Remarks by Bill Gates
Microsoft Research Faculty Summit
Redmond, Washington
July 29, 2002

RICK RASHID: So now I have the honor of introducing Bill Gates. So I was just a warm-up act. Next year I'll learn how to sing myself. But it's really difficult to actually do an introduction for Bill, because so many people know so much about him. I'll just say that he's our chairman, he's our chief software architect and he has an incredibly important role in shaping the future of our product in that role as chief software architect. And he also has a very deep passion for technology, passion for research and I know one of the things he enjoys is coming and speaking to this event.

So without further ado I'll introduce Bill and get out of the way.

(Applause.)

BILL GATES: Well, good morning and thanks for coming to Microsoft. You'll get a sense from us over these next couple of days that we're very excited about the decade ahead. We're very excited because we think that software and hardware during this decade can solve some of the tough problems that have been discussed during the past decade.

We call it the "Digital Decade" because we think the tools that will be created during this timeframe will be the best tools for empowerment and productivity the world has ever seen. Some of the dreams, like being able to communicate, sharing the screen, being able to get the information you care about when you want it, being able to secure information, allowing data between two different companies to be shared without all sorts of special code being written to map from one format to another; these things are achievable during this timeframe.

And over the last year some people have been a little bit concerned over what happened to the dot-com companies. Our message has been that we need to think long-term. We need to keep increasing our investments, investments in research as you heard from Rick. We're also increasing our investments in development as well, because we think this is the greatest era for software ever.

Now, part of this is the advances that come to us from the hardware industry. We're going to see devices, better devices than ever before. Some of the key assumptions we make include the pervasive availability of wireless networking, whether it's 802.11 or the things that that will evolve into, we see that being in place in universities, businesses, homes and so when we think about scenarios, scenarios for communicating, sharing photos, music, the whole way that software will locate the things that you're interested in, we keep that in mind.

And we think that computing will be much more about the individual and help the individual work across all these different devices than ever before.

The vision of the PC, although it got us to this point, it led to some really amazing things, it was somewhat centric on the individual systems and so we didn't have management software that could look across thousands of systems and make it easy to know what's going on. We didn't have software that automatically as you move from device to device would take the information that you care about and make sure that it showed up there.

The number of form factors we'll have will be quite diverse. I see Microsoft working on things like Xbox. Xbox brings a new level of video capability and over this next year one of the exciting things there is what we call Xbox Live, taking broadband connections, hooking people up so they can talk to each other, work with each other while they're playing fun games. So that's one extreme for us.

The other extreme is the work we do on high-end, high-performance servers, getting into the enterprise and saying that the scale-out type approach, where you use many machines and have software to do the partitioning and software to easily manage those multiple systems, that's a way to get a level of scalability that no system has ever delivered before.

So there's a lot of work to make sure all the pieces come together and all the tools are absolutely there.

We call our strategy to do this .NET. This is something we announced back in June 2000. This was our commitment to take XML, build it at the center of all our software and have a series of protocols around XML that it would allow for a rich information exchange, not just between our systems but between systems of any type.

It was really at the time kind of a surprising thing because some people had been saying, "Well, why don't we just rewrite all the software in a common language? Won't that do some nice things?" And our answer to that was no, there will continue to be language innovation, lots of languages and the idea of interoperation, the tough problems there to achieve applications like e-commerce really are much more about data representation.

And so the momentum that's been built around this approach over the last couple of years is very gratifying to us and it's fair to say that throughout both the academic and industrial environment the idea that XML and these rich protocols are going to be the key things, that's broadly accepted.

There's a lot of work going on now to put those standards in place, things like the reliable messaging, the transaction support, getting the protocol stacks factored in a way so that you can discover what another system has and then work with that system to do whatever type of transactions you want to carry out.

.NET is not an overnight thing. We find ourselves two years into it feeling very good about the direction, but it's another four or five years before all the promise of .NET really gets pulled together.

