COMDEX User Groups

Mr. Gates: Thank you.

Good evening. Well, it's great to be here. The whole PC movement is based on people like yourselves that really believe in it, and help others to get as much value out of the PC as possible. It's been a long time since Microsoft started, almost 25 years now, and yet the goal that we've had hasn't changed during that time.

The original vision that Paul and I wrote down when we started the company was a computer on every desk and in every home. I think we can say that we're at least half way there. About half of the U.S. homes have PCs today. And, of course, in business the penetration is substantially higher than that.

However, as we look at the impact of the PC and what it's done to date, we find that we are nowhere near achieving the PC’s true potential. And it's fantastic that people are probably more excited about the PC as a tool, whether it's in education, entertainment, or in business, they're more enthused today than ever about the possibilities of this technology.

It's amazing to think that Microsoft has grown to be a company with over 20,000 people. Every year when people ask me about the potential for Microsoft and how much is it going to grow, I would say that at most it would double in size. And so when we had 1,000 people I said, you know, at most 2,000 people. I always thought, you know, how much software can you write? You know, it doesn't take that many guys to write a lot of software. But, as the industry has grown, and as the scope of the things we need to do in terms of supporting users and connecting to other systems has really gone beyond anything we would have imagined, we continue to grow to meet our customers’ needs.

Now, part of the dream that we had about creating the new industry structure really came to pass. The structure of the industry before the PC came along, and before we promoted the new model, was one where companies were oriented towards very low volumes, very vertical structures, and all the pieces of a solution came from a single company. That had a certain benefit to it, because you knew that the pieces would fit together. But it wasn't something that was going to ever be able to deliver a product that was in the range that an individual could afford, nor could it build up the range of software applications that would make the PC relevant to a very broad audience. So, the structure that we really invented was one where we created a layer that was standard across all the different hardware, and allowed tremendous advances in software and hardware to take place independently. In turn this allows you to pick the hardware from whomever you wanted to. It allows you to pick from a wide range of applications. And the high has really bred an incredible success loop there, leading us to the point where we are today.

Last week I was playing around with a PC that cost less than $500 and, you know, it had a two gig drive, 32K in memory, V56 modem. Okay, it wasn't the world's best machine, but it wasn't such a bad machine. Even to me, where we've been at the center of creating this structure and partnering up with chip companies, and peripheral companies, system companies and software companies, it's kind of amazing to see what has come out of all of that. You know, it’s amazing that a machine with very little compromise that is incredibly affordable will continue to grow the market, not only in the United States but in the rest of the world where the household penetration is still incredibly small.

We're on the verge of some very major advances. We've got critical mass now in terms of the Internet protocols and standards. Certainly, I think the idea of connecting up to other devices, what we call personal companions, is going to really accelerate. Here at Comdex you're seeing a lot of the companies that have different ideas about those form factors. You're seeing the early stages of wireless networking. That's a piece that may take a few more years to come into place. But overall the message is clear: for a very low cost, people will have the ability to update their data on all of these machines.

I have no doubt that the hardware industry is going to do its part. In a meeting with Intel and AMD and the others in the last week, it’s clear they've got road maps that get to processors that are over 1000 megahertz, processors with 64-bit architecture. So the Moore's Law doubling of power effect we’ve seen over the past two decades certainly isn't going to slow down in the foreseeable future.

The disk drive guys are going even faster. If you can find a machine with a disk as small as two gigs, it's going to be a collector's item in a couple of years. It will be hard to buy a machine that doesn't have 10 gigs. And what are we going to do with it? Well, that's where we take on digital photography. That's where we also take on digital video, digital audio, and make those commonplace data types that are built into the system.

The hardware guys will be delivering things like USB connected cameras. They will be less than $50 for video capture, video conferencing, and even making that functionality part of the user interface. They'll be delivering hand-scanners and digital cameras that will bring a lot more data into the PC. And they'll be delivering flat screens that have prices comparable with CRT prices, and yet in terms of the portability and the readability, go way beyond the CRT.

I hope some of you have had a chance to go by our booth and see a breakthrough we made which I recently called ClearType, which takes advantage of the property of the color LCD to get much, much better text resolution. By tripling that resolution, we can make an LCD color screen go from something where you can just barely tolerate reading email and documents that are a couple screens in length, to something where you would even be glad to read hundreds of pages in front of that screen for eight or ten hours a day. It really crosses the threshold to change the behavior and use of these devices.

