Rick Rashid: MobiSys 2005

Remarks by Rick Rashid, Senior Vice President, Microsoft Research
MobiSys 2005
Seattle, Washington
June 6, 2005

RICK RASHID: What I want to do today is talk about a variety of things. When you starts putting together a talk like this and you send in your title and your abstract, over time your mind changes a bit as to what it is you actually want to say. I’m going to say a variety of things today but I’m going to start out by addressing this subject because it’s been on my mind now for the last almost a year and certainly fairly intensely for the last six months and it dovetails into the topic that was in my original abstract, and I’ll get there as well.

I mean, one of the things that has been concerning me over the last six months is really a feeling that there is something of a crisis of confidence broadly in the technology field but, I mean, I feel it particularly strongly when I visit universities and talk with some of the students and people that are out there.

I mean, a lot of this came in some sense out of the fairly apocalyptic dot-com bust — this is supposed to evoke the apocalypse here. I think I see the horses and some of the flames coming down. It wasn’t quite this bad but if you lost your job, it probably felt like it.

And what happened with the dot-com bust is it really made people start thinking about, wow, this whole thing is falling apart, maybe this computer thing, this technology thing is really coming to an end. And so you got articles, especially in 2002, 2003 timeframe where people were going out there saying, you know, well, IT doesn’t matter anymore, it’s not really helping people, it’s causing more problems than it’s worth. Dijkstra has this end–of-computing science title paper he put out, the “have we reached the end of IT?” history; I mean, these are pretty strong statements that make you hopefully feel pretty bad, since I assume that’s what they were intended to do.

This is a science editorial that Dave Patterson and Ed Lazowska put out last month and I don’t know if you read it or not but it’s basically pushing on a more recent aspect of the problem, which is the issue that it was the funding in universities and in particular the very significant pullback in funding for long term basic research at DARPA. So Ed and Dave have been really pushing on that and I was just in DC about a month ago addressing this issue with many congressmen and really trying to push on this notion that the National Science Foundation really needed to be appropriately funded.

But there’s another aspect of this and I know a number of you in university environments are seeing it, which is we’re seeing this basically freefall drop-off in interest in computer science among students. This is the latest CRA study, I think it came out about a month ago, and here it’s a chart from the study that sort of shows you what’s happening. And, of course, the scary part of the slide is the fact that since roughly 1999, 2000, the interest rate among students has fallen off almost linearly, so roughly speaking in 2002 the interest in computer science among incoming freshman had dropped from about 30 percent from its peak. By 2004, so last year’s number, it had dropped off by 60 percent from its peak. And you can see that for women in particular the numbers have dropped of below the height, the lowest previous historical point in terms of general interest in computer science.

Now, if you look at the diagram, you’d say, well, gosh, this has all happened before, this is a cyclic process, it’s happened before, maybe we shouldn’t be as concerned about it, but the issue here is that there is more to what’s going on now than what occurred before and my particular concern is you see a confluence of a variety of issues, a drop-off in interest in students, an overall decline in math skills and math training for incoming freshmen, which tends to predispose them away from science and engineering disciplines, a sense amongst students that if they take — you know, if the job market picks up, especially in the United States, that the jobs won’t be in the United States, they’ll be elsewhere, although this trend, by the way, is visible in Europe as well, it’s not quite as well pronounced but my European friends say that you’re also seeing a drop-off in computer science interest in Europe.

So it’s a very scary process and it’s even scarier because it’s not like there aren’t jobs. If you go to Microsoft and you look around the various product teams at Microsoft, there are tons of open positions. I mean, we can’t fill — we hired more students off college campuses this last year than we’d ever hired before; we could not fill all the positions that we wanted to fill from college campuses and broadly within the company, thousands of jobs are going begging and we just can’t find qualified people to take them.

So it’s a funny process; there are jobs available at least at the high end of the skill set but students perceive it to be the case that there really won’t be jobs or in any case they’re not interested in going into computer science.

There’s another part of this, which people are beginning to — I see some of my friends in the press and some of the more lay people that I talk to beginning to pick up on, which is the feeling that, well, maybe the computer thing has kind of petered out, Intel and AMD going out there basically saying, you know, we’re not going to make faster processors, or alternatively they may not say it quite that way; I’ve heard one spokesperson for AMD say, “Well, maybe we’ll be able to eke out a 10 percent a year improvement in clock speed, but just maybe.” I mean, we’re already up against the law of physical limits. One of the techniques that’s historically been used is dropping the voltage; well, the voltage has dropped so low now and so close to zero that the difference between a one and a zero is getting very, very close to thermal noise.

The biases, the geometries in the existing chips are such now that there is some substrate, some layers that are literally only four atoms thick. It’s not really obvious how you make it any thinner and still make it work.

So there’s a set of concerns that’s being raised now of, well, you know, maybe computers aren’t getting faster, maybe the jobs aren’t going to be there, maybe IT really isn’t helping people, so is this the end of this particular cycle and maybe we should all be doing something else.

My understanding is the fastest growing undergraduate major is recreational management. I don’t know what recreational management actually is, but somebody evidently thinks it’s a good thing to be doing.

So I think a question — actually it was funny, I put this in the slide and then on Friday I had somebody in my office ask me pretty much ask me exactly this same question. I mean, this was a layperson, someone that wasn’t in the technology community, but basically said, “Look, I’ve just been in Silicon Valley, things don’t look so great there, maybe we’re just done, maybe we should be thinking about something else.”