Of course, a key milestone for us was the shipment of Visual Studio .NET and the runtime, the common language runtime that works around that. That's really the kickoff and it's the reception to that has been very exciting for us.

Why are we pursuing this new architecture? Well, I think there are many reasons we would have chosen to go down this path. If you take four domains, information workers, IT, consumer, business processes, I think any one of the four would have said to us that this next step in distributed computing was absolutely necessary.

Take the IT department. They have a tough problem keeping all the software up to date, knowing which systems are working and not working, and it should be possible for them to say something like what kind of response times are people getting in e-mail for some corporate application and to literally say they want to probe that information and have that come back to them within a few hours. Well, it takes a very rich architecture to have that kind of schematization and be able to have those events go out, be gathered up, be put into a format so that you can browse the data.

These IT problems in a sense have gotten worse as they've had more and more service and more and more clients and nobody has stepped back and said, "OK, here's an architecture that can actually not only simplify that but make it as simple as it was back when you had very few systems." So tough problems like meta directory, monitoring, things like that, drive us towards the new architecture.

In the business process area, e-commerce, of course, is sort of the defining application. How do you do business with somebody where all the richness of the dialogue is encapsulated in the digital protocols, and you're doing business with somebody whose developer you never met and whose systems might not only be unreliable, but in some cases might actually be malicious, and yet you want your software to work perfectly even under that kind of an environment?

And people who've talked about e-commerce a little bit thought, well, maybe we can have middlemen there, these marketplace things, but that didn't work because it doesn't solve the tough problem of how do you find resources, how do you exchange rich information.

And so now we're back to doing the hard work, including the creation of a group called Web Services Interoperability, the WS-I group that companies like ourselves and IBM are very involved in, in getting the foundation laid there.

For knowledge workers this is probably in a sense the most mature area in these four domains. Microsoft Office is very popular, but we see that as just scratching the surface. The way that people exchange information, workflow type things, rich viewing, real time communication, being able to get information where they want it, all of those things are well short of the ideal that's possible.

And the benefit of making information workers more productive is quite phenomenal. The idea that they can have the latest software to empower them is something that's always gotten us very, very excited.

The final domain, the consumer domain is one of the toughest because the ease of use and the low prices that are necessary there make it a real challenge. But the idea that in the home environment you can get the music you want anywhere in the home, you can look at the family schedule, you can have very inexpensive screens that we just project information onto, we see this rich protocol environment as also making that possible.

So it's an ambitious agenda to have software that helps in each of those areas, and, of course, the boundaries between the areas benefit now that we have the one common architecture that's being used to address the issues in every one of those things.

Well, one of the key things is this idea of multiple devices, the pocket-sized device, the tablet-sized device, the large device, and so this demands very innovative thinking about the user interface. And a project that we're very excited about is one that has been done at the University of Maryland, and it's the idea of taking the .NET framework and doing some very, very innovative UI that operates across different devices.

So to show you some of the work that he's done I'd like to ask Ben Bederson to come on out and show us the great UI work that's come out of this.

Good morning, Ben.

So I want to show you this calendar interface that I've been developing for Pocket PC and it's running over here on this device. So our goal has been to support better planning and analysis tasks for calendar interfaces. So at the University of Maryland we developed this product called Date Lens. It's written in C#, and uses the .NET Compact Framework, and runs on Pocket PC, and there are other devices, as I'll show you.

So it uses visualization, simple interaction and fish-eye distortions to navigate and so you can quickly see how this works over here. We just have to go to different dates and move around like this.

We've got columns that show days of the week and in each row we have each different week, and then we have a little sort of Outlook-style visual representation of the individual days.

So let's take a look at my schedule, and this is my actual schedule. So we've looked around. We can see some kind of pattern going over there this week. So if I just drag the mouse we can get tool tips and see that I'm going to be on vacation in Windsor, Massachusetts, so that's nice.

But then when I come back we can see there are two columns here, and that actually represents a conflict, so we can zoom in and see that I'm having a meeting at the same time I'm suppose to be at the Human Centric Computing Conference.

