Inside Microsoft Research Cambridge

Microsoft recently held an event at Microsoft Research Cambridge, inviting thought leaders and innovators from business and bringing them together with our own 100-strong team of researchers. Filling the labs auditorium and seminar rooms, it was in itself, an experiment in creative cross-pollination.

Founded in 1997, the lab’s mission focuses on leading-edge research that advances the state of the art in computer science so that it can develop technology to transfer to the Microsoft business and help lead Microsoft into the future. But it’s not a product-development facility; as you’d expect from a lab based in Cambridge, it works closely with academia.

The role of innovation

According to Clay Christensen, the author of The Innovator’s Dilemma, large companies like Microsoft run the risk of being out-innovated by new market entrants who are not focused on optimising existing products. Instead, start-ups can focus on breakthrough technology that disrupts whole markets.

The pace of change is accelerating. As Rob Fraser, CTO of Cloud Services remarked, “S&P 500 companies have a half-life of seven years.” In other words, half of them won’t be in the list by the end of the decade, replaced by more successful companies.

Research and development labs like Microsoft Research Cambridge are the natural response. They turn a big company’s resources into breakthrough innovations such as the Kinect motion sensing input device and the advert prediction algorithm used in Bing, both products of the Cambridge lab.

Academic links are very valuable. Consider the central role of Stanford University in Silicon Valley or MIT on the East coast. In fact, MIT as a whole “has generated nearly half as many entrepreneurial spin-offs as the entire UK university sector,” according to a 1998 report by McKinsey. Cambridge is establishing itself as a similar hub for innovative companies.

Pure research, serendipity and competitive advantage

The body part recognition software in Kinect evolved out of research in a completely different field and it’s a great example of the way that pure academic research can filter into the commercial world. Cross-fertilisation with academia has direct benefits for Microsoft.

According to lab director Andrew Blake, Microsoft Research has influenced virtually every product that Microsoft has shipped and some have gone on to be huge successes. For example, Kinect has gone on to sell 18m units.

Another example of profitable serendipity are the algorithms developed by Microsoft to analyse big data. They were tested on historical chess match results and the by-product of the research was eight lines of magic code that drives Xbox Live’s player-matching technology. This tiny code fragment (and the deep thought behind it) is the secret sauce in Xbox Live. It makes the difference between a challenging game against a similarly-skilled opponent or an unsatisfying defeat or walkover.

It may seem counter-intuitive when everyone is trying to cut costs but in a recession, investing in R&D can be the smart move. A McKinsey study of the 1990-91 recession found that companies that the companies which invested in acquisition, R&D and advertising during the tough times emerged ahead of their competitors who cut back.

Four technologies to watch

At the event, researchers gave in-depth presentations about different projects at the lab. They are interesting in their own right but they are also prophets for future changes in IT.

  • Big data and analysis. We live in an information society. The big question now is “how do we aggregate, analyse and present information,” according to Kenji Takeda at the lab. This requires tools for working with large data sets, such as the Intellisense for Data tool that he is working on. It can pull information out of the World Bank’s 1.5m data sets with a single line of human-readable code. Much of the value of data lies in how it is used so the lab is investing effort into economics research. For example, what is the optimum pricing model for cloud infrastructure (if you want to create a secondary market between buyers), insurance (if you know how well or badly people drive in real time) or electricity (in a smart grid)? Machine learning is also evolving and John Bronskill is working on tools that will create an optimised model of a problem to make it easier to apply this technology to existing data sets.
  • Biological computing. It may sound Strangelovian but we’re rapidly approaching the era of biological computing where living cells can be used to run software. The lab is developing tools to model molecular mechanisms underlying human cancer along with software for programming molecular devices and compile software to DNA code. These are the first steps towards understanding what cells compute, how and why and it’s an extraordinary vision of the future.
  • Natural user interfaces. New trends are driving evolution in user interfaces and the traditional windows, icon and mouse pointer paradigm is making room for touch interfaces and, soon, direct physical and gestural interfaces (think Tom Cruise in Minority Report) and, eventually, augmented reality. We saw demos of Microsoft technology such as Kinect Fusion which creates 3D models of the real world, Holodesk which combines input and output in 3D and tools to convert a series of 2D images into an articulated 3D model.
  • Beyond the file system. Digital files, such as photos, are more than a collection of bits or a digital representation of a real world objects. As we move from the traditional desktop metaphor to a metaphor cloud-based, social media world, we need new ways of storing, searching and representing our data. Richard Harper and Richard Banks are working on new file systems that make help users do a better job of managing their files in this new world.

These talks were backed up by hands-on demos. Out of dozens, four in particulate stood out as examples of ways this new technology could impact real products.

  • The A-Brain Project. This uses ‘big data’ techniques to analyse neuroimaging and genetic information using Azure and MapReduce technology.
  • Project Colletta. This add-on for Windows pulls related documents and information together into a single collection that is easily accessible from the desktop, making it easier to manage your tasks, activities and related resources.
  • Digits. This wrist-worn sensor captures the full 3D pose of the user’s hand making it possible to control a user interface with gestures but without the need to wear a special glove.
  • Bio Model Analyzer. This is an early prototype of the software needed to actually program living cells. It lets users draw out a biological system by dragging and dropping cells, their contents (e.g. DNA) etc. on to a canvas and then analyse them and determine cellular stabilisation.

These highlights and common threads were only a subset of what we saw in November but they show the breadth and focus of Microsoft’s research. They also point to big innovations in IT that will soon affect businesses and consumers.

Next steps

More information:

To talk to the MS Horizons team, email:

To find out more about how Microsoft technology is supporting innovation in the Enterprise

By Tim Cozze-Young
Microsoft Enterprise Team