It was on November 12, 1990, in a speech at the Comdex/Fall show in Las Vegas, that Microsoft chairman Bill Gates first proclaimed his vision for "information at your fingertips" — software that would let people easily find the data they wanted, wherever it was on their computer or office network. The new system, he promised, would arrive within three years.

Those three years passed, of course. Then another 10. Today, Microsoft is still promising information at our fingertips, now in the guise of powerful search technology built into the next generation of Microsoft's operating system, code-name Longhorn. But Longhorn, originally expected this year, remains just beyond our grasp: It now isn't likely to reach customers until 2006.

Meantime, an upstart named Google, founded eight years after Gates announced his quest, has pretty much stolen the day. Google's string of Internet search innovations has not only won the hearts and minds of customers but also made a ton of money. It has spent a total of $233 million on research and development since 1998 — just 3.4% of Microsoft's annual R&D budget — yet its market value now tops $3.7 billion.

Which raises the question: Why not Microsoft? Here is a giant company that for two decades has dominated computer software. It generates $9 billion in cash flow and $8 billion in profits from $37 billion in sales, making it twice as operationally efficient as General Electric and more than twice as profitable as IBM. But when it comes to online search, arguably the hottest technology of the past five years, Microsoft has missed the boat. Heck, it hasn't even been near the dock.

Say the same for the Web browser (created by Netscape), the streaming media player (by RealNetworks), the game box (Sony), interactive television (TiVo), so-called smart phones (Nokia, Ericsson, and Motorola), and digital-music distribution (Napster and now Apple). Once the bellwether of the computing industry, Microsoft has watched from the sidelines as comparatively smaller, poorer companies brought to market virtually every important technical innovation of the past decade.

It's not the sort of track record that inspires confidence about Microsoft's prospects. "It's not good enough. And it's not just the incremental product innovations that matter. Microsoft's inability to create leadership in entirely new product areas . . . is a real problem," says Adrian Slywotzky, strategy guru at Mercer Consulting. "If they didn't have $61 billion in cash, if they had only $40 billion but they had dramatically stronger strategic positions in games, or [digital] music, or technology in the home — which is where they really want to be — then people would think very differently about this company."

Instead, Microsoft shares have gone nowhere in the past five years, more or less tracking the market in that time. In July, chief executive Steve Ballmer basically conceded that his company couldn't find enough opportunities in which to invest its enormous cash hoard, announcing plans to pay shareholders a $32 billion special dividend. And Wall Street analysts have quietly begun comparing Microsoft to a power utility circa 1990: well suited to generate steady cash flows and dividends, but not growth.

Naturally, Gates and Ballmer fiercely deny that their company is either past its prime or stuck in midlife, or even that it is a "fast follower" when it comes to innovation. "If you want to go with that myth, that's fine," Gates scoffed in an interview. "Were we a fast follower when we invented the PC idea?"

Well gosh, Bill, how to break the news? Yes, you were: Microsoft didn't invent the PC itself, or even the operating system that made PCs tick (it famously bought that from a third party). Its innovation — call it genius — was divorcing the operating system from hardware. It was this new business model that fueled the mass proliferation of personal computing in the 1980s and 1990s.

That distinction says something about Microsoft's dearth of innovation since: Redmond, Washington, has never been a great cradle of invention. In fact, Microsoft's research labs, though enormous and enormously well funded, are also strikingly inefficient. Beyond missing big opportunities, the company has placed some very poorly conceived bets (digital toilets — no kidding — come to mind) and poured hundreds of millions of dollars into technologies with neither sizzle nor apparent commercial prospects (SPOT Watch, anyone?).

But Microsoft's own missteps explain only so much. The truth is, the failure to consistently produce dramatic and successful innovations may be less a comment on Microsoft than it is on the nature of innovation itself. Innovation is, after all, capricious — a function of luck and good timing as much as brains. It's tough to score once, much less repeatedly, with big-money bets and sky-high ambitions. Harder still to feed and mine creativity in established organizations, where scale becomes the enemy. It's simply more difficult for a $37 billion business to find and commercialize inventions that will sustain profitable growth at the same rate as in smaller rivals.

