Cambridge has done very well out of Microsoft in the past five years, with one notable landmark being yesterday's official opening of the William Gates Building that now houses the Cambridge University Computer Lab. Behind that is another new building with much the same design: Microsoft Research's Cambridge computer lab. By hooking into the research and innovation at these labs, Microsoft hopes to do very well out of Cambridge.
The man at the fulcrum of this two-way exchange is Professor Roger Needham, who built the Cambridge computer department's global reputation, and was founding managing director of Microsoft Research's Cambridge operation in 1997.
A Salon headline writer wittily called him "Bill's Don", but that notion would not survive more than a few minutes' experience of the real thing. If Needham could be bought, then he wouldn't be much use. He was the main reason Microsoft chose Cambridge rather than somewhere else in Europe. Fellow academic Rick Rashid, who founded Microsoft Research in Seattle in 1991, says: "The opportunity to have Roger involved is what really brought us to Cambridge."
As head of the computer lab for 15 years, Needham had earned a reputation for plain speaking. And "one of the things I've noticed, having moved from being an academic to doing this job, is that I'm trusted now, and I wasn't then," he says.
So far, Needham, now 67, has recruited about 60 researchers from across Europe. If there has been any negative fallout from the US government's anti-trust case, "we've not noticed it hardly at all," he says. "We're respected because of who the people are, and they would be respected if they worked for IBM or the mafia. The standing of our people is so high, in the computer science world, that this completely overtrumps anything people might feel about the corporation as a whole."
Not that MSRC has much to do with the software factory in Redmond, Seattle. Needham says they try not to duplicate things being done at other Microsoft Research labs in California and China, but he decides who to hire and what to research, and in general, the results are published.
"When people have been with us for a couple of months, we send them to Redmond," he says, so they can get to know other researchers and people from product groups. "We are not working on next year's products, but we do know an awful lot, and sometimes it turns out we can be helpful. For example, we have had people putting generic functions into C# [the open computer language used in Microsoft.Net]. It's not research and I would not want them doing it for very long, but it was helpful."
Research rarely makes a visible impact on products, with ClearType - developed in Redmond - being one obvious exception. "Small pieces of technology typically get transferred, rather than large advances," says Needham. "You would have to do something rather special to divert the juggernaut, but you can make things better."
One example is the Cambridge work on image objects led by Professor Andrew Blake. He is working on improved filters that, for example, let users draw round and cut out part of an image then fill in the background. His demonstration involves removing one girlfriend from a tourist scene and replacing her with a different girlfriend. The aim, says Blake, is to develop something that is "smart on the inside, simple on the outside". A huge amount of complex maths is involved in getting a program to separate, say, a hairy animal from background vegetation. But for the user, it just delivers a slightly better result from one pull-down menu option in a paint program.
Some research requires less complex maths but could deliver obvious benefits. For example, Natasa Milic-Frayling is working on different ways to manage browsing. History Explorer, one of her additions to Internet Explorer, lets you see where you have been by keeping a series of thumbnail pictures of pages linked together. She can and probably has demonstrated the idea to the product teams, but there is no guarantee it will be adopted.
MSRC's importance has increased in the past two weeks, because of the expected closure of AT&T's Cambridge research lab. This was founded in 1986 by Olivetti (which had rescued Acorn, a local computer company) and Professor Andy Hopper from Cambridge University. It became the Olivetti & Oracle Research Laboratory until 1999, when AT&T - a company broken up by the US Justice Department - came in with a six-year investment.
"It's a tragedy that AT&T is closing them down," says Needham. "My understanding is that, over the years, they have been a profit centre, not a cost centre, because of the start-ups they have done. I am surprised and somewhat dismayed that no one's picked them up."
AT&T's withdrawal makes it hard to complain about Microsoft funding research in Cambridge when it would otherwise be done somewhere else in Europe, or in Seattle or California. But for anyone with around £6m a year to spare, there are still opportunities available.