But what I really want to do now is show you how we'll do some scheduling. So I want to try and schedule the human computer and traction labs colloquium series and at the same time make sure I avoid conflicting with the computer science department's colloquium series, which is also on Monday.

So the way we're going to do this is to search for a colloquium. So I'll just write "colloquium," search for that and it will highlight in yellow all of the days that have appointments that match colloquium, and so we can see here those two days, but I really want to see the rest of the semester. And so if we look on the scrollbar here, there are these little yellow marks and we can change the view so the marks are inside the current view, and we can also change the size of the scrollbar, and then we can see the rest of the semester's schedule, and now I can see the blank days are good days for scheduling our colloquium.

So searching is valuable, but text entry is a challenge, as you can see, sometimes, so we want to have an interface for searching without text. And a simple way to do that is to use a text that's already on the screen, and we can do that just by tapping on any date and clicking on any appointment and it will search for all days with that appointment and it puts exactly matching searches in yellow and inexactly matching searches in orange. So we can see I have a recurring appointment here for the ICDL Tech Meeting and over here we have ICDL Telecom and then there are some other things like you can look over here, ICDL launch, so that's our big party for the International Children's Digital Library Project.

You can also use pre-built searches. So, for instance, just where on my menu can we search for birthdays and then similarly it will do a free text search for "birthday" or "b-day" or "b-dash-day," and common spellings, and that's something that you'll be able to configure.

So this broad view is very nice for searching but you sometimes want to have more control of your view. So you can do that through these little check boxes along the sides. So, for instance, if I turn off the check box for October and September that allocates more space for the nearby month. Similarly, I can have less space allocated for weekends of if I want to schedule that camping trip then I can turn off weekdays.

We can control the view further through the scrollbar, as I showed you. We can go all the way down to small views. And, in fact, when you're in a small view then when you zoom into a day there's enough space to actually show a time-based view within the context of the rest of the system.

So the last point I want to make about Pocket PC is that it can be a little bit of fun. We found that artists sometimes paint to a grid and we can take those paintings and map them to the date lens grid. So we have a Mondrian skin that takes an arbitrary Mondrian painting, maps it to the calendar and then it continues to be a fully functional interacting calendar. So this is a kind of way to personalize your calendar.

So what fun! But does this thing really work? Well, we actually have some evidence that it does. We've done a collaborative study with Mary Czerwinski and George Robertson at Microsoft Research and done a benchmark study comparing it to the Pocket PC 2002 calendar and the results are very encouraging. There are some real advantages to the date lens and we'll talk about that study in detail this afternoon when we give a more detailed talk.

So the last thing I want to tell you about is how it scales to other devices. We designed this for a PDA but the design also scales for other devices. So we've got this Tablet PC device over here and running exactly the same code we can scale the interface down, and here it's many fewer pixels, about the size of a mobile phone, so I can have a month interface, an interface for navigating through a month's worth of calendar on a telephone.

Similarly, we can scale up to the Tablet PC and now exactly the same interface, right, now gives me enough room so I can see a six-month view and navigate around this way with all of the same kinds of features and interaction.

So this has been interesting, and especially because I want to say a little bit about my experience using this technology. I've been using Java to develop interfaces and visualizations for about five years, so I was a little bit skeptical when I started looking at this. But I have to admit our experience has really been positive. Between the language design and the APIs, our code is definitely a little smaller and leaner and it's completely a new idea that we can actually have the same source code running on both machines, so that's worked out very well.

So we are actively developing this and we hope to make it available others soon and I want to thank you for supporting our research and also a special thanks to Aaron Clamage, who's the developer who joined our team and who's just doing a great job with it.

BILL GATES: Well, that's fantastic to see what you've done with .NET, and that is a wonderful UI. I'd love to have it on my Pocket PC.

BEN BEDERSON: I'm sure we can arrange that.

BILL GATES: All right. Thanks very much.