A long time ago, people talked about using PCs to do presentations, but they didn't really ever do it. But today, think about the last time you saw somebody with slides. You know, the PC is totally accepted for that function. But today you don't talk about taking a PC to a meeting, some people do, but it's not a mainstream behavior. I'd say three or four years from now, that will have changed completely. Instead of taking paper notes in a meeting, you'll take your PC, annotate on that device, and then you won't have that disconnect between the information we have on paper and the information we've got there on the screen.

On the hardware side, there's a very, very bright future. One the software side, I'd say the same thing. The theme for Microsoft is absolutely bringing in the extra power while delivering better simplicity. As we look back at how much people have yet to learn to use the PC, we see what all of you are seeing, I'm sure, every time you help people use these machines -- there are a lot of ways that we can take the extra power and simplify the experience.

The experience when there's a problem is very opaque today. The experience of how many commands and utilities you have to learn is simply a lot more than most people have time for. Fortunately, the world of software is the world of rapid advances. So, you'll see in the next versions of Windows that we've really taken this priority to heart. We're really rewarding people inside Microsoft not for creating new commands, but for getting rid of commands; not for creating new file formats but by merging the existing ones we have until we get to one really rich store that you can navigate for everything.

So even if you want to query and say, for instance, what connections do I have? What fonts do I have? What users? Who are my contacts? All of that is done one simple way, and it's more like web browsing -- in fact richer than web browsing -- but it seems to be an evolution of the way web browsing works, and it combines the local information we have of all types and the remote information into a very seamless experience.

Being able to share data between multiple devices, being able to migrate easily between devices, all of those become very important as we're moving away from the idea of having only one information tool, just your PC, to having multiple PCs and the personal companions as well.

So, there's tons of work for Microsoft to do. We've increased R&D very rapidly over the years because some of these challenges like speech, handwriting and visual recognition, are areas where creating the software, refining it and testing it, requires quite a scale of activities to pull it together.

Fortunately, because of the volumes involved, we can take things that we've spent billions on, and price them for well under $100. And that magic of more powerful, yet cheaper will continue to drive this market forward. We're seeing the volume increase at every layer, at the Windows layer, the Office layer, and for those of you who get involved in servers, our BackOffice layer is actually our fastest growing business as well.

I'd say there are few but still notable things to be concerned about as we look forward. In my speech on Sunday night I talked about the concerns about privacy that we're hearing more and more about. And there is a lot of software effort needed there to fix that. Certainly, with these tools being as fantastic as they are, we want to make sure that either through schools or libraries that somehow people have access to them, so that everybody can benefit, almost like with literacy and libraries, where it's simply there for anybody who's interested. And we also have a lot of tough things to work through in terms of getting everybody hooked up at high speeds. Businesses are getting high-speed connections. Anybody who can afford $1,000 per month is going to have an almost infinite speed connection, and the ability to do video and audio at great distances. But, in the home, where the pricing has to be more on the order of $10 or $20 a month, the opportunity to get the high-speed connections is going very slowly.

Cable modems are the only bright spot there, and even that, with full speed deployment it will take three years before we'll get up to a couple million of those out there. So if you take it as a percentage of homes with PCs, or a percentage of homes using the Internet, or a percentage of total homes, it's still a fairly small number. And yet, the picture there is actually stronger in the United States than it is in most other countries. So that's a tough one, and yet we really need those high speed connections to fulfill the full promise of making this the new way that we communicate, superceding TV as we know it today, and finally essentially getting away from the use of paper at all, where the information you care about it immediately brought down to you and it's brought through systems that are very, very convenient systems that you trust to use your data to collaborate with people, whether it's a home type activity, or doing any kind of work.

So it's a very bright picture, it's certainly been great to be involved in all of this. The next 25 years really should see the fulfillment of any vision that people have had in this industry. I think 10 years we'll probably get most of it. But, you know, sometimes there are a few pieces that lag. One piece that lagged for a long time was connecting computers together. We talked a lot about it, and lightning struck. It was actually the standards coming out of the universities that allowed that to get to critical mass. So all that vision of information at your fingertips happened, and it happened suddenly. It wasn't a linear type thing, where year by year it's gotten greater. It was a discontinuity, and that's what you see when you have communications like technologies. That same discontinuity is driving across the economy, not just in the United States, but everywhere, where businesses will have to be on electronic mail and paperless in order to compete. They'll have to offer their products to customers through the Internet. They'll have to work with partners in that way. And actually today, businesses are clamoring to find out how they can get ahead of this, no matter what type of business they're in.