Well, what I want to say and what I want to emphasize, and I’m going to make these points as I go and really push on these issues, is that I look at what’s going on in the labs, I look at what’s going on in the universities, in research labs like Microsoft and IBM, what’s going on in companies and startups and to me it doesn’t look like we should be pessimistic, it really looks like we should be optimistic, that we’re just getting started. There are a lot of new opportunities that are opening up. I’ve listed a few here and I just want to talk about each one individually.

So one of the things that I think is happening right now that’s very exciting to me is that we’ve made a transition from what I would say is just being able to store files to the point where disk drives are now large enough, the storage we have available is large enough that we’re at a point now where an individual can store much of what they would generate during their lives.

So one way of thinking about this is a terabyte of disk is enough storage to keep all the conversations you’ve ever had from the time you’re born to the time you die — even for people from New York, especially for people from the south. It’s just you’re not going to fill it up.

I’m not suggesting you necessarily want to keep every conversation you’ve ever had, but at the same time the notion that you could augment your own memory could be very attractive. I mean, I know thinking about my own personal life, my father passed away about now I think six, seven years ago and after he passed away I realized that I had very little left of my father. I didn’t have that many photos because, really, photography really sort of blossomed after I left home and I didn’t get home all that often. I didn’t have any recorded conversations. I don’t actually have a record of his voice at all.

And so this notion that there may be things in your life that after the fact you may want to go back and get, your conversations with your parents, your relationships with your children, the things that occur, that just happen, that suddenly afterwards you say, boy, I wish I had a picture, I had that recorded, I had that available to me.

So that’s something, and it’s not just audio that you could keep; you can literally store a picture taken once a minute for your entire life, again on a terabyte of storage. And you could be still taking pictures and it will still work out.

So we’re really at this point now where the things that happen to a person we could imagine keeping track of and recording.

Now, that’s kind of an abstract way of talking about it. One of our research groups in Cambridge, I think partly because I’ve been talking about this for a while now, one of the researchers decided to actually build something that would do some of what I just said. And so they built something called the SenseCam and it’s really a portable device that some of you may have already heard about it or seen it, that amounts to a black box for a human being. And by black box, I mean it records a lot of the things that are associated with the person, much like a black box might record the things that are associated with an airplane or some other kind of vehicle. The notion is that it has a lot of technology in it that allows it to sense what’s going on in the world around you and keep track of information that occurs during the day.

So if you look at what’s actually in one of these SenseCams, it’s a lot of sensors, a lot of devices. There’s an accelerometer for motion detection, infrared sensors so you can know when a person has walked in front of you, light sensors, sound sensors. The newer ones have GPS and GPRS support. There’s a wide angle lens so that basically anything that would have been in your field of view would show up on the pictures that are taken. And every time something happens, the camera will take a picture and by something happening meaning some sensor shows a significant change in the environment around you. So maybe you’ve just walked from the dark to the light, maybe you’ve just had someone walk in front of you, maybe you’ve fallen — and you can’t get up — who knows what’s going on; well, it will be able to keep track of that. It knows when you’re walking, it knows when you’re stationary, it knows when there’s a loud noise that occurs around you.

Now, to give you a sense of the kinds of pictures — everybody likes the pictures part of this the most. To give you a sense of the kind of pictures that it takes, here are some examples. And these are wide angle lens pictures but you all know the mathematics of photography well enough to realize that if you wanted to take these wide angle images and turn them into things that look like normal pictures, you can easily do that, just through a certain amount of mathematics to undo the effects of a spherical lens.

Now, one of the issues when you start thinking about doing a device like this is you realize, well, you can’t just like randomly take pictures because the reality is that you’re moving all the time and it’s going to be jittery, things are going to go wrong, you need to make sure that the camera is taking a legitimate picture, otherwise it’s not going to be very valuable.

And so one of the things that the team did was to say we’ll take advantage of all these sensors that we have and in particular the accelerometer and the time sensors and so forth and say, well, what we’ll do is make sure that when we take a picture it’s appropriately stabilized and it uses some image stabilization techniques in the mathematics as well to make pictures look like they would be taken by a person in a reasonable environment.

So I just want to give you this is just a very quick little clip. This is Lindsay Williams walking through Cambridge and so she’s the person that originally developed the device and you can see her walking through Cambridge. There was just a black spot there. She had physically covered the lens because she was going into women’s clothing stores and she wasn’t that interested in having it take the pictures while she was there.

There’s a hand gesture you can use to tell it to stop and you do that by putting your hand in front of the lens and it recognizes that hand gesture telling it to stop and then you turn it back on again.

So that’s the sense of here so what kind of images you might imagine taking.

Now, of course, I said this is a black box for people but this is really just a sensing device that could be used almost anywhere. For fun — I think it was for fun — one of the researchers decided to stick the camera in a refrigerator at the lab and produced this little video. Notice the little girl stealing stuff out of the refrigerator.

Now, again, I don’t think there’s a really strong value associated with putting a camera in a refrigerator but purely from a research perspective, we have now proven beyond a shadow of a doubt that when you close the refrigerator door the light goes off. (Laughter.) So I don’t know when that’s going to be published but you can look for that in Nature or something coming up.

Now, what can you actually use this device for? Well, one of the things that frankly it was built without any really good reason, just as an experiment to see if we can do it, what would we actually do with it.

As it’s turned out though, we’re getting a lot of interest from a lot of people in this type of mobile device. One of the areas is clinical, particularly with two different kinds of patients. One, for the not to severe memory loss patients, for which a device like this can be an aid, can help them remember the things that happened to them during the day, they can keep a log, they can go back and check that. There’s a clinical trial going on right now in England using the device for that purpose and some of the preliminary results that are coming back are actually very positive so that’s very exciting.

