Going back just a few years, the corporate relationship between research scientists and the engineers and execs who built and sold products was entrenched, methodical, and often contentious. Top tech minds at companies like AT&T, Hewlett-Packard, IBM, and Xerox toiled in their labs for the next breakthrough, which, after sufficient gestation and a lot of luck, made its way through advanced engineering, product development, and marketing. Out of this process, which could easily take years, came landmark products such as the mainframe, the PC, laser printing, touch-tone dialing, and Unix.

Now, in trying to gain an edge in the fast-paced Internet software market, Google, Microsoft, and Yahoo are taking a wholly new approach to research. They're building labs focused on the problems and opportunities that have emerged with sleeker Web sites, the explosion of online video and photos, widespread broadband connections, and the soaring numbers of hours people spend online. Inventions can be tested on thousands of users at little cost, and adjusting an algorithm today can mean big gains in the effectiveness of a Web service tomorrow. The 20th century blueprint for research is "essentially mythical now," says Alan Eustace, Google's senior VP of engineering and research. "The model of research has changed."

Google's small research group is working on a speech-powered search engine, pages that answer queries with graphical time lines instead of lists of results, and software that can translate Web pages from Chinese to English. But the company hires "research-caliber people working on research-grade problems" for its product groups, too, and everybody shares the same offices, computers, data, and budget. "I don't separate research and product development," Eustace says. "The difference at Google is smart people are spread all over the company."

Research now reinvents the way we live every five years, says Prabhakar Raghavan, Yahoo's head of research. "You want to move really fast." Photo by Jeffery Newbury

Yahoo, too, is trying to rewrite the research formula. It's supplementing labs in Silicon Valley with groups in New York, Spain, and Chile, and it recently hired Ron Brachman, an artificial intelligence expert from the Defense Advanced Research Projects Agency, to expand its labs. Prabhakar Raghavan, Yahoo's head of research, has brought in economists, psycholo-gists, and social scientists from Harvard, Yale, and other top schools to help the company design and price its products, reflecting the Web's increasing resemblance to television. This approach lets Yahoo balance the scientific expertise and university partnerships of conventional labs with the Web's imperative for speed.

"PARC reinvented the way we all live," Raghavan says, referring to Xerox's famed Palo Alto Research Center. "But that revolution now isn't happening every 20 years, but every five. ... You want to remain scientifically plugged in and algorithmically deep, but you want to move really fast."

Microsoft, too, has plugged into the need for speedy Web development. In late January, it formed Live Labs, a small group of researchers who work on a range of projects: search algorithms that deliver results and serve ads based on a user's demographic profile; online visualization software developed by Seadragon Software, a startup Microsoft acquired in February, that can zoom in and out of high-res images for closer examination of digital maps, photo albums, and video clips; and scientific tools for grid computing and data mining.

Just as important, Live Labs aims to bridge the company's vaunted Microsoft Research group to its sometimes insular product teams, acting as a sort of brainiac job shop where engineers can quickly prototype new technology to (it's hoped) avoid missing deadlines. That's top of mind three weeks after the company botched yet another Windows Vista release date. "We can streamline the whole process of how we innovate," says Gary Flake, Live Labs leader and technical fellow, whom Microsoft hired away from Yahoo last year.

Deep thinking for the Web doesn't just involve software, either. Google, Microsoft, and Sun Microsystems each are granting $2.5 million to a new computing lab at the University of California, Berkeley, to study dependable, large-scale computer systems to serve the Web.