Colleges and high schools are also looking to figure out what this all means for their worlds? They're all about delivering information, satisfying people's curiosity, helping them certify their skills, and as we move into this digital age, all of those things will change as well. The availability of the best lecture, the availability of the best tests, that will be commonplace. The ability to share rich presentations on any kind of subject, connect up with people who have answers, or share some of those same questions will come through all of these systems.

Now, through all this, that basic insight of making the PC a personal tool, with high volumes, incredibly affordable, and simple enough for most people to use -- that's what we have to retain. The vision of Microsoft being very specialized, focused on the high volume building blocks of software, that's something we think is very important to retain. Given the success we've had and the size we've had, we're constantly asked when are you going into the system business, or the consulting business, or the network business, or even these high end enterprise applications? And our answer has been quite consistent that in the areas we operate, Windows, Office, BackOffice, a broad range of software, but rich, high volume software, we see enough challenges and opportunities that even if you look out 5 and 10 years, that's what will really characterize Microsoft.

We're delivering some of that now, more and more through services out on the Internet. And, in fact, the boundary between how you get support services will blur as you're downloading some of the software on a pretty regular basis over the Internet. You're starting to see the very beginnings of that in a couple of our products. Web TV is a product, when you dial in it will automatically update anything that's changed on the client. So new features just show up. Windows Update, of course, is an entry in Windows 98 that lets you get out there and see if there are new drivers or fixes, and we're getting a lot of feedback. It's a very new thing. But, it's really meant to be a constant service that will keep your Windows machine up to date, and help you diagnose what is going on. And it will allow us to stay in touch with customers and see where they are having problems and immediately when we solve it for one person, we can solve that for everyone.

So it's the most revolutionary industry there's ever been, doing a better job for customers than any industry ever has. And yet, throughout all that rapid innovation and change, the basic principles, such as open PC standards, focusing on the pieces that you're really good at, and driving that power while maintaining the simplicity is truly important. That's what this business is all about, and it makes it really one of the most fun businesses in the world.

You're role will be ever more important, as we conquer these frontiers. There will be a lot of people who are scared about what the PC means, a lot of people who want to dive in and push it to the limit. Fundamentally it's very exciting, and it's very positive. But change at this pace is really unheard of. Everything else that's been as impactful as the PC has moved out into society broadly over a period of several generations. And so anybody who thought of the car as coming along and being a strange thing, well, during their lifetime that was the case.

But here with the PC, the change is so much more rapid that in the course of someone's career you can go from the PC having been irrelevant to being absolutely an essential tool for them to be able to get their job done. And so this idea of reaching out, answering people's questions, helping them get excited, helping them help their friends with all of this, it's a very, very important thing, and I applaud you for all the energy you put into that. And certainly we're going to continue to do what we can to support you in that endeavor, because it's very important to us.

Thank you.

(Applause.)

Question: (Inaudible.)

(Applause.)

Question: (Inaudible.)

Mr. Gates: Well, there are things that used to be hard with the PC that are no longer hard, and you don't see anything about six-side bonds versus variable bonds. Pretty much printer drivers come with the operating system and not the application. Enhanced memory versus extended memory, for those of you who have been around for a long time. And, you know, even things like config.sys, although it's not completely buried, are no longer as visible as they used to be.

So, as we solve these challenges, people want to do more with the PC. It's the most general-purpose device there has ever been. And, all those nice new things, photography, music, transactions, coordinating schedules with other people, connecting up to the companion devices, every one of those certainly for the next decade will arrive in a way that will be absolutely perfect. They will be concepts that people have to be exposed to. You know, someday we may get to the point where the PC is doing everything we want it to do. And then, you know, we can say it's like the car. But, you know, that is not any time in the foreseeable future. It's not -- and certainly not in the next decade.