Potentially it’s also an aid for caregivers for severe memory loss patients. They probably can’t use the information coming from the device but it’s an opportunity for caregivers to keep track of what was happening to a patient while they were maybe being taken care of by someone else or while they were on their own for some short period of time.

It’s also being looked at as a tool for reflective practice, where you would like someone to be able to see what they’re doing or how they’re doing or what they’re doing, how it affects patients they might be working with, for example, psychologists.

We’re conducting privacy studies because obviously one of the issues that devices like this create, as in camera phones and many other kinds of video devices and audio devices that are out there now, is what does this do to people’s sense of privacy, how do they think about devices like this.

Because make no mistake, they will be out there. I mean, this is an early one but it’s a technology now that is available to make devices like this something that anybody can build and once that happens they will get out into the world, so we’re actually looking at what is the etiquette.

And one of the things we find is it varies. I mean, notions of privacy and privacy standard vary dramatically from one part of the world to the other. In the U.K., for example, there’s a much greater acceptance of cameras in public places than in the United States and you see that in the studies you do in the way you talk to people.

Another area that’s gotten a fair amount of interest is tourism. We’ve got a number of tourism companies that are saying, look, instead of our people spending their lives behind their cameras when they go on tour and they’ve never actually seen the Tower of Pisa, they’ve only seen the LCD display that shows the Tower of Pisa, now there’s an opportunity — or they’ve never seen their kid’s wedding, they’ve only see what the camera sees — now there’s an opportunity for devices like this to exist and keep track of this information and do just as good a job that the person was likely to do in taking pictures but now have a significant record that someone could carry with them. So we’re experimenting with things like that.

And we’re in the process now of building a new generation of these devices that we’ll be making available more broadly to the research community. We’ve gotten a lot of interest from researchers around the world, not to mention interest from DARPA, interest from the Metropolitan Police in London and a number of other parties that are interested in how a device like this might be able to help them with their particular setting.

So that’s this notion of kind of the human-scale storage and being able to augment human memories with new kinds of devices.

Another thing that’s happening out there, and I think this is pretty exciting, I was just at the Emerging Technologies Conference about a month ago in San Diego and I saw some work like this as well — what I’ll be showing you is work at Microsoft — is this notion that we’re moving into an era where almost any surface can become an input/output device. And let me just show you this little video clip and I’m going to turn the audio off so I can just talk about it.

This is work that is particularly done by Andy Wilson at Microsoft Research in Redmond. And what you’re seeing is the idea of tying together a short throw — in this case it’s a short throw LCD projector in combination — or actually it’s a DLP projector — in combination with a camera, sort of computer vision techniques. And the notion here is that you can use something like this to turn virtually any surface into an interactive computing surface that you can do a variety of things with.

Physical items that you might put down on the surface now can take on virtual appearances. So Andy’s put down a couple of physical things there and they can become spaceships, they can become little toy soldiers or whatever, if it’s a game environment for children. Throw down a piece of paper and suddenly it’s a video or a spreadsheet or something else.

Andy actually has the whole Tablet PC operating system working in this environment so he literally could write on this if he wanted to.

And again what you’re seeing in this is this notion — and I’ll just stop it right there — that almost anything can now be an interactive surface.

For those of you who have been keeping track of what’s happening with LCD displays, the rate at which LCD display technology is getting cheaper is such now that within probably about five years — at least that’s what people tell me — we will reach the crossover point where on a per square inch basis it will be cheaper to produce an LCD display than it will be to do a whiteboard, the way whiteboards are currently produced. And it’s very easy with these displays to put an input surface on them as well and, in fact, there’s some new companies in this space that have basically made that all but free.

And so now you have this prospect that almost any kind of surface that you may want to put out there could suddenly be an input and output surface and suddenly be something that you can interact with, and I think that’s going to change the way people think.

Here’s another example, this isn’t so much the whiteboard kind of case but this is again this notion that you can do new things with input/output devices and displays. This is Andy again and basically this is the same technology you saw with the surface computing, but instead of using a table, he’s now using holographic film that he’s projecting against. What that does is it gives you now the ability both to be able to project against the device, against the surface and have the images sit on the surface but also you can put cameras behind it and how things that are seen in the surface can be picked up on by the computer and interacted with.

So this is just Andy doing some registration work and one of the things you’ll notice is that the image you see of him, that’s actually projected back from the display, that’s not like a reflection. And the fact that it’s wider, closer, he’s just using that as part of the registration process to show that once the system is registered it knows what’s nearest the display and what’s farther away. It knows when things are being put in front of it and again I think he’s just going through a little it of registration here. It knows when things are being put in front of it, it knows when there’s a person there, it knows when there’s a hand or a pointer. You can use this for doing fancy colored light shows. This is again just a way of showing up how these things looks.

And this is very “Minority Report”-like. Actually, Andy was one of the people they consulted with for “Minority Report”.

I think this is the part I wanted to get to. But you can also take physical things and basically have the system take a picture of them and then interact with them. In this case, he’s just taking a magazine cover and he’s putting it up there and now he can physically interact with it in his hands in a very natural way, but you can imagine maps, other kinds of displays that you might want to physically manipulate, you can now imagine doing with technology that’s like this.

Okay, so that’s another area. I think again it’s exciting to me because it opens up a whole new range of things that you can do with computing devices if now any surface can be a way of interacting with those computing devices.