More than that, Microsoft may be the most striking example ever of the phenomenon that Harvard academic Clayton Christensen famously identified in his 1997 book, The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail (Harvard Business School Press). Good managers, Christensen wrote, tend to direct resources toward protecting established lines of business, usually by investing in incremental improvements that help pad profit margins.

Christensen called these "sustaining innovations." We call it playing defense. It's not that Microsoft lacks creative talent or that it simply has run out of ideas. In fact, the company has an abundance of both. But for most of the last quarter century, it has overwhelmingly devoted these vast resources to the innovative defense of its existing franchises. That's why it has missed opportunity after opportunity to launch important new businesses. And why, in all likelihood, it will miss many, many more.

The Department of Defense

Imagine working at the biggest, richest technology research-and-development operation on the planet. In its fiscal year ended in June, Microsoft spent $6.8 billion on research and development. For perspective, that is roughly equal to the annual R&D budgets of Oracle, Hewlett-Packard, Dell, Apple, and Sun Microsystems combined. Only IBM, with a budget of $5 billion, comes even close.

Microsoft's talent base is staggering, too. Most of its 57,000 employees are engineers, but fully 700 of them work exclusively for Microsoft Research (MSR), the company's quasi-academic think tank that is two-and-a-half times the size of Xerox Corp.'s famed Palo Alto Research Center in its 1970s heyday. With an annual budget of $250 million, MSR can afford to hire the best and brightest: Gary Starkweather, inventor of the laser printer at XeroxPARC; Jim Gray, a seminal developer of the distributed database; and Gordon Bell, known as the "father of the minicomputer" for his work with miniprocessors at Digital Equipment Corp. in the 1960s. Twelve of MSR's researchers have been inducted into the National Academy of Engineering. Two have won the Draper Prize; three more have won the A.M. Turing Award for computer science.

In the past five years, Microsoft has acquired 2,188 patents to protect its researchers' work — and the quality of those patents is outstanding. Patent Ratings LLC evaluates patents on three criteria: the number of unique technology claims, probable revenue and profits, and the likelihood that they can be defended. Over the past five years, Microsoft's average patent "intellectual property quotient," or IPQ, was 123, well above average. Because software patents are harder to acquire — and consequently harder to defend — than hardware patents, it's especially impressive that Microsoft's average IPQ is higher than those of hardware makers IBM and Hewlett-Packard — the two most prolific patent claimants in technology.

Moreover, Patent Ratings estimates that of those 2,188 patents, 75% are original, as opposed to continuing patents. This means that three-quarters of them were issued to protect entirely new technologies, not, as with continuing patents, to renew claims on old inventions. In the past decade, Microsoft was issued proportionally more original patents than either Intel (71%) or Apple (68%).

But here's the thing: For all of its cash, talent, and intellectual property, for all of its apparent research productivity, Microsoft isn't really set up to invent new stuff. Microsoft's co-chief technology officer, Craig Mundie, volunteers a remarkable statistic: Of every dollar the company spends on R&D, "probably something on the order of 90% is directly in line, or in service of, the existing business groups." Just 10%, he says, "is essentially [invested in] pure research or incubation [of new products]." So yes, Microsoft files many original patents, but if Mundie's assessment is on target, most of those probably represent merely innovative features for existing products.

"We have this interesting economic model, which is we sell a product that never wears out," Bill Gates argues. "So if I don't innovate in Office, I get zero revenue."

There is, of course, a compelling economic rationale for this defensive strategy, as Gates himself takes pains to explain. "We have this interesting economic model, which is we sell a product that never wears out. If you [have] a version of Office, I can't come to you five years later and say, 'Okay, it's worn out and you need a new one.' I have to come to you with innovation that makes it worth it for you [to] move up to [a new version of] the software. So if I don't innovate in Office, no matter what my market share is, I get zero revenue. If I don't innovate in Windows — you already have a copy, it is right there running on your machine — I get no revenue."

