By Rob Knies
July 12, 2007 2:00 PM PT
On July 15-17, Microsoft Research Redmond’s External Research & Programs (ER&P) group will host the eighth annual Faculty Summit, for which more than 350 academic leaders, representing 175 institutions, will gather at the Microsoft Conference Center to discuss how computing is being used in virtually every educational discipline. The event gives computer-science and information-technology professors a forum to share ideas, create curriculum changes, discuss innovation opportunities, and meet Microsoft researchers. In the days leading to the summit, Sailesh Chutani, director of External Research & Programs, found time to discuss this year’s event and some of the challenges being addressed by academia.
Q: The theme for this year’s Faculty Summit is “optimism for the future of computing.” What makes you feel optimistic about the future?
Chutani: Part of the optimism is rooted in the fact that some of the issues that led to a lack of optimism are being understood. Some specific actions are being taken by different organizations to address those. Compared to a couple of years ago, people understand the issues.
This is not unusual as certain paradigms and ways of doing things mature. Just as you’re getting to the maturity cusp, people think all the interesting problems have been solved. But then a couple of new discoveries happen, a couple of different approaches come about, and peoples’ excitement level goes up.
People are discovering a new range of interaction between computing devices and human beings. Robotics has begun to generate excitement, because people think this would be a good way to bring life back into artificial intelligence. People are getting excited about the possibility that cellphone-based infrastructure could help solve the access-to-healthcare problem inexpensively.
Looking for hard, deep problems that have deep consequences for humanity—how can you go wrong pursuing that? You need to tap into computing tools, computing abstractions, to have a seat at the table in solving those problems. Computing hardware and software advancements have opened new research opportunities, and researchers from diverse disciplinary areas are collaborating to exploit these advances. That’s exciting.
Q: What are the goals for this year’s Faculty Summit?
Chutani: We’ll be sharing our external research agenda with attendees and outlining research funding opportunities. But part of the goal is the one we’ve had for all the Faculty Summits: to have a forum where conversations can happen that help us define the research agenda in the academic community, where we identify things we want to work on as a community. When we bring these people together, we want to look at where we should apply our collective intelligence to pursue research.
The other part that remains constant is catalyzing relationships and interactions. The things we discuss at the summit help us focus more attention as a community on those problems. What we focus our attention on varies from year to year.
This year, it’s a bit of a landmark. We have accomplished what we had set out to do 3½ years ago when we laid out a model of engagement and our framework. We have populated the framework. We have filled out all the pieces, and we have some traction in each. We have set up 10 institutes across the world. We have got one of the largest request-for-proposal programs in the IT industry. The pipeline of ideas and projects is very, very robust. We want to continue to identify problems and issues the community cares about and make sure we stay relevant and fresh and play a catalytic role.
Starting next year, we will go to a two-year cycle. We will have a summit in 2008, but it won’t be like the Faculty Summit this year. This year’s summit 2007 summit is once again focused on bringing together academic leaders for two days of dialogue about opportunities for computing research. Next year, we will gather early-stage faculty and graduate students to examine the same questions from their unique standpoint, These two events then will alternate, each on its own two-year cycle.
Q: What is External Research & Programs, and how does it fits into Microsoft Research as a whole?
Chutani: We want to expand the range of things we think about and explore by working with faculty and graduate students across the world. That’s what ER&P does. We have created an engagement model that permits us to work with a very broad range of people, a very broad range of ideas, across the world. We want to go after the best ideas, the best people, wherever they may be.
We have a self-interest in making sure the academic ecosystem thrives. Our future is determined by the quality of the people who enter the discipline. We have a policy goal to work with government entities and academic leaders to make sure the discipline retains its vibrancy and attracts the best and the brightest. If we fail at that as a community, our future would be in serious question.
Q: What programs are aimed at invigorating the field of computer science?
Chutani: One of the positive outcomes, at least in retrospect, of the dot-com bust was that the drop in enrollment that followed made the faculty stand up and think whether what they’re doing in the classroom is the best way to go about teaching the material. As enrollments dropped, they had to ask some hard questions. How do you bring back the enrollment?
That’s led to very creative experiments. It prompted a very serious intellectual exercise and deconstruction of the curriculum and then reassembly of the curriculum that kept the rigor but introduced elements that were very compelling, that made the connection between what students will do in their lives and what they learn in the class. You will hear notions like contextualizing computing. It caused people to take the context they were using in the classroom much more seriously.
