Craig Mundie Talks with Tim O’Reilly at Gov 2.0 Summit
TIM O'REILLY: I'm not sure he remembers it. We were both speaking at an NTT event in Japan about the future of the Internet. Craig appeared at that point by video, which in 1994 was quite remarkable.
Craig is Microsoft's chief research and strategy officer, and the reason I wanted to have Craig here to talk with us is first of all because Microsoft is one of the great platform providers in the history of the technology industry, and obviously knows something profound about how to make successful platforms, but also because as the chief research and strategy officer, he's somebody who is tasked with thinking a lot about the future of technology. And it's really critical that when we start building applications, technology applications, whether in the private sector or for government, that we think about the future and where the future is taking us, so that we don't build for yesterday's technology. Particularly in a world where we have long procurement cycles, it's really important to shoot ahead of the curve rather than behind the curve, and Craig is here to help us think about that.
Please welcome Craig Mundie. (Applause.)
Video of discussion (full transcript below):
TIM O’REILLY: So, I really wanted to start with a conversation about how you see technology changing over the next three to five years, and how that might actually affect this discussion of government 2.0.
CRAIG MUNDIE: I think that some of the biggest changes in the last 20 or 30 years of the technology platform are going to actually occur in the next five to 10, and almost everything that we think about -- the computer itself, the model of interaction, how we write programs, the basic assumptions about connectivity, the role of the cloud-based environment relative to the local computing environment -- all of these things are going to change in some fairly substantial ways.
The microprocessors are all going to change architecturally, become wildly more powerful, and yet more efficient and cost about the same.
I think we're going to see novel forms of radio communications emerge that will make it easier and less expensive around the world to get connectivity, sort of like Wi-Fi on steroids.
I think that we've seen a huge, you could say, a couple decades where the personal computer was clearly a big platform that emerged and powered a lot of people to develop a lot of applications, and then along came the Internet, and Web browsers and e-mail and then all the things like tweeting and Facebook and social networking all were built on that.
But those two worlds were somewhat separate. There was sort of the world of the Internet, and the world of the local computing. And I think one of the biggest changes is in the future we'll think of that as one big distributed system, and we'll write applications that are spread across --
TIM O'REILLY: I think Clay Shirky sold the story about that in 2001 at my first conference on this topic of the Internet as platform, and he said, you know, Thomas Watson once said that he saw no need for more than five computers in the world, and Clay remarked, we now know that he was wrong, and everybody thought of the hundreds of millions of PCs, and he said, we now know that he overstated the number by four, and that is, in fact, a lot of where we are headed: big, global computer with services that are accessed by a variety of devices.
CRAIG MUNDIE: Yeah, although the thing I think will change also fundamentally is the model of man-machine interaction. And so many people would say, well, if I could compute it all in the cloud, can I just make the local device simple? And I think the answer is the local device will become in some ways technologically even more sophisticated, but a lot of that sophistication will be used for the part of the interaction that you can't really remote the computation of, and that's the man-machine interface.
TIM O'REILLY: But there's amazing experimentation. What's the name of the game that you guys are developing where you've got the cameras looking at you, and your body is, in fact, the game controller?
CRAIG MUNDIE: Yeah, you are the controller. That's a perfect example where gestures and machine vision become an integral part of how you interact with in that case the game, but I've given a demo recently of using exactly the same kind of technology to just interact with your computer in an office environment, and you can essentially gesture and move things on walls and very large display environments.
TIM O'REILLY: Right, even in an intermediate stage we're starting to see applications that are starting to use the phone as if it were a 3D mouse, just waving it around and using the accelerometer in there to control some other application.
CRAIG MUNDIE: For this conference, I think it's particularly important to focus a lot on this evolution of the user interface, because the fundamental role of government you could say is to provide services, and they have to be provided in a very universal way. And one of the big questions that is always presented is, how do the less technologically advanced, or the people who don't have access to a lot of these technologies, how do they get those government services?
So, everybody wants to move in that direction and government is always faced with the challenge, both in this country and elsewhere, how do I make it more uniformly available.
So, I think this --
TIM O'REILLY: Do you have an opinion about the whole digital divide? Is that something that eventually technology just catches up with or there need to be progressive efforts to deal with that?
CRAIG MUNDIE: In a way, I believe that the consumerization of information technology, which is well underway, will go a huge distance toward eliminating this, quote, digital divide. Even in the poorest places in the United States and elsewhere in the world people have televisions, and increasingly they have cell phones.