Another thing that’s happening out there is the combination of storage and these incredibly high performance networks, I mean, we’re now doing basically what amounts to circuit-switched optical networks of 10, some people are now looking at 100 gigabits from point to point across the country where you can now move huge amounts of data around, and you’re seeing it change the way scientists and many disciplines are thinking about doing their work.

Here this is just a pointer, if you haven’t seen it, to something called the SkyServer, that’s done by Jim Gray in San Francisco. And Jim worked with the astronomy community to basically create a global, federated database of astronomical information, starting with the Sloan Digital Sky Survey but then moving on to a broad collection of different telescopes. And, in fact, this is by the way if you go to the SkyServer Web site, SkyServer.sdss.org, this is what you’ll see and there’s a ton of stuff here. It’s great for the layman, it’s great for the scientist and it’s great for students and there’s 150 hours of educational material on here for students and teachers that Curtis Long in our research lab in Redmond helped to design.

So there’s a huge amount of material here and it’s allowing scientists to do data mining against a large number of different data sources, using Standard SQL, using standard schema to represent the information that’s out there. And it’s changing the way they think about doing their science; they’ve now found a number of new astronomical phenomenon simply by being able to data mine this data source. Literally with three lines of SQL you can go in and find all the quasars in the Sloan Digital Sky Survey. I mean, in the old days that’s just huge bodies of dusty deck C or FORTRAN code and who knows what format the database was stored in and so forth. Now you can imagine putting that online.

And the Sky Query Portal is the place you’d go to actually query against all the archives that are online, currently there are 20 that are online. We started with four and it’s up now to 20.

And you can plan queries. In fact, one of the things that Jim did is create an environment there where you can store the temporary results of queries, manipulate those results and then get them back to you and it’s all programmable through a standard either Visual Studio .NET but it’s all Web services and if you want to use Linux and eMacs you can do that, too, not that I would personally recommend it.

OK, so those are some of the changes that are going on that are exciting and those aren’t the only ones, but obviously I mean the subject of this conference on mobility, mobile devices, interconnectivity, I mean, those are the things that are also really driving I think a new era of computing. You’re seeing evolve now what I think is really an ecology of mobile and fixed computing devices that start with things like laptops and PCs but include the handhelds and mobile devices and increasingly smart objects, smart input and output devices, smart effectors, sensors, things of that sort and it’s really changing the way I think people think about computing in a significant way.

I mean, here this is just a cute picture saying pretty much the same thing but this notion that you can tie these smart devices, you can tie sensors like health sensors in and a variety of different things.

Obviously computing devices, these smaller devices are just growing by leaps and bounds. All you have to do is go to China or India and see the rate at which the cell phones are penetrating that space.

And these are real computing platforms. Here is the example of the Smart Phone computing platform. You can do as much or more with these as we were able to do not very many years ago with the best computers we had available and you’ve built up a huge collection of software components that lets you do exciting new applications, whether those be communication applications, sensor applications or what have you.

And it’s not just things like phones. You think of a phone, okay, well, a phone is big enough I can imagine a processor in there, I can imagine doing a fair amount of computing but things as simple as watches now have pretty substantial programming environments on them.

We came out about a year and a half or so ago with the SPOT watch, this Smart Personal Objects watch, and this is one of them. This is the one I like the best and it’s also the most expensive from Tissot, but there are quite a few different models that are out now and you can see some pictures of them here.

And the idea here was to put a substantial computing environment on the smallest possible chips that we could find that could operate at power levels that you could literally put in a watch and have that watch work for several days.

We had to put a network to support it and so we used the FM subcarrier concept to do that, developed a new way of doing FM subcarrier so you could get more data and have better error handling characteristics for the data; designed a chipset based around the ARM 7, running at about 27 megahertz; and really built an entire infrastructure to support those types of devices.

Running on this, again much like the Smart Phone, it has a very substantial programming environment, you program these the same way you program your handheld or your PC or something else, it’s done in our case using the Common Language Runtime environment. You can use C#, you can use actually any of the languages that are supported by the CLR and Visual Studio to program against it. And so you can imagine now building a pretty broad collection of different kinds of devices around this type of thing.

Now, what does that enable? Well, as we see these devices proliferate, you now have the opportunity to do just enormous numbers of new applications. Here is one I particularly like, this was developed by one of our researchers, Eric Horvitz and his team did this in Redmond. It’s the notion that by collecting a lot of data from sensors — in this case we’re actually using sensors that are stored under the streets in the Seattle area — pulling that data in, reasoning about it, you can now begin to do a lot of prediction and provide services to people to allow them to get around traffic or traffic problems.

So the general technology here is called JamBayes and it’s all based on Bayesian reasoning and it helps you get rid of traffic jams so that’s why it’s called JamBayes. It really fits into what’s called streaming intelligence in Eric’s team, which is this notion that you can stream data in, constantly be assessing it from an environment and then make decisions and get that information out.

In this case, the output device is a mobile phone and there’s an application on the mobile phone called Smart Flow, and quite literally I used this, this morning, to get in here. I saved probably about 15 minutes because I looked at it as I was driving, I realized that there was a traffic problem on the 520 bridge that was going to slow things down.

And one of the things that the software is able to do is it actually does prediction, it can tell you roughly how long a traffic jam is likely to stay in place. And the reason it’s able to do that is it has historical data over a number of years about traffic flow, time of day, weather, various events, traffic incidents that occur and it’s able to then use that to say historically this has happened, so what is the probability that the traffic jam will last for a particular period of time and it’s able to compute that. And it does a really good job of it, so in my case I looked at my phone and I said, oh, hey, there’s a problem here on the 520 bridge, it’s not going to clear up for about an hour, so I’m going to go the other way and it saved me a considerable amount of time.

