnerable to competition from 'commodity' suppliers of computer equipment.

The first manufacturers to spot this chink in IBM's defences were manufacturers of peripheral equipment who began to sell 'plug-compatible' tape and disc storage devices that could simply be hooked up to any IBM installation. Fortunately for IBM, its capability in electro-mechanical engineering was second to none and this - with its enormous advantages of manufacturing scale - limited the damage to its profits.

IBM had a much tougher time competing with its electronics technology. Electronics was not a real strength in IBM and, although it relaunched its computer range as System/370 in 1970 using state-of-the- art integrated circuits, it remained an organisational weakness. Recognising this, Gene Amdahl - a defector from IBM who had been one of its principal computer designers - decided to manufacture a plug- compatible processor. This would be functionally identical to an IBM mainframe, enabling it to run the same software - whether written by IBM, the customer or a software house. An IBM-compatible processor would be able to replace the very heart of an IBM installation. Eventually it would be possible to run an entire IBM-compatible installation without buying a single item from IBM.

Amdahl made a licensing arrangement with Fujitsu and by 1975 was selling processors with more than twice the price-performance of IBM's. This was the turning point for the Japanese computer industry. Soon Fujitsu was followed by the likes of Hitachi and NEC, selling high-performance, low-price mainframes.

IBM responded by making a huge investment in semiconductor technology and, in 1979-80, relaunched its mainframe range with dramatically improved price-performance. This sent mainframe profit margins tumbling around the globe and was, for example, the principal factor in ICL's near-bankruptcy in 1981. Like Amdahl, ICL turned to Fujitsu for semiconductor technology - and this eventually led to it being taken over by the Japanese company in 1990.

Since then mainframe profit margins have continued to be eroded by the twin forces of Japanese competition in big computers and the burgeoning market for personal computers. Last year declining mainframe profits led to IBM recording the largest loss in corporate history.

In 1996 IBM will celebrate its centenary - assuming it is in any mood to celebrate. Between the two world wars it became stupendously profitable through its phenomenally successful '400 series' punched-card accounting machines. They reigned for 30 years - it was not until 1962 that IBM's computer revenue outstripped the income from its electro-mechanical products.

If IBM had not been able to leap on to the new technology of computers, it would have been doomed to slow decline. Instead it grasped the computer like no other company and became the third largest commercial enterprise in history.

There is a compelling parallel between IBM's pre- war accounting machines and System/360 as it, in turn, reaches the 30th year of its product life cycle - and, surely, approaches its demise.

Alas, as far as anyone outside the company knows, there is no new information technology waiting in the wings to propel IBM another 30 years into the 21st century. Unless it comes up with something, IBM faces the sure fate of any dinosaur.

The author is a computer historian at Warwick University.