You know, the PC market has changed, and it's become larger, the basis of competition has become so intense that the margins are a lot lower, particularly on the hardware side, which is where most of the revenue is. You know, the hardware piece of it, as intensely competitive as that is, is still about eight times what the software side is. And so, a little bit of that has taken some of the marketing budget and budget for activities with user groups out of the picture, which is definitely an unfortunate thing. But I don't see that there's any radical change that will change the need for this sort of thing.

You will see more variety of users. The range from that hardcore user who wants to reassemble the motherboard with a different part, all the way down to the person who can't seem to get comfortable with electronic mail and URLs. And that's got to be tough for user groups to know where, you know, you hit an audience that everybody who is sitting there will either find it interesting, but it's still not over their heads. So, that diversity has got to lead to more specialization and creative ways of dealing with that.

But, absolutely, the PC will -- the fact that it's needed this kind of activity isn't changing.

Question: (Inaudible.)

Question: (Inaudible.)

Mr. Gates: I mean, we've added the ability to have a single worldwide binary for the Office products, and for the NT products. And so, you can create documents that use any type of thing. You can run Japanese software, you can buy a machine here in the United States, use the language quiz, run the software on it. We've done something that's never been done in an operating system before in terms of the broad multilingual support. And certainly we support the English products in Japan, we sell them in Japan through the different vendors. So, you know, maybe we should follow-up and understand what more we need to do.

Question: (Inaudible.)

Mr. Gates: We have a lot of different BUSes and connectors. And in some ways we're going to have to focus in on a few. Right now I'd say it's kind of unclear. In the home, you know, 1394 maybe a standard for connecting consumer products. But, some of the companies that were involved in creating it are charging huge patent fees. So that could hold that back. USB has a lot of great capabilities, in fact, we can take USBs to much higher speeds than we have today. And so that's a contender for connecting digital peripherals in the home. Then, of course, you have Ethernet, where you decide that everything is just, you know, an IP address network device. And Ethernet has the beauty of being high volume and fairly low price, but it doesn't have some of the features, like self discovery, today that the other BUSes have. We want to hide the difference of these BUSes, but also working with partners like Compaq and Intel, we want to really pick which ones we get the peripheral vendors to focus on.

In addition, there are things like various wireless ways of sending data inside the home. There's an idea of sending data across the AC power line, so you can just plug in a computer and the data comes across that way. That actually works amazingly well. And in terms of not having to run wires between various rooms, I think will be one of the approaches that are important there.

So with 1394, Microsoft has been very much behind it. We've gotten the consumer electronics companies to do things with it. But, the relative role of that BUS versus USB, versus Ethernet, I'd say is a bit up in the air. And I hope in the next six months we can really clarify that, so that the people who are investing in peripherals know which way we want them to go.

Question: (Inaudible.)

Mr. Gates: Well, if you believe in what we've got here as the communication tool of the future, then you do want everybody to be drawn in, and you know, have a chance to use it, whether it's to learn or to qualify for the jobs that involve these skills. I certainly believe it's true, and that's why in philanthropy a lot of things I've done have been focused on access to computer technology. My wife and I created a foundation that has a very simple goal, and that is to give to every library in the country PCs and software and support, so that anybody who can get to that library can get access. And we're pretty far along with that.

The first state we went into was Alabama. And so we picked that, because we thought, you know, we went to where the lowest percentage of households had PCs, because that's where the universal access can have the biggest impact. And then we picked six additional states that we'll get done in the next six months. So within the next three years all 16,000 libraries throughout the entire country, and in Canada, will get the PCs and the hookups.

(Applause.)

Mr. Gates: The reason I'm excited about that is the libraries, you know, they're a universal institution, and the librarians have been very enthused about making it happen. There still is a tough problem for some of the rural libraries, how much it costs to get a decent line into the Internet. And so we're working with them on that problem. But, everything else has gone incredibly well there. We'd like to see the same thing happen in schools, in terms of broad use, and there's a lot of people looking at that, and how we might be able to accelerate it. So I agree that access is, you know, that's the -- you take the good of all this technology, and you say why shouldn't everyone benefit from it.

Question: (Inaudible.)

Question: (Inaudible.)

