Microsoft lab has its sights set on future

Prof Roger Needham, head of the British research laboratory, is full of enthusiasm for the work being done at his lab, writes…

Prof Roger Needham, head of the British research laboratory, is full of enthusiasm for the work being done at his lab, writes Karlin Lillington

At Microsoft's Cambridge research laboratory, they're not interested in where you want to go today. They prefer to think about where you might want to go tomorrow.

"We're not working on this year's products. We're not working on next year's products. We're working on the technology that might help improve the technologies of the future," says Prof Roger Needham, a former Cambridge student who became a professor, then vice-chancellor at his Alma Mater.

He now heads up the Microsoft lab in Cambridge, running it in a low-key, quirkily individualistic way that is full of enthusiasm but quite removed from the relentlessly perky optimism of the American lab at Microsoft's headquarters in Redmond, Washington.

READ MORE

Along the way, Prof Needham, an expert in computer security and encryption, has built up a curriculum vitae that includes some of the best-known research labs in Silicon Valley such as Xerox's Palo Alto Research Centre, where the modern computer was by and large created, and the former Digital's Palo Alto laboratory.

While working as a researcher, Prof Needham got the call from the Washington software giant to run a new, English lab.

After "about a millisecond" of thought, says Prof Needham, the Cambridge lab had its head.

Nearly five years later - and preparing to officially celebrate its fifth anniversary - the lab is eagerly pursuing work in such topical areas as peer-to-peer (P2P) computing, self-organising networks, 3D holograms that tilt and roll on handheld devices, and picture-editing tools based on probability theory.

The Cambridge region is already well-known as a technology hotbed that has produced several successful companies, so the university was an obvious choice for the lab, says Prof Needham.

And Cambridge's broader science and technology credentials are formidably impressive, including heavyweights such as Sir Isaac Newton, "father of computing" Sir Charles Babbage, DNA pioneers Crick and Watson, mathematician and code-breaker Alan Turing, and Prof Stephen Hawking.

Despite a reputation for speaking his mind - which reportedly has kept the folks in Redmond slightly nervous - Prof Needham has managed the lab without the controversy and conflict many predicted at its inception.

Many wondered how the big corporate lab would fit into a fairly eccentric and free-thinking British university environment, and how the independent Prof Needham would fare as a deputy to Bill Gates.

But it has managed fairly well all around, it would seem. Since its set up in 1997, the Cambridge lab has grown from 13 researchers squashed in temporary quarters, using computers bought with assistant director Dr Chris Bishop's own credit card, to 65 researchers in a specially built facility in West Cambridge.

If the researchers tend to spend a lot of time emphasising to visitors how much they enjoy working for Microsoft, to be fair, they do get asked regularly whether they do.

However, a few will admit on the sly that having eight time zones between the US west coast and Cambridge offers a pleasant sense of distance and freedom from corporate constraints.

The lab also has an overwhelmingly European feel. Prof Needham says that Microsoft's head of research, Dr Rick Rashid, told him not to forget that the lab was "a European enterprise, not a British one". Now, a large Gallic contingent means French is as likely to be spoken in the halls as English, along with dollops of Spanish, Italian, Portuguese and other European tongues.

The research focus at Cambridge has not developed around a deliberate agenda, Prof Needham says, but incorporates the interests of its incoming researchers.

At the moment, those include machine learning, distributed computing, computer vision, mobile computing and computer languages.

According to Dr Bishop: "Probability theory influences as much as half of all research here." That translates into projects such as one called Pastry (researcher Dr Andrew Herbert declined to explain the odd origin of the name), an algorithm (mathematical formula) that Herbert believes will help improve the way in which P2P networks manage and distribute content.

Rather than a P2P network having to ask all computers on its network if they have a given piece of information, Pastry relies on clusters of computers routing queries only to the most relevant machines, improving traffic efficiency by 60 per cent.

"It's a very efficient algorithm - a funnel that comes in very, very quickly," Dr Herbert says. He thinks Pastry could help very large networks to be self-managing, taking the tedium out of system administration.

He adds, "It's something on the trajectory of where we see the .NET strategy going." .NET is Microsoft's expansive - and controversial - plan for a network of internet services based around Microsoft products. However, Pastry will adhere to open standards and not be a proprietary technology, he said.

Another project incorporating probability theory teaches computers how to "see" a given object against a background. Such abilities could help a computer to accurately outline an object in a photo, cut and paste it, and refill the gap with a correctly synthesised background - a boon to users of photo-editing software.

But in the future, according to researcher Dr Andrew Blake, such innovations could help teach computers to really see, recognising individuals against complex background environments.

Researcher Dr Lyndsay Williams' Holosim project looks at ways of placing 3D objects on handheld devices equipped with tilt sensors. "By tilting, you can look around any part of a 3D image," she says.

Tilting removes the need for a mouse, a button or a pen for scrolling and is a more natural way to examine an object in the hand, she says. The same concept would also allow someone to view a flat image, such as a document.

SmartView, another handheld development, is a way of segmenting large Web pages into smaller, logical units so they can be viewed on handheld computers or PDAs. The project of researcher Mr Ralph Sommerer, SmartView analyses Web pages by tables and cells, not by text, and clusters related elements together into smaller, easily viewed chunks, he says.

SmartView has a fairly obvious product application - it may well go into future versions of handheld operating system Windows CE. Other projects, such as Holosim, still seem more Star Trek than Compustore.

But the lab has applied for almost 50 patents, and researchers have published 93 papers and parts of five books.

Yet even at Microsoft, not everything needs to end in a product.

"Some things we do just because it's good science to investigate," says Dr Blake. "We don't really know how it's going to be used."