TIM O'REILLY: Oh, and, in fact, in some countries they've got -- yeah, I was just interested, some of the things you've got in Africa on the mobile world; it would be great if we had those services here.
CRAIG MUNDIE: Right. And so I think the combination of television, cell phones, low-cost personal computers, all of these things getting down into this price point of $100 to $200, which is really pretty much here today, is going to mean that there's no reason that almost any person in this society won't own the technology to get it, and those few that don't will be able to get it in some kind of shared access environment.
TIM O'REILLY: But what kinds of things should government be thinking about doing to move this along faster or should government just get out of the way like John Gulverson (ph) would like it to do?
CRAIG MUNDIE: I actually think of the problem in two separate halves. There's the problem of what can we do to use advanced technologies to make government services more efficient, more available. It doesn't matter whether you're looking at health care or education or almost any problem; government really has sort of three separate issues. They have the question of access, how do people get access. They have the question of outcomes, for what money goes in do we get a good result out. And they have the question of cost containment. And I think that information technologies in particular give us a new set of tools to address each of those three things in part, in any area we choose to apply them.
TIM O'REILLY: And the one I framed is actually we have such huge problems that if we can save money, we can spend it on other things that need doing more.
CRAIG MUNDIE: Right, but I think that is true in every one of these disciplines. So, for example, I mean, a lot of the discussion today around health care in this city is about access. And it doesn't really contemplate very much yet how technology might actually change the whole practice of medicine or dramatically lower the cost of delivering services.
TIM O'REILLY: But how do we do that? In the private sector if a company comes up with a killer idea, killer app, they're able to displace the competition. But governments are forever or at least a lot longer than companies. We don't have the same kind of creative destruction that we see in the marketplace. How do we actually get a better idea into government, if, in fact, there's huge entrenched interests that are keeping things the way they are?
CRAIG MUNDIE: Well, maybe I'll say the other half of the way I think about government is its accountability. To any society the government is ultimately accountable to the people.
And I think one of the other things that was mentioned this morning and people talk about quite a bit is in this open government sense where they seek transparency is to think of that as the way in which the government holds itself accountable to the people.
So, providing data and allowing people to analyze it is a way of expanding on what you might think was historically done just through the role of the media, and we now have a richer way to think about creating that accountability, that participation.
But the situation gets very difficult when you try to conflate those two things. So, in this conference I think it would be great if you could get people to say, look, I'm here to talk about services and how we can make them better, cheaper, more efficient, et cetera, or I'm here to talk about the accountability gap, and I think the way we approach them will be a little different.
TIM O'REILLY: But in terms of better services we are in some ways -- you could look at all the hackers who are building these outside government services, everything from EveryBlock or the MySociety stuff in the UK. In some sense people are building alternate interfaces to government services. And now the government is responding to that, what I would see as the heart of this gov 2.0 movement is they're saying, OK, wow, they've raised the bar. We've got to do some of that, too, and now you're seeing these innovative applications coming out of government, partly in response to the private sector or NGOs kind of using government data to create new services.
So, we are starting to see that interface as a fertile area for experimentation and innovation, and that comes because the government is starting now to say, okay, so and so did this cool application; how can we make it easier to do more of that kind of thing.
CRAIG MUNDIE: Well, I think -- and you and I have talked for many years about the idea that it's easy to talk about platforms, whether they're data platforms, computer platforms, programming platforms, but every one of these things that has really been big -- the personal computer, the Internet, the iPhone -- they all got launched on the back of some killer application. I mean, PCs got made on the back of word processing and spreadsheets. The Internet as a phenomenon was made by e-mail clients and the Web browser. The iPhone and platforms like that were made on the fact that it was just a great phone experience that integrated the browsing environment with the phone.
TIM O'REILLY: Well, yes and no, though. I mean, if you think about DOS, for example, existed before you had the killer apps.
CRAIG MUNDIE: But DOS --
TIM O'REILLY: You know, Windows drove forward -- you know, you had Windows before you had Excel.
CRAIG MUNDIE: But actually if you think back, I mean, my first experience with the PC was running Lotus 1-2-3 on DOS.
TIM O'REILLY: Yeah, and it was a great experience.
