In the waning days of the second World War, Vannevar Bush, science adviser to President Franklin Roosevelt, penned a report that served as the blueprint for what would become America's enormously successful information-technology industry in the second half of the 20th century. With the grandiose title “Science, The Endless Frontier”, Bush (no relation to the current president) laid out a vision for government-funded science and engineering that would unite academia, industry and (this being wartime) the armed forces. This it achieved by, in effect, keeping them apart.

Under Bush's plan, universities researched basic science and then industry developed these findings to the point where they could get to market. The idea of R&D as two distinct activities was born. Firms soon organized themselves along similar lines, keeping white-coated scientists safely apart from scruffy engineers.

This approach was a stunning success. AT&T's Bell Labs earned six Nobel prizes for inventions such as the laser and the transistor. IBM picked up three, two from its Zurich Research Laboratory alone. And Xerox's Palo Alto Research Centre (PARC) devised the personal computer's distinctive elements, including the mouse, the graphical user interface and the Ethernet protocol for computer networking (although it was criticized for failing to commercialize such leaps forward).

Now the big corporate laboratories are either gone or a shadow of what they were. Companies tinker with today's products rather than pay researchers to think big thoughts. More often than not, firms hungry for innovation look to mergers and acquisitions with their peers, partnerships with universities and takeovers of venture-capital-backed start-ups. The traditional separation of research and development enshrined by Bush in 1945 is rapidly disappearing, especially in the information-technology industry. Does this mean the days when companies came up with big breakthroughs are over, too?

Not necessarily. The approach to R&D is changing because long-term research was a luxury only a monopoly could afford. In their heyday, the big firms dominated their markets. AT&T ran the telephone network, IBM dominated the mainframe-computer business and Xerox was a synonym for photocopying. The companies themselves saw the cost of basic scientific research as a small price to pay for such power.

Modern technology firms are much less vertically integrated. They use networks of outsourced suppliers and assemblers, which has led to the splintering of research divisions. Even though big American firms still spend billions of dollars on R&D, none has any intention of filling the shoes left empty by Bell Labs or Xerox PARC. The research and development that Bush tore asunder are once again becoming entwined. Old-fashioned R&D is losing its ampersand.

On the surface, American innovation has never been stronger. American firms spend around $200 billion on R&D annually, much of it on computing and communications. Microsoft, for example, spent around $6.6 billion last year; IBM and Intel about $6 billion each; and Cisco Systems and Hewlett-Packard (HP) around $4 billion each. Most of this money went into making small incremental improvements and getting new ideas to market fast.

For example, IBM has eight laboratories on three continents, each with its own personality and expertise. At its Zurich Research Laboratory around 300 scientists representing over 20 nationalities concentrate on areas such as microelectronics, nanotechnology and computer security. Only a few years ago researchers were judged on the basis of patents and papers, but today they roll up their shirtsleeves and work alongside the company's consultants, explains Douglas Dykeman, one of the laboratory's managers.

This reflects IBM's transition into “services science”. The services business is becoming commoditized, as hardware did before it, and IBM knows it must add intellectual property to its offerings. Putting its researchers on the case is one way to charge clients a premium.

The fusion of research and development is meant to solve the central shortcoming of Bush's plan: how to turn ideas into commercial innovations. Great ideas may moulder without a way to develop them. In America the link between industry and government-funded research was reinforced in the 1980 Bayh-Dole Act. This expected recipients of federal funding to patent their innovations as an incentive for them to leave the laboratory. So, for example, when Google listed its shares in 2004, Stanford University received around $200M worth since research by Google's co-founders on search algorithms had been partly financed by the National Science Foundation. Moreover, the rise of venture capital has smoothed the progress of new ideas into products.

One reason for the shift towards more commercially minded research in technology companies is that the nature of IT has changed so much. In Bush's time the science that went into computing was itself closer to basic research. By contrast, many of the big scientific questions in computing have been answered—at least well enough for companies to find that innovation emerges from new ways of arranging today's technologies rather than inventing new ones. Dell's innovation was a business model that used extreme supply-chain efficiency to create bespoke computers. Likewise, Apple's iPod is a new interface atop standard industry parts.

The post-war R&D model needs updating not simply because of the new way of innovating, but also because companies now have a greater choice of where to shop for ideas. “If you go back to that period of AT&T, there wasn't the same kind of engineering going on at universities, and Bell Labs could not rely on an external cadre of engineers for their research—they really had to do it internally,” says Charles Giancarlo of Cisco.

He admits his company has long been accused of “R&D by M&A”—but he does not see why that should be a problem. Cisco spends $4.3 billion a year on employing over 16,000 engineers around the world, in addition to tapping into the venture-capital industry's start-ups. Basic research inside companies is impossible in a competitive industry, according to Mr. Giancarlo. “We might decry this on a public-policy basis, but at least as far as public markets are concerned it is a Darwinian world. You live or die by that.”

Some private institutions have sought to enter the field of “basic science with a business plan”, so to speak. But their success remains to be seen. In the 1990s Paul Allen, one of Microsoft's founders, set up Interval Research to great fanfare, but closed it within a decade after spending $100M with little to show. More recently, Nathan Myhrvold, Microsoft's former chief technology officer, created Intellectual Ventures, to come up with futuristic technologies and license them widely. It has filed over 800 patent applications, but has yet to publicize any big innovation.

Perhaps all this would have made Bush weep. “Industry is generally inhibited by preconceived goals, by its own clearly defined standards, and by the constant pressure of commercial necessity,” he wrote in 1945. “Basic research is performed without thought of practical ends.”

But the message from mammon is different. “When I started out running PARC, I thought 99% of the work was creating the innovation, and then throwing it over the transom for dumb marketers to figure out how to market it,” says Mr. Brown. “And now I realize that there is at least as much creativity in finding ways to take the idea to market as coming up with the idea in the first place. I would have spent my time differently had I figured this out early on.”

For the entire article, go to www.economist.com.