Through programs such as our Institute for Personal Robots in Education at Georgia Tech, launched last year, we have been exploring whether gaming and robotics can make a difference in attracting a broader range of new students and keeping them excited about the discipline.
There are many, many such experiments happening. My guess is, one or two years out, we will begin to see consensus emerge on a couple of new modalities. There won’t be just one modality; there will be multiple contexts. The notion that you get people more excited and that learning takes place more effectively when you let them experience it by doing things, by appealing to the things that they’re interested in—I think that will become mainstream and established, and the impact of that will be quite significant.
Q: Let’s focus on some emerging research opportunities. Multicore, many-core computing: Where might this lead?
Chutani: This is a major conceptual shift that is redefining research priorities. For at least 20 years, industrial and academic research have profited from Moore’s Law, with cores doubling in capability with every generation. Further increases in computing capability will result from the rapidly increasing number of cores on a chip, as well as the opportunities of heterogeneous multiprocessing.
On the hardware level, these advances provide increased flexibility. But a lot of programming models didn’t have that level of concurrency in mind, so that necessitates a deep rethinking of tools and methods and abstractions that will enable people to build systems around multicore.
What will it mean for specific disciplines? There is a class of algorithms from artificial intelligence, machine learning, and machine vision that may be able to leverage this functionality.
Q: How about using the cellphone as a mobile-computing platform?
Chutani: I think this trend will continue. This is the only computing platform that most people in the world will have for the near future, but there are lots of roadblocks. It’s difficult to program. Relatively few APIs have been built with the view that somebody will want to use them to extend that platform. The power of the platform is manifest when you have services in the cloud to leverage. Currently, that’s not easy, either. And depending on which part of the world you live, the carriers control this tightly, so it’s not easy for people to experiment with the functionality.
But those things will get sorted out, because the value proposition is too strong. We have seen evidence of this in the work we have done in digital inclusion: Yes, this can be leveraged and used effectively as a platform.
Q: What are the specific parts of this grand challenge that the academic research community can address?
Chutani: You could start by thinking about what it means to design for human interaction when you have so little visual real estate and the keyboards are very small. A lot of our interaction paradigms take for granted that people will be sitting in front of a screen and have a keyboard. As people begin to use the cellphone as a platform, how they interact with it will create some very interesting research issues. How do you build applications for these platforms? What kind of tools do you need? What kind of reliability is required? What happens to privacy concerns?
There’s a whole set of research issues that were transferred from the PC space to this space. At some level, you’d think they’re the same problems, but they’re not. In most of the world, the cellphone is still a fairly personal device. It’s one-to-one mapping, but in some cases, it’s a community-owned device, especially in poorer economies. How do you solve some of the problems if you have a shared device? It will be interesting. People are beginning to look at being able to do microfinance using cellphones. How do you do that while preserving privacy and security?
Q: How about search? What new directions remain to be forged in that area?
Chutani: Search is an interesting space on multiple levels. What you’ve seen in the last decade or so is conventional information retrieval that analyzed text to do text-base search. How do you improve the effectiveness of search based on the semantic understanding of what you’re searching for? Can you leverage semantic approaches where if you know that what you’re searching for is something in astronomy, you target the search to leverage that knowledge? So far we haven’t leveraged the semantics of the information we are searching for and the information that we have.
Second, we haven’t really taken advantage of the context in which the search is originating, especially on mobile devices. If a cellphone user is seeking some piece of information, the search has a lot of context about the person doing the query. As an example, since phones are beginning to have GPS, location information provides underlying context to the search. This metadata enables the search to be a lot more specific.
The blocking factor in search that is preventing academics from fully engaging in the research is the shortage of large data sets. If you come up with new algorithms, how do you try them out? Currently, the only way to do that is if you work as contractors, consultants, or employees for three or four companies in the world—Microsoft, Google, Yahoo! That’s how researchers can get access to the data. Clearly, that’s not scalable. It’s important to figure out ways to make large-scale data sets available to the academic community and do it while helping protect personally identifiable information. It’s a sensitive issue. If we solve that, then you could see an explosion of new activities around search.
Q: How can computing be used to address fundamental problems in the sciences?