Mr. Gates: Well, certainly the effect of technology has been very positive in terms of spreading access to information, giving people a chance to learn things in better ways. There are things that we knew hundreds of years ago that we don't know today. And, you know, you can assign your own moral value to the fact that we don't know how to plant seeds and grow food -- at least I don't, and probably most people don't. So, there is, over time, a shift in what things are important to have as skill sets. I think kids are writing more when they use email than before they had electronic mail. Inside Microsoft, you have to have the eloquence of putting your ideas in written form, because it's not just meetings, it's not just phone calls, it's written communication. And the same way as it was in the days before the telephone, you have a written record there of how people were discussing issues, and how they're going through things. So, I don't see anything that moves away from writing. In fact, I see an increase in writing.

In terms of handwriting, okay, that's unfortunate, and it's partly because the computer doesn't recognize handwriting. We will get it so it does. We'll even be able to read somebody's handwriting in terms of how legible it actually is. You know, you've got to motivate people to learn things because they're interested in those things. And getting people to be curious about the underlying mechanism, why do you use mathematics in different ways, what problems would it solve, that's the challenge.

The fact that the computer can actually do the multiplication is not really a problem if you've motivated the person to be interested in the fundamental subject area. And so as long as they are reading and writing and even using the computer to reach out to new things, it's definitely a huge positive from what it can mean. You know, I know when I go into classrooms that are using computers in neat new ways, and I've talked to the kids, I've never come across kids saying, boy I don't have to learn anything because of the computer.

But, you know, there's no new technology that's not without its controversies. There's a general mind-set, if you think, in general, technology has been making the world worse over the last 100 years, that the phone, the car, that these things made the world worse, then you'll probably feel the same way about the computers, because it's the same directional vector.

My view is that if computers were used only for one thing, just for medical research alone, and not used for anything else, that we could justify everything that goes on in computing, because the miracles we're going to have in terms of curing various diseases, that alone is a fantastic contribution to the quality of life here. And, you know, so I definitely see computers in so many areas as a positive thing.

(Applause.)

Question: (Inaudible.)

Mr. Gates: Well, the most dramatic thing will be the virtual elimination of paper and use of tablet-based computers on a pervasive basis, and use of natural input techniques including handwriting and speech recognition. And so, the effect is that the device, the way you think of the device and use the device, will be radically altered. In fact, the design will be the classic PC industry design. You know, those will be the companies that are putting it forth, and will actually have upwards compatibility as we get into those things that lets us use the investments that we've made in the past.

But those things, as they usher in PCs in different form factors, and usher in greater and greater use of the Internet to exchange information, those will be the things where people will say, okay, the computer as I used to think of it, I knew I didn't think that would change the world, but not that you've got them in this form, it's obvious that they should be used in a very, very broad fashion.

And because those are discontinuities, you know, there's a point at which that screen computer is cheap enough for everybody to want to use it, and the readability is there, or the speech recognition gets there, it's not something where you can say, okay, the market is so big today and it will be twice as big next year. It will be something that will surprise people, just like the Internet did, because when the pieces are inadequate it doesn't gel together and develop the momentum. But I feel confident, certainly in the next ten years, and probably even in the next five, those will change and become the way we think of the PC.

Question: (Inaudible.)

Mr. Gates: Well, we definitely need to have a scalable interface where people aren't exposed more than they need to. Now, in some cases, like the file system, we're exposing people to things that even experts don't need to see. You know, for example, if you have Word why shouldn't you just see Word, all of that code, as a single file, and think of it as one thing you can remove or install, rather than being sprayed in various directories all over the hard disk.

We do have this open structure where manufacturers can put in a variety of utilities, and third parties come along with device drivers. And it is a challenge to us, one we have to meet, but to have the best of both worlds, that is to have the openness where anybody can come up with a device driver, and yet have some way of certifying those things, and getting their interface to fit into a framework that makes it all pretty seamless, and particularly when you have people like the graphic centers, who are trying to do better on this related benchmark. You know, they're revving their drivers, and not taking the time to go through test cycles, and do the user interface. So, we're definitely going to have to come up with a way of making sure all the pieces that go into the system integrate as a whole without choking off the flexibility that the PC architecture has had in a very unique way.

The multilevel effect is definitely something to keep looking at. We tried that in Office, and what we found is that although nobody uses all the commands, the subset of commands they use varies quite a bit. And so you get this discontinuity that if there is this one feature that's not in your set, then you have to change the mode, and you're not used to what's there. And your friend who knows a little bit more doesn't understand your subset and what you have. And so, you get a "which does the book talk about" scenario, the Windows Level 3, Windows Level 2, Windows Level 1?

