The history of the computer industry is littered with companies that had a head start in some promising technology, only to squander it, leaving the field to rivals who were more alert.

Xerox, for instance, invented the graphical user interface, which is now central to desktop computing, but never used it to break out of the photocopier business. International Business Machines (IBM), which employs 4,500 people in the Republic and is king of the mainframe, missed the mini-computer boom.

Big companies in all industries routinely fail to exploit the new markets emerging under their noses. Nowhere, though, have the mistakes been as glaring as in computing, thanks to the frequent technological shifts that have opened the door to successive generations of newcomers.

IBM has weathered the transitions better than many: it is at least still around to tell the tale and, after being derided for missing out on the final leg of the tech-industry boom, is showing greater resilience than most to the bust that has followed. But, conscious of the big markets it has failed to conquer, it has set out to hone the way it identifies and pursues promising new ideas.

It is too early to tell whether the company will succeed - but its methods bear scrutiny by all big companies that fear missing the next big thing.

The motivation for IBM's initiative rests partly in its sheer size. With annual sales of $90 billion (#102 billion), meeting a growth target of nearly 10 per cent a year means leaving no stone unturned in the search for extra revenue.

"That's like creating an Oracle or an EMC every year," says John Thompson, IBM's vice-chairman. `It's not a small task to grow the elephant.`

Nearly a year ago, Mr Thompson was handed the job of nurturing the businesses that have the potential to be the next blockbusters. Only those thought capable of revenues of more than $1 billion within a few years make the cut.

The process begins in what looks like traditional IBM fashion, with a formalised and centralised process of review and decision-making. The company that turned corporate bureaucracy into an art form has not completely changed its ways.

Two studies are produced annually for the benefit of the company's top executives. One, called "Global Market Trends", synthesises the forecasts of academics, consultants and other industry experts to produce a list of the eight to 10 trends that are expected to have the most significant impact on the technology industry.

A separate study, produced internally, focuses on the big developments under way in particular technologies, resulting in a list of the 10 to 12 most important technological advances.

Mr Thompson makes no apologies for this resolutely topdown approach. The organisational structure of a big company such as IBM means that good ideas might otherwise never be identified or percolate up. New businesses "sometimes fall in the white spaces between organisational boundaries. Sometimes they conflict with existing business units," he says.

IBM's current list of most promising new businesses is a bit of a mish-mash. It includes initiatives tied to particular technologies - the company's well publicised backing for the Linux open-source operating system, for instance, or a unit developing microprocessors for data networking equipment.

Others are geared to a particular type of customer, such as a business unit that develops technology for the life sciences industry, a market that IBM believes will reach $30 billion within a few years. Several initiatives span IBM's operations.

For instance, a unit formed to pursue "pervasive computing" - the extension of computing power from the personal computer to mobile handsets and other devices - is intended to stimulate sales of a range of IBM's chips, servers and software. What links these ideas is the belief that they need a different management approach to that typically applied.

Internally, they are dubbed "Horizon Three" operations - as opposed to Horizon One (mature businesses such as the mainframe) and Horizon Two (growth businesses that are extensions of existing operations such as adding web-hosting to the outsourcing business - which IBM looks to for the bulk of its growth in the short term).

Horizon Three businesses have more in common with start-ups and need to be protected from the pressures that can squash them in a big company. "We have been good at seeing the technologies but not so good at developing them," says Mr Thompson. "The problem was, we didn't put our best people on them, or we didn't see them through." That has led to a number of strategies designed to manage these projects differently.

The first lesson, according to IBM executives, is that new ideas need to be separated from normal operations. "You almost always have to put a dedicated team on it," says Mr Thompson.

Setting up a separate unit to pursue ideas in pervasive computing was needed "to give it visibility and management sponsorship - to protect it", says Michel Mayer, head of that initiative.

Such teams overlap extensively with IBM's more established business units, however, and must find ways of working with them. The membership of the specialised pervasive computing team has now climbed to 600 people, but "two or three times that number" work on pervasive computing projects inside other units, says Mr Mayer.

To promote cooperation, the new units use their own investment budgets to back joint investments with established divisions. The heads of these divisions "understand the importance of finding new areas and growing their business", says Caroline Kovac, head of the life sciences group.

There has also been an attempt to insulate young businesses from IBM's traditional management methods and performance yardsticks.

"Sometimes we make the mistake of managing every business the same," says Mr Thompson. He says that in a big company, the accountants and efficiency experts get their hands on potentially promising ideas and can squeeze the life out of them. Rather than being obliged to meet detailed financial plans, new businesses are given broad revenue and investment targets.

The fledgling operations often need to be measured differently, he adds. As a former head of IBM's services division, he found himself championing a unit whose profit margins are lower than those in IBM's hardware and software businesses. But when judged in terms of return on capital, the services business produces attractive results.

"We make a mistake often of putting a new idea inside a business that is measured in a different way," he says.

A new initiative will survive only if middle managers with the power to block development realise that it has active support at more senior levels.

Some degree of resistance is inevitable. "It happens in all big companies," says Mr Thompson. "People fall into silos and develop their own agendas."

To overcome that, senior executives have to take clear leadership in backing new ideas. "Because I'm a watchdog over it, no one can chip away at it," says Mr Thompson.

"The days are gone when a manager down in the trenches can decide, `I don't like that project'." The heads of IBM's Horizon Three businesses say that they rely on this personal sponsorship. "Occasionally, I will call on him to help resolve issues when we have to make a trade-off between the long term and the short term," says Ms Kovac.

"People need to have a champion. They need to be protected."

Mr Thompson, 59, will be around to provide that personal sponsorship only for another year or two before he retires. By then, Lou Gerstner, chief executive, may also have gone. If IBM's new growth initiatives are to be kept from fizzling out, they will have to find the same level of support under the next management team.