Moreover, I mean, as you think about this, we’re collecting the data for the system on the road system itself but it’s actually something if you imagine phones becoming smarter, GPS devices being located in cars, an ability to transmit that data back up to a central source, we’re not very far off the point where literally any vehicle will be able to collect data on what’s happening to traffic and that could feed into a system like this and then be able to predict traffic and tell you what’s going on not just on the major highways but on every highway, every road in the system. And in some senses it’s a collection of vehicles that are moving and it’s a better sensing network for this kind of information than anything you could pick up in a normal way. We’re actually licensing this technology out to a small company that’s a startup in this space that really is planning to get into exactly that business.

Well, that sort of brings me to this whole question of how ubiquitous computing is really going to impact sort of the broader population. If you look at the world out there, there’s a lot of people that haven’t really quite made it into the world economy and they live in villages, they live in rural areas and they’re effectively not able to get information that’s important to them and to be able to do the right things, whether it’s plant crops at the right time, solve simple medical problems, deal with situations, find out about the weather and so forth.

So there are a lot of livelihood decisions that a lot of people out there have to make with very incomplete information and because they’re outside the sort of normal part of the connected world they’re not able to get that and they have to pay a premium for it.

Now, I don’t want to suggest that computing is going to solve the problems of the world because there are lots of issues and it’s not just about technology, but I do think the technology has an opportunity to help.

Now, I’ve already said it’s not just, you know, technology, there is a thirst for connection, a thirst for access that exists everywhere. I mean, in the developed world you can see all the new projects that have been started up, these are projects in 2004 where small towns, cities, big cities are starting to create infrastructures for providing wireless connectivity to their citizens 24 hours a day, seven days a week, and really satisfying a need for access to information.

It’s particularly important because I mentioned in the Third World there’s a desire or need for better information but even in the developed world if you look at the income levels of people, what you’ll find is that the lower income levels, people are not very well connected.

So this is a chart here where you see that if you’re over $100,000 in family income, you’re in good shape, the chances are 85 percent that you’re going to have broadband access but if you’re under $35,000 here, the chances are good that you’re not going to have any kind of good online access to information.

Of course, in the developing world, I already talked about that a little bit. I love this picture, actually Victor Ball gave me this, I don’t know where he got it from but this is a great picture. It gives you a sense of the sort of juxtaposition of this need for access to information along with some of the realities of living in rural areas.

There are many projects that are being developed, these are a number of ones that are going on in India to provide interconnectivity. And you see this great juxtaposition of straw huts with satellite dishes on them; again, as an indication of sort of how much people would like to get access to information.

Now, in urban areas, if you ask people what they would use computing for or what they want to use computing for, this is the kind of information that you get back. And I’ll particularly note here that this is actually the result of a study and set of focus groups, in particular I know the Seattle area, I think the Austin area were the two places they collected this, and you can see people want inexpensive access to the Internet but they want to buy goods, they want to game, they want to be able to feel more secure in their environment, they want to get access to medical information and, of course, they want inexpensive access to other people, voice over IP, chat, e-mail and things of that sort. They want to be connected, they want to be socially interactive.

You can do the same kind of study in the rural areas. Kentaro Toyama, who is one of the assistant managing directors in the lab we’ve just started in Bangalore, has been going around to rural villages and at this point visited a ton of them and really trying to find out what are the potential uses for the interconnectivity in these rural environments and what do people want; things like support for agriculture. Obviously their livelihood depends on their ability to produce crops.

E-government, and you wouldn’t think that would be such a big deal but there are a lot of issues that these rural people face in terms of working with their government. Simple things like land transactions and knowing who owns what, those are critical issues and historically have been hard to work with the central governments to get this and so there are opportunities to do it.

Computer training, and the reason computer training tends to show up very high on this list is because people want better lives for their children. I mean, they want to be trained because it gives them potentially access to new moneymaking opportunities but they want their children trained because they view that as a way out, a way for them to have an opportunity that they didn’t necessarily have.

Telemedicine, a simple thing. I met with several ministers when I was in India recently and this notion kept coming home of very simple interventions can make huge differences in some of these rural villages. Just being able to tell someone what to do when they have severe diarrhea or when they’ve eaten something that may be poisonous or if they’ve got a common infectious disease that needs to be treated in a particular way or they need to get access to certain medicines; getting that knowledge to them is critical and they view it as really important and it can make a life or death difference more so than perhaps in developed countries.

And, of course, communication, that’s critical. People have families, people have extended families, they want to talk to people in other villages, they want to interact and communicate.

Now, obviously some day we could build wired networks out to everybody in the Third World or for that matter maybe we could build out to everybody in this country. We don’t have that today. My mother, for example, lives in rural Kentucky. The only thing that connects her to me is the telephone, which is actually very bad because she had a brain hemorrhage about 10 years ago and so she can’t actually talk; I mean, she can say a few words but they’re not actually related to what she wants to say, and so it’s a very strange conversation to have. I’ve tried putting one of these little video cameras that work over the telephone to her, doesn’t work, too complicated, she can’t figure out how to make it work on her end. So the only way we’ve ever made that work is if we get some young relative to come over and set the thing up for her. That’s not a great solution.

And if you look at some of the published numbers about what it takes to wire an environment, the costs are just too high. If you really wanted to do that everywhere in the U.S., it would just be a huge expense.