But there are some ways of making it so that aren't discontinuities there. That it's scalable in a fairly straightforward fashion. And I definitely think that's part of the way forward is to figure out the things that the entry level users need to see, and although we'll let them get to those other things, we don't force them in their face.

Most of the problems people have with the PCs aren't necessarily they got up in some commands and messed things up, it's more that there's a conflict between the various things they have on the system. Those are the toughest problems to solve.

Question: (Inaudible.)

Mr. Gates: Well, if we want to make that donation, you know, then we will make the donation. It's -- I mean, it's not something where we just say, oh it's -- you decide, you know, who should have free software. There has to be a framework there that doesn't eliminate the value of the software altogether. So, you know, certainly we give a lot of software away to charities. And if the issue here is the system software, we can talk with people about it.

I do think, you know, these recycle programs do -- you know, there is a lot of value to them. It is tricky, though, because you have to look very carefully. When you can buy a PC for $399 that really works, you know, versus a keyboard that's four years old, and a hard disk that people were buying three or four years ago, in some cases, you know, you have to decide which is -- in terms of is -- you know, take people's time. People's time is fairly valuable. And so in some cases, like in schools, it's kind of a shame for them not to have access to the latest educational software, all of which would run on that $399 machine. So you hope that over time people have access, and they're not essentially second class in terms of what they're able to do.

If you go back, three and four year old machines, it is tough the expertise of solving problems on those machines is not as wide spread as it used to be. But, for the system software we'll come up with a very clear framework for what goes on there, in terms of these recycled machines.

(Applause.)

Question: (Inaudible.)

Mr. Gates: Well, everything we do is based on instrumenting Windows and taking user feedback on what people are interested in. So when we create a product, let's say, Microsoft Project, it sells because people want it and use it. And we have literally tens of thousands of copies where we instrument what commands are they using, what commands are they not using, what errors do they get. So there's a very tight feedback loop that drives everything that we're doing there. You can go through and look at, you know, what have we added to the word processor, do people use spell checking, do they use the integration of the email and the word processor, do they use tables, do they use drawings, do they use, you know, any of these new features? The answer is absolutely they do. You know, people who design products are very serious people, who are coming in every day and designing these things with these capabilities.

If people don't want to use the capabilities, they can buy Microsoft Works, you know, what's the volume ratio of Microsoft Works to Microsoft Office? About 40 to 1, in favor of Office. And you know, people actually send fairly rich documents around at work. You know, take PowerPoint, we could not find a single command that over 30 percent of the users didn't use in PowerPoint. And these are the instrumented versions that are out there.

So in terms of are people using the full power in the machine, no they never will. You know, when you buy a four-wheel drive vehicle, do you drive on snow all the time? Probably not. You know, there's a lot of power there. If somebody emails you a document that's very rich, you're going to be able to view that document, you're going to be able to edit that document and send it back to them with total fidelity, in terms of what's out there.

The industry survives in a market place where if people don't want the new versions, they won't buy the new versions. If they don't want the new PCs, they won't buy the PCs. But, the change, in terms of how ambitious people are in using these things has gone up really incredibly. I mean, people are really mailing around rich documents. They're not just mailing around little notes to each other. They are creating electronic forms, mailing those things around, and you know, think about the PC of three years ago. You had to go buy -- you had to go out buy your own TCP/IP stack. You had to go out and buy the NetMeeting software, and the telephony add-ons, and all those different components.

So, you know, we think we're doing things that people use. We try and measure those things as much as we can. And for the user who doesn't want to use these new things, there are products out there, from us and from other people, that are aimed very, very much at that entry level of usage. People do like a lot of headroom. When they learn a product, people like to know that if they ever want to put a footnote in a document or something, that that capability actually exists there. I do think that we have reached some limits. For example, the technique we use for presenting interfaces. Today we use 2D menus. And 2D menus should present about 30 or 40 options. That's all. And so once you've used 30 or 40, if you're going to add features to those products, you'd better come up with a new way to do it. The new way to do it, fundamentally, is that you allow people to type in essentially arbitrary English. And you saw this feature come out in Office 95, and it was greatly enhanced in Office 97. We used it in the help interface, where you could ask questions.

