Figuring out how to blow $5 billion (€5.5 billion) in a year is a challenge - even for one of the world's richest men.

Founder and chairman of Microsoft Mr Bill Gates admitted as much last week at Microsoft's Washington headquarters, when he spoke of how the software company spends its close-to-$5 billion annual research budget.

"I never thought I'd get up and be proud of spending $5 billion," joked Mr Gates, dressed in classic American "business casual" style of open-necked blue shirt and dark slacks.

His speech, given at a "Future Forum" marking the tenth anniversary of Microsoft Research (MSR), profiled the company's research programmes.

But it was equally an opportunity for Mr Gates to evangelise, and to subtly address some of the criticisms made of the company, still embroiled in an antitrust suit with the US government.

The primary message from Mr Gates's speech was that the company, confident of its own longevity, is not for turning.

A key emphasis was on the importance of long-term research and the implication was that Microsoft would be around to see its fruition: "The problems we're tackling are not problems that lend themselves to overnight success. Most of the goals we've set out on will probably take most of the next decade to resolve themselves," he said.

He also spoke of Microsoft's commitment to distributed computing, which uses the capabilities of many computers linked across a network to tackle problems or provide new services.

That notion is the basis of Microsoft's new - and controversial - broad-ranging computing strategy, called .NET.

Critics worry that Microsoft could become a dominant force on the internet, presenting Microsoft's products and services as a person's main link into the Web.

Mr Gates said the company would continue to add features to its software, arguing that people often didn't know what they wanted in a program until it became a standard offering.

Microsoft is often accused of creating "bloatware" - programs containing dozens of features that computer-users rarely or never use, but which increase the size of applications like Word or Excel.

Microsoft is also eager to discuss its advancements in what Mr Gates called a "holy grail" of computing - the ability to read a computer screen as comfortably as paper.

One of the most dramatic presentations by researchers showed a prototype "Web tablet", a full-strength, 2.5lb hardback-sized computer on which people can write with a specialised pen.

Software converts handwriting into searchable notes and boosts screen readability to almost magazine print quality. Several manufacturers are expected to offer tablets next year.

While Mr Gates painted a broad picture of the company's overall research goals, other Microsoft researchers filled in the detail.

Dr Rick Rashid, senior vice-president of MSR, said the research division was set up in 1991. "Microsoft was an awfully small company back then," he said, although Microsoft's concept of "small" is certainly a relative one - it had $1 billion in sales then, and 5,000 employees.

Key products at the time included operating systems DOS and Windows 3.0, and three youthful productivity applications: Word, Excel and Powerpoint.

Dr Rashid joined MSR from Carnegie Mellon University, where he says he earned the name Dr No from Microsoft for refusing the company's offer to head up MSR for many months. Now, the division he runs boasts 600 staff working in 40 research areas and has labs in Redmond, San Francisco, Cambridge, Beijing, and most recently, Mountain View, California.

"When I came to Microsoft, my idea was not to create a traditional research section, but something more like a university research department," he said.

"Our first goal in research is to expand the state of the art. If we're not pushing, expanding, getting out in journals, then we're not expanding the state of the art.

"Our next goal is to take that research and move it rapidly into products." The move to get research into products is central to the company's survival, he said: "Our goal is to continue to ensure that there's going to be a Microsoft."

The directors of the California, Cambridge and Beijing labs presented work done at each facility. Perhaps most intriguing is the Chinese lab, because it is tackling problems of which the West is barely aware.

For example, while voice recognition software remains a minor interest in Europe and the US, it turns out to be an enormous boon to Asian languages.

Keyboards cannot adequately accommodate the thousands of characters in languages such as Chinese, said Mr Ya-qin Zhang, managing director of MSR Beijing.

So voice recognition enables people to enter characters by voice at speeds that far exceed keyboard users.

Dr Chris Bishop, assistant director of MSR Cambridge, outlined some of the lab's work in the complex area of computer vision. Recognising the features of a human face is an "enormous" challenge, he said, demonstrating some of the ways in which researchers are tackling the problem.

"We're so used to being able to turn around and recognise someone's face in a fraction of a second that we take it for granted," he said.

"But a third of our cerebral cortex is devoted to vision because it's such a complex problem."

MSR is not just computer scientists, but includes many sociologists and psychologists who try to understand how people work alone and together, and what they would like to be able to do with computers, several researchers said.

Some of their findings have been included in Microsoft's new operating system, Windows XP, they said, although a demonstration of such features encountered the gremlins that often plague live previews of products.

Some of the more interesting developments in software capabilities shown by MSR fall into the area of natural user interfaces - finding better ways for humans to interact with computers.

Researcher Mr Kai-fu Lee said MSR was trying to get computers to learn human language, rather than the other way around.

He also showed intriguing ways in which linguists were helping the company to make search programs work more efficiently and understand better what people were actually looking for, from brief verbal cues.

Software could automatically present the best search engines from the context of a request, for example.

One researcher in this area provoked laughter when she demonstrated how such a search engine would work by looking for an antitrust lawyer.

For the hard-core tech addicts, the most compelling session of the seminar was probably a researchers' round-table, which tackled questions from the audience. However, as questions had to be handed in on cards in advance, researchers, and a cautious Microsoft, could safely circumvent any awkward grilling by press attendees.

The tough questions will undoubtedly be teased out by the lawyers in coming weeks as Microsoft returns to the courts.

And then we'll also know whether the multiplicity of research projects shown at the forum will have a future as well.