That's only partly true: Microsoft collects revenue for every copy of Windows, innovative or not, delivered with a new PC. But let Gates continue: "So, we wake up every day and say, 'Okay, what are the breakthroughs there [in Windows and Office]?' Workflow, business intelligence, integration with the phone. And [we] think, How do we drive those things forward?"

That's why, with its release of Office 2003, Microsoft added a new information-rights management feature — pioneered by others — that allows users to give sensitive documents expiration dates, after which no one can view or share them. It also created a "research and reference task pane," giving workers access to dictionaries and online search sites without the hassle of having to leave the programs in which they are working to, say, open a Web-browser window. (For either feature to work, customers must buy a $400 license to Windows Server 2003 software. That's two new sources of revenue for Microsoft!)

On one level, it's hard to fault Microsoft for pursuing a strategy that protects businesses that have served it so well. After all, Windows and Office provide close to 60% of its revenue and nearly all of its operating income. But Slywotzky and other outsiders believe Microsoft is spending more than it has to on defensive R&D. By Mundie's estimate, the company invested $6.1 billion last year in research to support existing franchises. But those categories produced revenue growth of only $2.7 billion. Wouldn't it be better off pursuing bolder new bets? "Microsoft has been too timid," says Howard Anderson, founder of YankeeTek Ventures and a lecturer at MIT's Sloan School of Management. "Are they spending enough? Clearly. But they either come to market late, after others have proved the value of a new product, or they [invest] in little fixes to products that they already have on the field."

To be fair, Microsoft has spent lavishly over the years on dozens of new opportunities. It just hasn't fared very well. Take the company's decadelong investment in interactive television. Gates really was in the vanguard when, in a 1992 speech, he predicted that television would one day be delivered over Internet protocol and enhanced with "rich software."

"I said, 'Hey it could happen any time.' Well, you know, about eight years went by and nothing happened," Gates admits with a chuckle. "[But] Microsoft was [still] spending over $100 million a year on research and development on that." In 1997, after investing half a billion dollars in its own technologies (none of which worked), Microsoft shelled out $425 million to buy the startup WebTV Networks. It made more investments in television through its satellite venture Ultimate TV, as well as in cable through AT&T (now Comcast). Together, these stakes cost Microsoft over $5 billion more. Both WebTV and Ultimate TV foundered and have since been folded into MSN.

Just Ahead of its Time

This is the central tension for Gates, and for Microsoft: Given a strict choice, it will nearly always be easier, and probably more fruitful, to plow research funding back into a proven technology. Sinking money into wisps of new ideas can be interpreted as bold and forward-looking — but also, for a market leader, as strategically irrational and financially untenable. "Internally," Gates admits, again chuckling, "people give me no end of grief about things like TV."

Gates straddles this crucial fault line with apparent ambivalence. He makes a compelling case for his company's predominantly defensive research strategy. But when pressed, he proudly ticks off long-term product investments, where, he claims, Microsoft was either first or at least "resilient," meaning the time horizons were just as long as with interactive television — and the bills nearly as big.

"We've invested 10 years in the Tablet PC" — Microsoft's platform for pen-computing, which has won only modest adoption. "We were in there even before the fad came along and after the fad went away. . . . We've been working on speech recognition for a long, long, long, long time!" The Speech Server project began around 1995. It too has cost close to $100 million a year. "The SPOT Watch, we've been working on that for over 10 years." The SPOT Watch (for smart personal objects technology), is a tepid PDA-like wristwatch released last January.

While missing the biggest computing trends, Microsoft has sunk fortunes into technologies that proved either way ahead of their time or, simply, wrong.

This is par for the course at MSR. While missing the biggest computing trends, Microsoft has sunk fortunes into technologies that proved either way ahead of their time or, simply, wrong. Or, just as troubling, into innovations that most considered inconsequential. Even Lawrence Karr, whose decades-old patents for a method of transmitting data via low-grade FM radio are what make the SPOT Watch work, still seems dumbfounded by the attention Microsoft has lavished on his technology. "I never thought anyone would want this stuff," Karr says. "I never thought I'd be working on something this mundane."