Chutani: We have some early indicators of what that would look like. Climate modeling and environmental modeling are heavily computing-intensive. People are using computing tools to describe and capture biodiversity. You want to bring the physical and the cyberworld together. We want to make it easy for people to go back and forth. I think that will have deep consequences on how research happens.
The world of biology has already been transformed. As genomes have been sequenced, that information will be used more and more in research, to synthesize new drugs and to enable personalized medicine. Being able to analyze genome sequences will become critical, and you can’t do that without the help of not only computing infrastructure, but also sophisticated algorithms and computing abstractions.
What’s also fascinating to me is that some of the social sciences are getting transformed. If you can scan the full text of the literature you’re studying, you can do analysis that was not possible before. You can do correlations that were not possible before. The humanities are really waking up to this potential. That’s exciting, too.
Q: What do you hope to learn about sustainability?
Chutani: Given where we are, my hope is to get more people excited about this problem and take it seriously, because it is an important problem to solve. Currently, there’s not much attention being paid to it, so my goal is to get people excited about this area as we did with digital inclusion three years ago.
Q: What is the problem that you’re looking to address?
Chutani: Can we build our computing systems—the building blocks of the computing systems—so they use a small fraction of the energy that they use today for the same kind of work and the same amount of work? Can we build systems so that they do not generate the level of electronic waste that they do today? Can I build systems that allow me to provide new functionality without having to throw away the whole architecture?
As a crude level of approximation, when we upgrade to a new computing system, we change the whole infrastructure. That doesn’t make sense. It should be possible to build systems where if I want to have new functionality in the system, I can swap in some small pieces and go forward. It’s conceptually possible to do so, but we haven’t looked at the architectural approaches that will make it easy.
We want to get to the point where, four or five years out, the computing systems consume a fraction of the energy, and we think it’s possible to do that and that they can last much longer. We have a responsibility, given the climate crisis, to do what we can. It’s our responsibility to stimulate innovation that will minimize our energy and material footprint.
Q: You and your team have spent a lot of time thinking through the history of computing, the present-day situation, and where the field is going. What are the most notable successes your team has been able to accomplish?
Chutani: I feel very good about the focus we have brought to contextualizing computing as a way to reinvigorate education. That’s something that will have a pretty big impact on the student body.
I feel very good about the areas we’ve pioneered for the company. They have become big focus areas for our businesses. The robotics group was created directly in consequence of our early work. The attention that the cellphone gets as a platform within the company and within research—I think we can take at least partial credit for that. Same thing with the focus on emerging markets. We started looking at that three years ago, and now it has become a big focus area for the company.
There are many areas I can point to where we have done the right things. The time frame is long in some, but our work in digital ink and pen is beginning to bear fruit. The folks in PowerPoint are talking to Andy van Dam of Brown University to see if it makes sense for them to incorporate pen-based interaction in PowerPoint. How many hundreds of millions of people use PowerPoint? If that bears fruit, it will change how they interact.
We have planted many, many seeds, and some of them are showing promise, and some of them have already come to fruition, in areas like robotics and gaming. My goal is to stay ahead of the curve and plant more such seeds.
Q: What sort of responses do you receive from the academic world to the work your group has done and the work your group is planning to do?
Chutani: I would say enthusiastic—even from people who don’t get funded by us. They appreciate the openness and transparency of our external research model. They understand the strategy and how to engage with us. By and large, I would say, they like the choice of problems we’re going after. It’s not very unusual for me to hear, “Oh, why aren’t you looking at this, as well?” But I don’t get questions that say, “Why are you going after this?”
Sometimes people might say: “Wow, this is interesting. So Microsoft is actually focused on healthcare? Microsoft is interested in synthetic biology?” They’re surprised sometimes, because they don’t think of Microsoft as engaged in those areas, but they’ve never been surprised and come back to us and said: “This is not an interesting problem. Why are you doing this?” Nobody questions that what we are working in is interesting, it’s valid, it’s meaningful. Of course, we are making choices like anyone else does, driven by our understanding of where the outcomes may be the best and where we have assets, tools, software, expertise, and, to some extent, the interest of people we have in the group.
That’s what goes into making the choices that we do, and I think the model and the choices resonate with the community. We don’t hear much dissent where people think what we’re doing doesn’t make sense. We do get a lot of requests to do more, which is a good place to be.