BEN BEDERSON: Thank you very much.

(Applause.)

BILL GATES: One thing that I think is very important is the kind of relationship that we've had with software work going on in all the universities. Even when we started Microsoft we thought this would be important, but as we've gotten more successful we've been able to scale up these programs.

A key part of this is our own research group, and Rick talked about the progress there. I'm really proud of what's gone on there. A lot of the most fun parts of my job are going in and sitting down with our research people, hearing about what they're doing, and the various collaborations that they have.

So software is most interesting because of taking on tough challenges, things like Tablet PC, where it's a frontier that a lot of tough things have to be solved, these distributed computing problems, the tool development problems, and there's no doubt that the strength of the commercial software industry really comes because of the great work that goes on in the universities.

And so we're getting smarter about how we can work together all the time. I think an event like this one you'll give us a chance to tell us where we can do more of these things, but the key point is we see it as very much a virtuous cycle where we both benefit and that it leads to great software creation.

Well, during the last year what were some of the milestones in these relationships? Well, for us getting .NET out there was a very big thing and expanding a lot of the work we do where we take Microsoft source codes and make those available, particularly with Win CE, but also with broader licensing to the Windows source code we've done a lot more. We are doing a lot to use online communities as a way to get engaged down all the way to the mass of students and hear how they're using these things and what they think about them and then we've done more sponsored projects. Although these tend to be fairly small scale a lot of them come up with some really very interesting results. So the last year was one of growing activity in academic relations.

Visual Studio, one of the key philosophies behind it is that we don't want to be tied to a single computer language. We have some like Visual Basic, C#, that we think are incredibly important and so we'll put particularly energy into those. We have a lot of people who have been using C++, and so we're going to continue to support that.

But we see this era where we're getting rich heterogeneous data to be widely available and widely browse-able, that there's an amazing amount that should be done in language innovation and that will come in two ways. One is taking the existing languages and extending those and the other is coming in with new languages.

And so we wanted to make sure we had an environment that encouraged that kind of experimentation. To us the idea of somebody saying this is the last computer language and one entity ought to review any new things that go on in that, it's at least not appropriate to this era of software advances.

And so what we've seen as we've put out Visual Studio .NET and said come and use this and build on top of it and yet you don't have to build the new editor environment or a new framework for debug or a rich runtime, we've tried to make it flexible enough to accommodate all the different languages; a lot's come out of that. We have over 20 different languages hosted here and it has forced us to think about the runtime and some of the issues, how we do the typing systems and the performance and we've made extensions and I think those extensions have benefited not just the individual languages we've done them for but the whole idea of that universal runtime. I think a lot more is going to come out of this because of something that we're very enthused about and very committed to.

Now, part of the idea here is to have grassroots activity, just have very, very broad usage. We want to have our software really easy for you to get access to.

One of the programs we have is the MSDN Academic Alliance that allows for very modest fees for the department to use basically all of the different tools. We have 300 universities working with us under that program. It's something that we'd like to expand as well.

We also have a lot of things that we've put out that we're just making free. One that came out just about four weeks ago, and that I'm very impressed with, is what we all the ASP .NET Web Matrix, which lets people sit down and build these Web service applications really quite rapidly, and I think we're going to see very extensive use of that even in quite serious projects, and then the .NET Framework SDK itself is also something that we make free and available there.

All of these things require innovative approaches to make sure that we are making it as easy as we can to build on top of what we're doing, and to give us feedback about these things, so I hope there's a lot of discussion about the reaction that people have to these programs and where we ought to go with those.

We did, in one of the relationships where we put out the source code, have some cases where people did work that we then were able to license back in, and as we looked through that we decided that it would be great to give an award to somebody who's done the best work on that and, in fact, Lancaster University was chosen to receive that award. They took our Windows Embedded product, they saw some neat things there and they knew they wanted to work with it, and they found that one of the things that's important in their environment, which is IPv6 support, was missing. And so they actually took that, did a really great implementation of it, and then we worked with them to actually get that back in to be a feature of the product.