Wireless is obviously a potential solution but if you want wireless to work, it has to be easy. I mean, the thing that made wired networks expensive isn’t the wire, it’s the truck roll, it’s the helping people, it’s the solving their problems when things don’t work, it’s the administration and the support and all the robustness problems that you run into, the tech support issues.

So you need to be able to build things that are self-managing, self-connected and interactive. And obviously one of the areas that your community has been looking at is this notion of using things like wireless mesh as a connectivity solution, whether it’s in a hub and spoke kind of system, it’s a little bit more classical, more like the way the cell phone network works, or more of a real mesh environment.

And one of the values of thinking about real mesh technology where everybody potentially can route and manage traffic is that you can imagine building systems to be much more redundant, to be much more usable in environments where you otherwise would not have access to that kind of technical support.

A lot of people out there are doing this work, there are some companies in the Seattle area, there’s a company called Seattle Wireless that’s working on this. Obviously there are a number of universities, both in the United States and in some of the developing countries, that are working on projects to address these kinds of issues.

We within Microsoft Research in Redmond have been particularly looking at this whole notion of how you would build robust wireless Mesh Networks that you could literally plop down anywhere and provide access and connectivity.

Now, let me just show you a quick little video that Victor Ball and some of the people on that team put together to talk about what they’re doing.

(Video segment.)

RICK RASHID: And by the way, I don’t know if anybody has seen some of the announcements on this but this notion that you sort of plop down satellite connectivity somewhere and plug it into this kind of a mesh is starting to get some traction in the sense that you’re getting companies out there now that are designing devices that basically are very small, very simple, provide Internet satellite connectivity, cost about I think the most recent one I’ve seen, about $3,500, requires no intelligence on the part of the person other than to drag the physical device somewhere where it can actually see the sky. It’s about five inches thick, about this big around. It was really designed to be working on top of a vehicle in motion, an SUV or RV or van. And those devices will be out in the market beginning in the U.S. beginning in the fall of this year. And by the way, it includes an 802.11g gateway built into it. So all you do is plug it into the power and that’s it, just give it power. In fact, you can put a solar panel on it and then you’re done.

So there are some pretty exciting things happening in that space.

Now, there’s a lot of work to do. I mean, we’re not — well, the computer field in general has never been very good at things that are really truly self managing and autonomous and so forth but this is an example where we really need to do the job right, otherwise it’s just not going to work. You can’t have trained people out in rural environments able to make things work for someone when they go wrong. You need to be able to have these be self-managing, you need to make good use of the spectrum, we need better tools for analyzing what we’re doing and there are lots of issues — privacy, fairness, security; all the things that plague us in the wired world really are critical in this environment are well.

Now, one of the things we’ve been doing in the research group at Microsoft is really trying to reach out to a broader community to really help move these ideas and these kinds of technologies forward. Last year I participated in the Mesh Networking conference and summit that we had. That was great. We brought people from academia and people from small companies, people from large companies together to really talk about Mesh Networking, some of the issues, some of the problems and some of the things that are going on.

Just last week we held a self-managing network summit, again trying to bring people from many different parts of the community together to look at the issues associated with building and keeping going these kinds of self-managing networks.

Another thing that we’re doing is the software that we’ve been developing for the Mesh Networking work that we’re doing, we’re going to be making available. I think this is a prototype, I think it’s available in about a month, right about the time we do our faculty summit event in Redmond. We’re going to be making this available broadly to the community. It’s all the software we’ve been developing for Mesh Networking in source form, along with whole bundles and oodles and gobs of all the necessary Microsoft tools and things that go with it to allow you to do the things that you need to do, and a lot of other technical information.

So this is an example of again us just trying to work with the community and get stuff out there and make it broadly available to people.

And we’ve been doing our piece in getting a number of universities, especially both in the U.S. but also in India looking at exactly these issues and here are just some of the projects that we’re in the process of funding and working with people in the outside. So we’re sort of putting our money where our mouth is in a number of these areas.

So let me just summarize a bit. I’ve talked about a variety of things. I started out really saying, look, I go out there, I see people depressed. I don’t like seeing people depressed, I’m kind of a positive, optimistic kind of person. You pretty much have to be if you work in the operating systems field; otherwise, you’d give up. (Laughter.) So I want to get people out of their depression.

I look at what the opportunities are out there and many of these are your opportunities. Many of these are opportunities in the mobile computing, mobile device space. And there’s so much going on now, so many new technologies being brought to bear that I think will really allow us to do completely new kinds of applications, get technology in the hands of ordinary people, not just getting everybody in the world to have a PC but getting everybody in the world to be able to get the data that they need, the information, knowing when to do something, being able to communicate with their family and friends, being able to train or educate their children.

These are really critical problems. When I go to China, I meet with the ministers in China; literally, the Minister of Education, having a conversation with the minister and there were sort of a sense of despair, how do we get education to the — we have 300 million, maybe going over 350 now that are moving into the developed world economy in China and we’ve got a billion people sitting out there that aren’t. It’s hard to get teachers to a lot of these places, it’s hard to get information to a lot of these places. The same thing is true in India, 650,000 villages in India, how do we reach them, what kind of technologies can be brought to bear, how do we get information out.

Well, I think we as a field can both accelerate what happens to the other sciences, we can do good in the world in many, many ways. And to some extent, when you talk to students, if you say what is it that motivates young people, going back to one of my earlier slides, it’s that ability to change the world to do good, to make the world better that I think motivates and excites young children, young kids, middle schoolers, high schoolers and students if they be in college. I think we can go out and put some of the romance back into the field of computer science by going out and telling these stories, getting the young people to realize what the opportunities are, where some of the change can come and how that can impact the people that they care about and the people that they know.