CRAIG MUNDIE: And I'm just saying, you know, it was the combination of these things that put power in the hands of the individual. Individuals made that purchase decision that they either wanted to do their own word processing or they wanted to create their own spreadsheets, which was a new way of modeling. Prior to that time, you had to go hire a programmer if you wanted to compute something or you had to know enough programming. And so it was an empowering capability.
TIM O'REILLY: So, the Web empowered people at a whole other level --
CRAIG MUNDIE: At a whole other level.
TIM O'REILLY: -- people who weren't even programmers could start to build applications.
CRAIG MUNDIE: And so the question is, with the new tools, the new models of man-machine interaction, a more natural way of interacting with computing, higher order ways of expressing programs or asking questions of programs, are these the tools that at least on the government efficiency side, the services side, really ought to be the focus in terms of improving efficiency.
TIM O'REILLY: But it also seems to me that following this metaphor that we're building of some kind of operating system for the Internet, just as on the PC you have a subsystem to control access to say the disk or the keyboard or the screen, now you have subsystems that control access to location or identity or various kinds of really databases. And some of those databases can be richly informed by government data. When we start thinking about location-based services, certainly there are a lot of commercial things -- oh, there's a Starbucks around the corner -- but there's also this huge outpouring of applications showing you EPA data relative to your location or civic information relative to your location.
How do you think about sort of the information infrastructure of that next generation of applications, and how do we make it better?
CRAIG MUNDIE: Well, I think one thing that's really interesting for government people to think about is that there is essentially a new government service, and that is the data service. So, not all government data is interesting. Some of it you could say, hey, if we published everything and you parsed it out and said, what would it be good for, some would be good for the accountability, some would be good for facilitating an improved UI to the old government services, and some would be data that people can say, hey, I'm just glad that that's available now because there's no other source for it, whether it's census data or like the comment this morning about the GPS system.
And I think that there is an opportunity again to think discretely about each of these things, and say, OK, you know, my job in the government is make my services that I've had more efficient or, hey, I've got a bunch of data and if I made it available, it would be valuable to the public, and so let's figure out how that gets used.
TIM O'REILLY: Well, I think that's a really important takeaway is think about the citizen as a consumer, you know, that effectively government is trying to make things better for the person who the government is supposedly here to serve.
Can we put the Twitter feed on the screens down front, and also if we could take questions from the audience. Do we have a mic out there?
But let's sort of continue this thread of services. Are there any particular areas where you look at, you know, whether it's health care, education, where you say, hey, the government out to be doing something more aggressively or ought not to be doing something more aggressively? Broadband: another issue.
CRAIG MUNDIE: Well, again, broadband is a place where government policy can affect the outcome in terms of how we improve broadband connectivity for the country. That's where I made the comment earlier that some of the things that have been changed, like the whitespace decision that happened on election day, I think time will tell but that that will probably create some at least modest revolution --
TIM O'REILLY: So, more unlicensed spectrum would be good.
CRAIG MUNDIE: More unlicensed spectrum I think will create ways to improve broadband connectivity.
You know, how the government thinks about the next generation of performance in the backbone of the Internet, and how that relates to the research institutions in the country, it's sort of like Internet 2 for the universities. But if you want to make that applicable to a broader array of services or where there's a public-private partnership between the government and its service functions and some of these other institutions, I think there are some very clever things that could be done there, and it will depend on government taking good actions in order to see those things emerge.
TIM O'REILLY: So, do you see any concerns? You know, we tend to think of broadband as wireline infrastructure, and yet so much of our future is wireless. Are we thinking enough about what's the infrastructure that's needed for wireless plus cloud as the dominant paradigm, as opposed to people being wired in their homes or in their schools or in their offices?
CRAIG MUNDIE: I think the global market for connectivity is going to drive a lot of invention in this space, that in the established countries we have both some fixed wire infrastructures, whether it's cable or traditional telephone infrastructures, and then we've supplemented them, driven by mobile telephony, to provide another form of broadband connectivity. And then in a local area sense we've put things like Wi-Fi somewhere in the middle.
I think that it will be possible to get more advanced wireless techniques than are currently being used in the mobile area in order to provide even higher data rates in a mobile or place where you don't have fixed infrastructure.
TIM O'REILLY: So, do we have any questions from the audience? If you want to be recognized, please stand up and Gina will come find you.
Also if you could make the Twitter feed bigger on the screen down there, we might be able to read it.
Do we have anybody who has a question for Craig? If not, we will keep going. All right.
CRAIG MUNDIE: We could read it back there.