What it's moving to, though, is to be the primary interface, where the menus are there for the things you use commonly, and we'll reduce the size of those menus. But, if you want the broad general functionality, either through speaking to the computer or typing in sentences, that's the way you'll access that functionality. So it will go from being a way navigating the help, to be a way of actually telling the computer what you want to do. And there are a lot of benefits of going that way. You know, we've demonstrated how we do that in the database for navigating data. That's the trend in terms of Internet search engines, to actually parse and understand sentences. Nobody's done it yet, but Microsoft and others are working in that direction. And it all fits right into the natural interface approach that we're taking.

So PCs will be overpowered. In a sense, you know, in terms of using up your disk space, people used to really use up their disk space. And you didn't have to feel bad that on that 10 megabyte machine that people were wasting disk space. Nowadays, people are wasting disk space. There's a lot of empty disk space out there. And to the degree that it's really expensive, then you say, hey, it's bad. You know, your PC is sitting idle, why don't you take it over to someone else's place and let them use it. Well, if it's expensive enough, then absolutely you should do that. But, you know, when you weigh in your time and the complexity of those things, the bottom line is that you want to make it cheaper, you want to make it more powerful, and you want to make the access to that capability there.

So we are in an interesting juncture, where the 2D menu structure, and the wide range of utilities that are in the system, and the different ways we store information, has reached an upper threshold, in terms of the complexity that people are willing to learn. And so we do have to have unification. We do have to have an elimination of commands. But, I don't see us backing off on the power that's in that system. You know, the rich documents you can create today with computers, I believe you'll be able to make those rich documents and more, with these computers in the years ahead. We just have to make sure that it doesn't burden the person who doesn't want to use that feature.

Question: (Inaudible.)

Mr. Gates: Well, there's a lot of people who have done voice add-ons to Microsoft Office. In this whole area of speech recognition, you know, the problem is a quality problem. The demos are incredibly good, you know, it looks like, wow, you know, you see the demo and you think, I'm going to use this. Well, 90 percent of the people who buy speech recognition products, three months later they're not using the speech recognition product.

Now, you know, just because it's over 90 percent, that is still an improvement. It used to be 100 percent.

(Laughter.)

Mr. Gates: And, you know, there's a question of when should somebody with a lot of credibility come out and suggest that anything like that is going to be mainstream? Honestly, the recognition quality is not there today, except in very narrow cases where you have an absolute need not to use the keyboard for some reason, or you have a very narrow command set, or you just happen to get such a kick out of it that that's the way you want to use the machine.

So, it is coming. We are working with all the vendors in this area. We're actually the biggest investor in research ourselves in this area. We own a substantial part of Lernout & Hauspie. We work with Dragon, we work with IBM, and all of them do add-ons to Microsoft Office.

The interesting juncture is when you get to where you do a fundamental design of Office that assumes speech recognition, and we've been prototyping that. We have some great, neat agent stuff you can download from the web where we use speech recognition that's proven to be very popular. A lot of people are playing around with that. But I don't expect you to do mainstream dictation for probably three or four years. And even three or four years may be too optimistic. The markets where this will catch on first are in Japan and China, where the keyboard is particularly difficult to use because of the size of the alphabet. And both by working with a third party, and by developing our own internal speech technology, we'll be there when we think it's got the right quality level.

You are seeing today in our products things like speech synthesis. Encarta and Encarta Africana ship with speech synthesis and Windows 2000 will include speech synthesis. You're seeing the IntelliWizard, you're seeing the English language query, and we'll put the voice recognizer into products where the narrow vocabulary makes that work. The auto PC, which is the Windows CE product that you use in your car, where you want to ask for directions and control the radio, that's based on speech recognition because it's a small vocabulary, and you don't want to take your eyes off the road to look down at some screen-type interface.

And so, every year is a year of progress in this area, and we absolutely believe in it, but we have to be careful not to bring out a product like the Newton was, that even though from a technology point of view, it was an incredible piece of work, but putting it out in the market prematurely, it stunted the opportunity for a handwriting-based product for quite some time, because it just wasn't ready for prime time. And my belief, based on personal use of all these products, is that we've got a lot of work to do before we meet the threshold. And certainly for a product run at Microsoft, I want to make sure that it really is mainstream before we go out with it.

(Applause and end of event.)