So that was a very much win-win situation for us, getting them to have the Win CE platform as one of the things that's there for them to do research with and to make our products even better.

And so the fact that the source is available and we have those programs, we're seeing concrete benefits come out of that and that's certainly encouraging us to push forward and do more things of that nature.

Community, as I mentioned, that's something that we see a lot of potential innovation to build community support into our products, make it so that when you have a question it's natural to go out to the community not just the local help file, to make it natural to be able to make a contribution.

There's a lot that's going on in terms of community software to encourage the people who make good contributions, to let you navigate around and find what the new things are, it's another area of hot advances that we in our collaborations want to make sure that we're driving that forward.

In the academic area we've got three communities that are worth mentioning: Dev Hood, which actually was a grassroots effort that grew up with students. It's a lot of students registered and over 70,000 messages up there ,even though that's a fairly new thing that's come out, very focused on .NET. Main Function is a very broad academic site that we created that's got curriculum and faculty exchange across all the different Microsoft tools. And then finally .NET is another environment that we've created where we've got hundreds of schools registered there and a lot of exchange there about different programs that are building.

Of course, we're going to cross-link the communities where that makes sense and drive this really to a new level. The idea of using online to help people, to get more feedback, that's something that we really see as being in its infancy but something we definitely have a serious long-term commitment to.

I talked about sponsored projects. Over the last year it was about $4.5 million for these things. We have a number of areas of focus. Mobility and wireless: That's a very hot area, whether it's higher-speed transports that are coming on or how you build software that you can have self-organizing networks, so-called mesh networks, we think something very important has to come out of that to get broadband accessibility to the point that we all want it to be at.

Security, I'm going to talk about that specifically. That's a big priority for Microsoft and a hot topic for our industry as a whole.

.NET is, of course, very important to us, the language extensions and the idea of Web services. The thing that Rich showed where Web services were applied in a particular domain, which was astronomy, to turn that into sort of a digital lab science where anybody with an idea has access to the information, we see that being spread to other domains as well and every new domain we get into we understand how to make the Web service generic platform that much richer. It's something that not only all the different areas of science, even tough problems like biology, but also different areas of commerce, the specific Web service innovation should be pulled together.

Learning systems, that's a very big deal for us. In many ways what can be done in the universities to improve learning, there's a very direct analogy in the business world for those information workers who are sitting in meetings and being trained, and so any advances that help in academia we think we can get into commercial products and improve productivity in the business world as well.

And finally embedded systems, getting software into an unbelievable variety of devices, whether it's the car and some neat things going on there to medical systems and even areas that wouldn't be expected. We have had over a hundred different projects and very good work that's come out of those.

One that I think is very cool and there's a lot that we can do with this now that we've got it kicked off is something called Conference XP and we're going to give you a little demo of this. I'd like to ask Jay Beavers, one of the people from Microsoft Research who's involved in this, to come out and tell us about it and see if we can put it to work.

JAY BEAVERS: Hi, Bill.

BILL GATES: Morning, Jay.

JAY BEAVERS: Good morning.

Conference XP is one of the sponsored projects that you were talking about that we're doing with a number of our partner universities. We're working on creating a high quality, low latency videoconferencing system that works over the Internet too for use in distance education.

We're able to do some pretty cool things today because of the advances that have been made in commodity PCs lately. We have gigahertz class CPUs, new graphics processors like the NVIDIA G-Force. And we're developing this as a research platform for other people to take it and do innovative things with conferencing and distance education in general.

So we've built it as a series of layered APIs. We start, of course, with the .NET framework, which provides all the base functionality. On top of that we've added a network API based on RTP, or the Real Time Protocol, that allows you to create interactive applications over the network pretty easily. An example of that, which you'll see, is the distributed PowerPoint work that the University of Washington has done, here later today.