So with that, I will stop and I’ll be happy to take a few minutes worth of questions. I’m not sure what the etiquette is but I see microphones here so I bet that’s part of it. (Applause.)

QUESTION: I enjoyed your talk. I applaud efforts to bring technology to the developing world, I agree with you, it can make a big difference. I have a question about how we might take the technology once its deployed out from around the poor area, and the countryside is littered with exciting things that have been done in agriculture and so on and they weren’t able to sustain it, so abandoned projects basically. I’m just wondering if you have any thoughts on how we might sustain these networks and so on. They need people on the ground to maintain them and so on.

RICK RASHID: Well, I think the thing I’ve heard over and over again, when I’ve been in India, when I’ve been in China, when I’ve been in the U.S. talking with NGOs and some of the people that are out there in that space, is that ultimately this can’t be an act of charity, right, this can’t be sort of an act of benevolent government because you wouldn’t be able to sustain it. There has to be an economic value, there has to be something that brings more money into the community or into an environment because the technology exists. Then it costs to keep the technology working or to keep the pieces going.

And if you can’t do that, then this is just not going to work. If that means that you’re doing Internet kiosks, that the person that runs the kiosk in a town, they need to be able to make money, it needs to be a job that they’re actually making money.

And when Kentaro has gone around to different villages, he sort of categorized what were the success places and what were the ones where just technology got brought in and it just went into the woodwork and disappeared and stopped being used.

And the difference was, were the people who were in charge of technology, the person that ran the Internet kiosk, the person that ran the access point or whatever, was it a business for them, did they make money, because if they did, then they maintain it, they’ll learn more, they’re willing to make an investment. Other people see that, they think it’s valuable, they think it’s an investment.

So it’s interesting, last week or a week and a half ago, I was at the Fire Conference down in San Diego and the CEO of iRobot was there and he made this comment about robotics and it was a very similar comment, which is he said, well, there are a lot of companies that have been in robotics, especially consumer robotics just because it looks good, it’s a cute little dog that helps your brand, it’s a walking robot that costs $4,000 that makes people feel good about you and now you want to buy their products, he said, but the business that they’ve gotten into with the Roomba was the business where you actually built something that cost the consumer less to buy than it was worth to them. And we’ve got to do that in the Third World, too. It has to be the case that it is self-sustaining and there’s an economic business associated with it. And I know in the work that we’re trying to do now as we’re working with some of the NGOs both in the big cities in India and the rural villages is to make sure that there is an economic model, because it’s otherwise not sustainable, charity doesn’t work.

Yes?

QUESTION: I think a lot of these positive effects that, I mean, for a cheaper technology has a lot to do with the applications we make of technology. So like my question is just like as a computer scientist and as an academic discipline, how agnostic can we afford to be with regards to applications? And if we get deeper and deeper into applied kinds of things, then it’s going to affect the reputation of computer science as a science, for example.

RICK RASHID: Well, it’s interesting. You know, I was at a CRA leadership summit on this subject about three months ago now, I think. I’m not sure I remember the exact date so it might have been two and a half months or three and a half months ago. And somebody actually raised this point, they said, well, you know, one of the things that’s going on in the field right now is that more and more you see computer science being critical to biology and, for example, just in my research group we’ve just started a computational biology center in Trento, in Italy, and that’s an area that we’re working in. I have researchers here in Redmond that are doing work on developing an AIDS vaccine. And the reason isn’t because they’re working in the AIDS vaccine, it’s because the algorithms and the technology and the ideas that they came up with were applicable and they had friends in that community and they built up relationships and something happened.

I think increasingly you see a lot of this crossover and somebody raised this point and said, well, are we going to lose the identity of the field. And it seems odd because in my generation, and some of the people out here are in my generation, I see a few, in my generation, computer science wasn’t much of a field. I mean, I came from mathematics, I think most people that I know came from physics, mathematics, biology, chemistry or music or something else. And so we went through probably a more introspective period when we were more about and less about other people and now I think there’s the opportunity to carry that back out.

And, yeah, I think it brings some risk associated with it. I think from a career perspective there’s always a question of how universities, tenure committees really treat that. I think from a funding perspective I’ve seen issues, especially between the different NIH, NSF, DOE; there are things that fall between the cracks and sometimes just don’t get funded.

And I think one of the things that I’ve made a priority for my organization, we spend about 15 percent of all my basic research money actually goes out to universities and one of the priorities that I’ve put on the money that we spend, I mean, we’ll never spend enough to replace the NSF, is to say let’s do the things that these agencies or the universities themselves might not be comfortable doing, let’s be a catalyst, especially in the cracks where they’re either new fields or crossover opportunities where we can help something get going while we’re waiting, while the rest of the world catches up with that as a legitimate enterprise.

So I think it’s a legitimate concern but at the same time I think that’s where we’re doing.

Yeah, another question?

QUESTION: I was wondering if the tendency for less interest in computer science is worldwide or it’s just a U.S. phenomenon right now.

RICK RASHID: Well, I don’t have good numbers outside the U.S. Luckily, the CRA has been doing really good numbers keeping and has been doing it long enough to get the trends there.

Anecdotally, I’m told that by people that I know in Europe that the same problems are occurring in Europe, there’s a drop-off in interest.