TIM O'REILLY: Yeah, we could turn it around and read it back there.
So, lessons. Microsoft has been a dominant company for many years, now facing some pretty serious competition. We have first the Internet coming along, now Apple has sort of reinvented itself with the iPhone. How do you see the competitive landscape in technology shaping up? I'm not just talking about what's new in tech but how are companies doing out there? I mean, do we have enough competition in the marketplace?
CRAIG MUNDIE: I think there's a lot of flux on a global basis in the technology business, and I think that's true in both the software area, the communications area, and even the device area.
For many years, I mean, 17 years ago I went to Microsoft to start non-PC computing, right?
TIM O'REILLY: It took a lot longer than you thought, huh?
CRAIG MUNDIE: It takes a long time.
TIM O'REILLY: That, by the way, is a really important point I think everybody needs to be aware of in something like this government arena: Platforms take a long time to get adopted. I mean, how long has GPS been out there before we started seeing really ubiquitous location-based services?
CRAIG MUNDIE: Well, these fundamental technologies on which all these things emerge, none of them -- I mean, even though they sort of appear in the public eye like they came from nowhere, have required oftentimes 10 to 15 years worth of diligent development by many companies in order to make them happen.
You know, I talked about the parallel computing evolution that's going to happen. You know, I mean, we've had active programs with Intel and AMD and others for more than a decade getting warmed up for this.
So, while for the average programmer in the world or the beneficiary in terms of the consuming public who says, eh, I can't believe how fast my toaster oven computes now, that may suddenly appear but it will have been many, many years of effort to get there.
TIM O'REILLY: Let me ask you about decentralization. It seems to me that one of the aspects of the personal computer revolution that people have sort of forgotten was it was fundamentally a decentralized revolution: We put the power in the hands of individuals. And the Internet, also decentralized revolution: We put the power in the hands of individuals.
So, when we look at a problem like say the smart grid, are we thinking broadly enough?
The reason I'm thinking about this right now, I talked with Lin Wells from National Defense University last night, and he talked to me about this amazing program in Afghanistan where they're figuring out how to drop in these nodes with distributed hydro power, distributed cell phone coverage, so people can literally kind of set up this distributed power grid, and I was like, God, wouldn't it be great to have that here, just to kind of bootstrap past this entire monolithic infrastructure that we have, and work from the other end.
CRAIG MUNDIE: When I think of the energy problem, again I sort of divide it into two problems. One is long term, and forever there will always be the problem of conservation. The less you need to use, the better off you're going to be, no matter what your sources are. And then there's the longer term problem that says if the world ultimately has to migrate to an all zero carbon generation capability, is it going to come from some zero carbon replacement for the base load generation we have today like nuclear, some novel form of nuclear power, or are you going to go to a fully distributed grid, cogeneration kind of environment like they have.
I tend to think that we will get distributed action soonest on the conservation side, and that's something that everybody can participate in. Microsoft, we built this new product called Microsoft Hohm, and just launched it a few months ago, which is exactly that, a tool to allow each individual to do an analysis to develop a model in a simple way of their home, and then make individual decisions about how they can lower their consumption.
TIM O'REILLY: Absolutely. You know, that brings up such a critical point of how much in the way of Web 2.0 and I think gov 2.0 is about measurement, and so that people get a feedback loop and they can actually respond to the data that they're seeing.
CRAIG MUNDIE: This ability to close that loop where we have a huge amount of low cost sensing and data generation capability, and yet most people can't develop a model that says, you know, would I be better -- if I want to get green, and my family wants to conserve energy, what would I be better to do: double glaze my windows and replace my water heater, get a high efficiency furnace? I mean, there's a huge amount of information and modeling that actually for each individual house would allow that to be done. Many people have done that.
The Department of Energy spent years studying those problems. The question was, how do you put it forward? In our case we actually took the work from the Department of Energy, put a Web friendly face on it, put a cloud computing facility behind it, and said, OK, now anybody can answer some simple questions and end up feeding the data that they have into this model. They get their company's energy data, their energy provider data.
So, I think this is the kind of public-private partnership where you put an easy to use interface on stuff that's very complex, but you end up empowering individual action.
TIM O'REILLY: All right. So, that's ease of use empowers individual action. Good quote with which to end our discussion.
CRAIG MUNDIE: OK, great.
TIM O'REILLY: Thank you very much, Craig.
CRAIG MUNDIE: Great to see you.