And what of its megastar researchers? They aren't toiling on the latest Longhorn bell or whistle — but neither are they pursuing anything with apparent immediate commercial potential. Jim Gray is developing the "World Wide Telescope," an Internet database that, Microsoft says, will eventually host "the entire world's astronomy data." Gary Starkweather is refining his design for a circular, "super panovision" computer monitor. Gordon Bell has spent his time recently on MyLifeBits, an online database aggregating all of his personal information, including 20,000 documents, 60,000 emails, 15,000 photos, and all of his music and videos.

Gates says he's content with that approach. "The market will tell us if we're ahead of our time," he says. "That's okay." But is it really? Experts in innovation argue that Microsoft just isn't investing enough in the sort of offensive innovation that should define its future. And the research it does do seems wildly inefficient. Over the past five years, Microsoft has spent an average of $9 million per patent, nearly twice the average for its software peer group. (In July, the company told analysts it would aim to double the number of patents it seeks. But in light of Microsoft's campaign to reconcile intellectual property disputes in the European Union and with competitors such as Sun Microsystems, this appears to reflect legal strategy as much as a renewed zeal to improve the research effort itself.)

Of course, it's impossible to say exactly what the right balance is between offense and defense, or between the short and long term. Ultimately, observes Gary Hamel, innovation guru and chairman of Strategos, "really good ideas are just few and far between. Forgive the metaphor, but it is a little like the process of sperm trying to fertilize an egg. Increasing the number of really strong swimmers doesn't increase your success rate. That's not how biology works. Only one [sperm] gets to fertilize the egg."

In other words, innovation is often just a happy, creative accident, where the laws of numbers can actually get in the way. In this sense, Microsoft's size and wealth become obstacles rather than assets. Venture capitalists typically pore over 50 to 100 deals to find a good $20 million software investment. By that logic, Microsoft, with its $6.8 billion annual R&D budget, must consider as many as 35,000 new ideas just to find a few hundred worth investing in every year. Is it any wonder the company invests in so many dogs — or, for that matter, toilets?

No surprise that the parking lots of the Courtyard, Fairfield, and Residence Inns opposite Microsoft's headquarters are always full. The rooms there are choked with engineers who have been summoned by Microsoft on the basis of particular knowledge or expertise that may someday turn into a commercially viable innovation. Some will sign on as consultants and a few others as employees. But the motels are always booked: Microsoft is that starved for ideas.

Even then, there just might not be enough legitimate opportunities out there for Microsoft to employ its vast resources effectively. "There are only about 10 software companies in the world with more than $1 billion in sales," observes Brian Skiba, managing director of the San Francisco-based hedge fund Viant Group. In the software industry, a research project can be considered successful if it spurs annual revenue growth of 10% to 15%. By this metric, "Microsoft has to create the third- or fourth-largest software company in the world every year to be considered innovative," says Skiba. Even if Microsoft had been able to replicate Google's dominance in search technology, Google's $1.5 billion in revenue would have lifted Microsoft's own top line by just 4%.

That's why it serves Microsoft to avoid the connection between research and revenue growth. "I think you're too focused on the top line as opposed to the bottom line," CTO Mundie said, when we put this calculus to him. Mundie argues that successful R&D is judged ultimately by profit growth. How much? Mundie was vague. "It's the maximum that we can achieve." Later, he added, "We don't measure these things."

This is the luxury of entrenchment. So long as Microsoft enjoys its dual annuities from Windows and Office, it needn't worry much about the revenue (or lack thereof) generated by its long-term research investments. It simply plows money into ideas good and bad, then sits back and waits — even decades — until the market is prepared to validate these investments. "Because we've got Office and Windows," Gates says, "we can afford to take some risk in these new areas and make sure we're out in front. The real sin for us is if we miss something."

But the reality is, Microsoft has already missed plenty. And that's the downside of entrenchment: Even as the company throws resources at the future, its strategic gaze rests squarely on the past. As Microsoft forfeits future revenue growth for current income, it continues to cede genuine innovations and important new markets to future upstarts with bigger ideas — and far less to lose.

Carleen Hawn is a writer based in Los Angeles.