I have been told by some of the people that I know in Asia that although the absolute numbers are very high, I mean, the number of engineers being graduated out of China, for example, is huge, like 350,000 a year, but that they see a drop-off in interest as well. I think some of that is that other fields are making themselves more attractive but I also think that there’s always been a bit of a relationship between the sort of economic status of individuals and their families and what fields they chose to go into. In the U.S., for example, you’re much more likely to be a scientist or engineer if you came from a poorer family than if you came from a rich family. That’s just a fact. So as the world becomes richer, maybe fewer people find that as an opportunity to move up.

QUESTION: So I want to follow-up on this question of a cultural shift amongst computer scientists because I’m wondering if there’s anything we can do other than waiting for it to happen and sort of funding the cracks in the interim. So as an example, I’ve been writing a lot of grants recently that have usability as a piece of this overall distributed system, operating systems project, what have you. And the reviews always come back, well, the technology piece is really interesting, forget this usability stuff, not interesting, don’t do that. And, of course, the program managers cut the budget correspondingly. So what can we do to help our colleagues see the importance of this problem?

RICK RASHID: Well, I think again part of what you can do is talk with them and raise these issues. I mean, some of these issues occur within Microsoft, too, to be honest, I mean, in talking to our product groups you get some of the same concerns where fundamentally companies like Microsoft, Hewlett-Packard, Intel, we’re technologists, right, we’re not user-interface design engineers, we’re not designers and so you get some of those concerns that get raised.

One of the things we’ve done is to try to raise, both by creating research groups specifically focused on social computing and user interface design that bring in sociologists and ethnographers and anthropologists and other design people to raise this as an area. For the last couple of years we’ve sponsored social computing symposia really focused on making it more of a legitimate field that can be talked about and you can write papers for and get published in that field.

And we’ve been trying to sponsor more activities that sort of cross over between design schools and the computer field. So for the last three years now, we’ve had something we call Design Expo that this year actually dovetails with our faculty summit. And the key thing there is what we’re doing is we’re getting the top design schools around the world and it’s very, very global — China, India, Germany, France, Latin America — in fact, last year we had a great group from Brazil — get these top design schools to run a semester course focused on designing computer interfaces for particular problems, often focused around the Third World, at least the last couple of years it has been, and social problems, and then running a competition in that space and then bringing those people to Redmond both so that other researchers can see them but also so that our product people can see it.

And a lot of this is just pushing on it, saying, yeah, this is a legitimate area and it’s not just the case that — I mean, I’ve heard people say anybody can design a user interface and then they wind up with eMac. And eMac works fine —

QUESTION: Which computer scientists think is really cool.

RICK RASHID: Yeah, it’s great, OK, it’s fine and it addresses the need for a particular group of people, but it’s not a broad stroke kind of interface and people lose track of the fact that there are a lot of people with different ways of doing things.

So I think part of it is just keep working to raise that awareness and try to build communities because communities are what then get steam and have an impact on the scientific organizations.

I think I’m supposed to stop pretty much after this last question because it’s 10:00.

QUESTION: You talked about sort of the economic moderation to keep some of the investments in other areas going but what about the economic motivation to get them there in the first place? I mean, it seems like most of these areas do not really have a lot of money to buy the technologies in the initial stage. So what are sort of the key economic motivating factors that would, say, motivate somebody, an organization or a profit-making enterprise to actually go out there and enable technologies that sort of bridge the digital divide? Because I think that’s one aspect that we need to focus on in a big way, to sort of get the technology out there in the first place that costs a lot of money. Yes, you can reduce the cost but what is the economic model to get it out there in the first place, can we create something like that?

RICK RASHID: Well, there have been a number of approaches to this. One of the ideas that worked in parts of India and I guess Bangladesh was this notion of basically having kind of a bank associated with a particular enterprise that would allow micro-loans to be taking out to pay for simple bits of technology and the first example of that was phones where someone in a village could effectively buy a cell phone on payments through a loan from this bank and the notion there was that over time the technology would pay for itself if the person who had the technology could really make a business of it and that if you had an ability to get money out there, capital so that someone could capitalize effectively a small business, that once they did that, that over time if they could make a profit, they could turn it into a business.

And really one of the issues in a lot of — I know certainly in a lot of the developing countries is this issue of how you get capital into these regions. I mean, in a lot of villages in India there is no bank. The closest thing to a bank is the post office where you can put your money in and they’ll just hold it for you.

And so I think this notion of being able to have sort of — and I know there are a number of NGOs that are out there working in this space, I know there’s one in the Seattle area specifically that’s looking at how do we fund these kinds of developments in such a way that ultimately you don’t lose money. And part of it is how do you make sure that the people who can take the loans are trustworthy about paying them back, how do you make sure there’s enough of a business model for them to make it make sense.

But again I think it’s the same old thing, get capital out there in a form that people can actually use, create businesses that actually bring more money in than they spend and that’s the way to get the cycle going.

And we’ve had these, I mean, it’s not like this is just a new problem. A lot of these issues were true in the United States at points in the past; I mean, this notion of how do you get capital out, how do you take advantage of it, how do you use it. The developed world has gone through so there’s some opportunities and experiences we can bring to that as well.

And with that, I think I’m done and thank you very much, you’re a great audience. (Applause.)

Related Posts

Rick Rashid: Microsoft Research 15th Anniversary Celebration

A transcript of remarks by Microsoft Senior Vice President Rick Rashid, University of Washington President Mark Emmert, Microsoft Corporate Vice President Dan Ling and other Microsoft officials during the celebrations of the 15 anniversary of Microsoft Research, September 26, 2006.

Q & A: Rick Rashid Looks Ahead

In a conversation with PressPass, the head of Microsoft Research discusses natural language processing and the future of computer technology.