On top of that, we've added a conferencing API so that you can create new UIs and adapt them to your environment. An example you'll also see today is that we've integrated in with Windows Messenger, so we use Windows Messenger to start up the classroom.

And then finally we have an extensible UI on top of that, which we call the Edge, which allows us to integrate other learning tools into the environment so you could add feedback systems, et cetera.

So what I would like to do is set up a conversation using the system. We've got a parallel system deployed here to what we used with the University of Washington in their masters class this last spring set up in the convention center and Dr. Lazowska is hosting a class on Tablet PC development that he's inviting Bill to attend as a guest speaker from offsite.

So he's sent us an instant message here that we'll use to accept and join the classroom. And the system starts up here, and after a few seconds for our key frame to hit.

DR. ED LAZOWSKA: Hey, Bill, thank you for joining our class today.

BILL GATES: Hi, Ed.

DR. ED LAZOWSKA: So let me give you a bit of background. Some universities, including Brown and the University of Washington, have been working with Randy Hinrichs' Learning in Science and Technology Group on a set of Tablet PC learning technologies, both developing them and trying them in class, et cetera. So this low latency Internet conference that Jay is showing you is an example. There are a set of sort of teacher experience applications such as the (Virtual IR ?) you hopefully you see up on your display, built by Richard Anderson from the University of Washington. There are a set of student experience applications that Andy van Dam and the students at Brown are building.

So in our test PC development class today we're talking about ink realism and what I'm interested in is what you could tell us a bit about the steps you've taken to achieve ink realism in the Tablet PC and if you'll wait a sec I'll take some notes here.

BILL GATES: All right, great.

Well, the key to Tablet PC is making ink really rich, something that you like to work with. It's not based on the idea that you always recognize it and turn it into text. We leave it as ink in a lot of places. We've insisted, though, all the tablet platforms have a digitizer that samples the pen position 133 times a second. It can actually sense the pen even when the pen is hovering above the service and you get both pressure and angle information. And so as you do the drawing you find that it's very realistic. In fact, you end up loving the ink, you almost never want to convert it and there's a lot that people are going to want to do in applications to take advantage of this.

DR. ED LAZOWSKA: Well, it certainly looks like my handwriting. That's bad for my students, I'm afraid.

STEVE WOLFMAN: So, Bill, this is Steve Wolfman from the University of Washington. I was wondering what sort of tasks do you think will really benefit from high quality ink?

BILL GATES: Well, what we're seeing right now is that in every different application the idea of annotation is very important. You can annotate schedules, documents, Web pages and it should just be a natural thing that you'd have annotations that you'd keep to yourself, for example, if you're sitting in a presentation you may not want your notes to go out to everyone else, or you have annotations that you share to a subset of those people.

We're also seeing people who do user interface and drawing programs really able to rethink some of the things that have gone on. We've been using the mouse for so long and the mouse is very limiting compared to the pen. And so there are a lot of new UI approaches.

Here at Microsoft, I'm considered a pro-gesture person. Some people think gestures, there's a limit to what you can do with them, but I think we'll have different gestures and those will grow up in individual applications and then we'll really coalesce that around system wide gestures that make the system easier to work with.

DR. ED LAZOWSKA: Very interesting. I think lots of those are really core applications for education, so I think all of us think of the education market as really a sort of killer app for the tablet.

Bill, thanks very much for joining us. I see you've got 325 people to talk to some more, so we appreciate your time and we'll talk to you later.

BILL GATES: Thanks, Ed.

JAY BEAVERS: Thanks.

So I'd like to announce that we've got the binaries available today on the CDs that you received with your package, so you can try out the technology yourself. We've also launched the community site on www.conferencexp.net that you can go ahead and join and get some more detailed information about what we're doing.

We'll have binary publicly available for download from the Web on August 15th as well as source code available under research and educational use license on September 1st.

We're not just working on conferencing in the learning sciences. We're trying to put together a platform for people to do learning science research. So you can see on the screen that we've got a number of areas we're investigating.

