But many businesses have come to view the sieve as a tool that's too rigid for modern times.
In its place is a model in which ideas come from both internal development and by licensing technology from others. Some ideas result in products for the company that came up the concepts, while other ideas can be licensed out.
The result, proponents say, is a model of "open innovation" in which good ideas are not lost simply because they don't fit with the company that developed them.
"We actually have to open up our business models," said Henry Chesbrough, who wrote a book titled "Open Innovation" and serves as executive director of the Center for Open Innovation at the University of California's Haas School of Business.
Chesbrough was part of a panel that discussed the shifting innovation model Wednesday at an event at Microsoft's Silicon Valley offices.
The panel was tilted heavily in favor of those who say the shift is both happening and mandatory.
"Big companies, they are either going to have to learn from this or they are going to have to die," said panelist Jim Huston, a former Intel Capital executive who's now a managing director for Blueprint Ventures, a San Francisco Bay Area-based firm that helps big companies spin-out ideas that don't fit with their business plans.
Chesbrough said the sieve notion made sense when big companies were responsible for the bulk of research spending. "In the recent past, the logic really eroded," he said
That's because a growing amount of research is coming from smaller companies. In 1981, companies with more than 25,000 employees accounted for $7 out of every $10 spent on research and development. By 2001, they accounted for less than $4 of the $10 dollars spent, with a quarter of research money coming from companies with less than 1,000 employees.
Microsoft was held up as a leader among big companies, even though it's only in recent years that the software powerhouse has really stepped up its licensing effort, both to bring technology in and to serve up its unused know-how to other companies.
Microsoft announced in December 2003, that it would. The company began focusing attention on technology . A month later it said it would also look to that want to build a business around its technology. This week, the company announced the latest such deal, to Eon Reality.
It's been a challenging shift, said David Kaefer, one of the lawyers who leads Microsoft's intellectual-property licensing effort.
"In any large company there is some skepticism about opening up your stuff to other people," Kaefer said. "What I've observed, over time, is a greater willingness to go experiment."
Chesbrough pointed out that the software industry is one that's heavily affected by the shift, but he said all manner of fields are touched, noting that Kraft Foods has a senior vice president of open innovation. There are some disciplines that are less likely to see the product development process open up.
"The nuclear power industry is not going to open innovation anytime soon," he said.
But in the software industry, he said, all kinds of new business model possibilities will emerge once intellectual property is seen as something that can freely move back and forth.
Companies have emerged and will continue to spring up to buy technology that is not being fully used.
"You can start to imagine bounty hunters who go out and look for this stuff," Chesbrough said.
Businesses could decide to sell even those technologies they are using, leasing back the right to use it in their industry while allowing another company to put the technology to work in some other application.
Indeed, the idea of a secondary market for software patents is one that's gotten a great deal of attention lately.
But panelists cautioned that the market is not yet here.
"Today there isn't really a secondary market," Huston said. There are indications that one is emerging. The challenge is that patents are hard to value and it takes a lot of work to evaluate whether a particular one is any good.
"Intellectual property is not like pork bellies," Huston said. "Valuing a patent and understanding what it is, is devilishly hard."
Companies also face a number of pressures that make it easier to keep even unused patents, rather than look for alternatives.
One problem is that because patents are still legal weapons, you never know when you might need one as a negotiating tool or other such implement. Huston pointed to a time when he was at Intel and the company was looking to donate or let lapse a number of underutilized patents. There was one that was close to being given away, but a lawyer noticed that it could be critical to boosting Intel's bargaining position in a pending cross-license deal.
"It probably saved Intel $100 million," Huston said. "This was a patent that was within weeks of being given away."
Another issue is the "winner's curse," in which executives stand to be criticized for licensing out a technology that ultimately makes another company millions--criticized more, in fact, than they would be for having let the technology languish.
Many say the benefits of open innovation could be hampered if companies fail to recognize that there's more to innovation than filing a patent.
Huston said that without proving a technology can work in the marketplace, it's an invention, not an innovation.
"That's why most patents are worthless," Huston said. "They don't do anything anyone would pay for."
Often singled out for criticism is Intellectual Ventures, the Nathan Myrhvold-led company that has been scooping up patent portfolios and says it will build a business licensing out the technology.
The potential benefit, Chesbrough said, could come by companies like Intellectual Ventures increasing the value of independent invention. That said, he acknowledged that what one person sees as an increased value for innovation "is another person's extortion."
Though there's a chance that companies could try to wait for other companies to develop an infringing technology and then seek royalties, Chesbrough said that would take a lot longer and run the risk of spurring adverse legislation.
The better approach, he said, is to license technology early on and seek only a reasonable cut. If they seek too much money, it won't be profitable to use their stuff. "That will set a practical limit on how much they extort, or charge."
Both Huston and Chesbrough point out, though, that even under the best circumstances, a company devoted purely to intellectual-property licensing is a tough proposition. Those who believe the future of such companies is rosy often point to Qualcomm, which today gets most of its money from licensing its cell phone network technology.
"Today they've got this lean, mean IP machine," Chesbrough said, adding that its current business stemmed from a huge investment of several hundred million dollars to demonstrate its technology. Key in that, he said, was straightening out the kinks. Only by actually building out its technology, Chesbrough notes, did Qualcomm realize it needed to vary its signal strength, boosting the signal in some areas and weakening it in others. That kind of knowledge can't come from a company that doesn't build anything.
He said the Intellectual Ventures business model doesn't appear to cover that. "They are not going to accumulate this knowledge," he said. "Somebody is going to have to do that. That's something I'm not sure those folks have fully thought through."