Some of the projects we're doing along with our partners are looking at the student experience, assuming the students have a Tablet PC, what can we do differently in the classroom, the professor experience, some of which you've seen with the distributed PowerPoint with inking annotation and various notes and feedback capabilities, as well as the TA or the mentor experience.

There's a lot more detail, of course, to talk about in these areas so please feel free to come by our breakout sessions or see us in the demo fest and I'd like to invite you to join us, along with our existing partner universities, such as the University of Washington, Brown, Carnegie Mellon and Berkeley, in doing more research in the learning sciences area.

Thank you.

BILL GATES: Thanks, Jay. It looks good. (Applause.)

Now, that to me is a very exciting application because it takes the high-speed bandwidth of Internet 2 and new tools like the tablet as far as what can we do with the learning experience, and I don't think anybody really has the answer there but it's a hot area for experimentation that I'd love to see some neat work come out of that.

Really the goals we have to have mutually are to solve some very tough problems, making computers a lot easier to use, so that both the things you do with them today are more efficient, more predictable and so that you're willing to do lots more with the machines.

Part of that has to do with conquering the challenges of natural interface. These challenges have been out there for many decades, but I think we can actually be optimistic that this is the decade where we'll make the most progress on these things and actually bring them into mainstream use.

Certainly with handwriting and gesturing, we see that being very soon with the launch of the tablet, taking place in November, and with speech we also see it coming into the mainstream fairly rapidly. We're getting microphones into all of the different machines and, whether it's voice annotation, real time communication or actual command recognition, we think that will be a standard part of the PC experience.

Vision, speech synthesis, those are also super important things that with the power that we get and with the new software approaches there ought to be some great progress there.

So quality, distributed computing, the really tough problems are all extremely relevant now. These aren't just problems that it would be fun to solve. Actually solving them, any kind of progress will make a big difference in terms of overall productivity.

The final one I've got listed here is security and you probably heard that over the last year we've really taken our development teams and asked them to take many months of time and just focus in on the security question, what is the threat model, where is it in the design in the code that there are potential exploits. We look at the code robustness in making more than an order of magnitude improvement there. We look at propagating updates so that when any problem is found the solution to that propagates faster than the exploit and finally issues of recovery, if there are any problems with these systems exactly how can they be up and running in a few minutes time without the opaqueness and confusion that results today when systems aren't doing what people expect.

This is so important to us that we've decided to put together a very focused effort around the overall Trustworthy Computing effort; very broad goals here. It's not just the security but also issues around privacy, reliability, the integrity of the different systems and how they work.

To us this is probably the toughest problem that stands between where we are today and people taking full advantage of what's going to be possible as the Web services type capabilities get out there.

There's no single solution that's going to bring this to an end, so we often say it's a journey not a destination. It is a journey that we want to make sure our partnership with academia helps us on in a fairly deep way. A lot of universities are thinking about research and so we're creating a board that's going to in particular focus on these trustworthy computing issues and this will be a board with about 15 people. It's just being announced today with the initial four or five members. It is something that people are particularly interested in the topic. We'd love to hear from you and talk about in what way we can work together on the topic. It's definitely an area where some advances are required.

The idea of the board is pretty straightforward: Feedback, review, making sure we're aware of the latest techniques, making sure that the way we're thinking about the tradeoffs in these areas that we've got all the best thinking available to understand what should go on there.

The initial board, I'm very pleased with the people we've got on that, very fun group, so I'm sure it will be an enjoyable board to be a part of.

So where does this leave us? Well, it leaves us here at early in the digital decade with some very ambitious goals about software, software really being the magic element that can take all those hardware advances and build things that are really cool and fun and make people productive.

There's a real reason why from the beginning we decided we wanted to focus on software. We see this as not a set of overnight features but a long-term quest and in that the fundamental research that all of you are so involved in will be a key element that really drives the advances.

So we've got a long-term commitment to working with you and I'm sure the event this week is going to help us manage that and work with you in an even better way.

Thank you.

(Applause.)