Bill Gates (Microsoft), "The Internet: Today and Tomorrow” - Lab for Computer Science Distinguished Lecturer Series

[MUSIC PLAYING]

MODERATOR: Wow. In 20 years of distinguished lectures of the MIT Lab for Computer Science, I think this is the largest crowd that we've had. Congratulations, Bill.

As all of you know, we're celebrating the 20th year of the distinguished lectures of the MIT Lab for Computer Science. And our policy is never to invite a speaker again. But since this is the 20th year, we broke the policy. And Bill is coming here for the second time. And I think that is very appropriate.

Those of you who know about the series, know that we like to introduce our speakers with a little tidbit from their personal lives that may not be known. Since there are so many young people in the audience, and after all, these distinguished people are great role models. Bill, as you all know, grew up in Seattle, went to school there, then went to Harvard. And I think everybody knows what happened from that point on. So let's not talk about that.

Instead let's talk about two things. And I'll keep this very short. What turned him into science, technology, software? What were the earliest forces? Because I know many of you have been through the same kind of experiences.

Well Bill says that until the seventh grade, he was getting lousy grades in math. And then a great math teacher came and said to him, I have confidence in you. You can do much better. Then the teacher roped him in to do math puzzles. Then after he got excited with math puzzles, the teacher got him to do them faster. And one thing led to the other and the fire was ignited.

The other big thing that I ask all our distinguished lecturers to do for the young people in the audience is to summarize their entire life's experience into one or two sentences, giving the young people a charge, a mission for what to do in their careers. Here is Bill Gates' mission to you.

He said, I don't know how it all happened. I love to play with software. It's a fascinating thing. I said, Bill, give me a sentence or two. He said, find something you really like to do and about which you have a great deal of curiosity. And it may be very surprising to you and to the world, even if it doesn't look like it might work at first. It might work after all. Bill Gates.

[APPLAUSE]

GATES: Well, good afternoon. What I'm going to go through today is talk a little bit about how the world of computing has arrived on the communications scene and going to have a pretty broad impact because of that, and talk about some of the research areas that suggests for software that are very exciting and at the same time very difficult. And then I'll leave as much time as I can for questions at the end.

Well as what was said, I got into this thing by dropping out of school. And I'm not recommending that. But the thing you're passionate about, you think there's only a critical time window to get involved in. Sometimes it does work.

And for me that was the kit computer on the cover of a Popular Electronics in January 1975. My friend Paul Allen and I had been working in software, and we thought, boy, we want to be there early on and try and contribute some software. And that was based on Paul's educating me about Moore's law, that these machines would become immensely powerful, more powerful than even the most expensive computers of the day.

And so we started off on that course really believing that software was quite distinct from hardware. And that if you wanted to have a company doing state of the art software, it would be very neutral to the various hardware providers and really optimize hiring and training around the software because it is quite different.

The PC has come a long ways in the little bit over 20 years since then. Not only the power, as Moore's law has done its work, but also the overall experience, the use of graphics, the communications in and out of the machine. It's easy to look at the machines we have today and say, boy, how did we ever use the machines of four or five years ago. But certainly, four or five years from now, we'll be seeing the same thing. We'll be saying you know why didn't we have a better user interface? Why didn't we have a motion video 3-D graphics built in?

I think it's not being too presumptuous now to say that the PC, as it's connected to the internet, will be as important as any major advance in communications. People talk about this being the information age.

Well, what the heck does it mean information age? Very strange term. Well, what it means is that in the same way we take electricity or running water, we take that completely for granted now. It's part of our regular experience. Well, take for granted the idea of using a variety of information appliances connected up to the internet as our way of getting information, of staying up to date, transacting business, of finding people with common interests.

And unlike those previous advances, however, this one in the period of a decade will become pervasive. People have a tendency to overestimate what can happen in two years and underestimate what can happen in 10 years. And certainly, with the internet, I think we're seeing a lot of that right now.

People who debunk it and say it will never be mainstream, they're just wrong. At least, we're betting in a big way that they're wrong. But people who think all these problems in acceptance will come in two or three or four years time, I think they're underestimating that it does take time for these things to be adopted.

Even today, you know, when I go to read the newspaper, I go up and I look at The Wall Street Journal online. And I look at the trade magazines online. But I always find myself picking up the real Wall Street Journal and scanning it.

Now it partly has to do with the resolution of computer screens. It partly has to do with familiarity. But I know very few people who moved all together to rely on the internet for all the information they get. So it will take some time.

And yet, if you compare it to these previous advances, it's far more rapid than any one of those. In fact, these advances took place over at least one generation. So somebody who was around in working when the invention came through didn't really have to care about it. It was only people who were growing up, the next generation who had to think how to make a part of their everyday activity.

Now, as this is all happening, of course, the machines-- and I use the term PC very broadly-- will be improving quite dramatically. Larger storage is important so that we can start to do caching at many levels. So all those pages that you like to go out and see on a regular basis, they're just going to be there. The faster processor is important to create a 3D virtual environment to do the kind of recognition and inference that you want to have in these environments. I think voice input is absolutely necessary to get into the mainstream.

I think people underestimate the importance of screen technology. And it's wonderful to see there are literally dozens of approaches for getting low cost, much higher dots per inch type screens available very broadly.

Eventually I think we'll talk about computer devices, primarily just by saying what the size and the resolution of the screen is all the way from the pocket-sized device to the desktop device to the wall-sized device to the stadium device. That will be the one unique distinguishing characteristic.

This internet phenomena is an incredible thing. And although for many, many years, we were optimistic about online services that people would dial up 1,200 baud and find something cool there. It was disappointing. It just never caught on. It never got enough people for it to make sense to build the community or to invest in publishing or to think about advertising as a meaningful thing there.

And it certainly was a surprise to me that in '93 and '94, a lot of things that have been around for some time, TCP IP and some newer protocols, particularly HTTP, HTML, actually got electronic communication to critical mass. I started to get a glimpse of when a lot of the people who worked for me went back to his university, Cornell, and said that it wasn't just computer scientists out there on the web, it was everybody in the university. And that was '93.

By '94, the phenomena was in full force, and a phenomenon with a lot more attention being paid to it than any that I've ever seen before. When the PC industry was getting going, there were a lonely group of people, a few people who believed in-- we'd get together in these small little industry shows and say, we know and IBM doesn't know. Well, later we had to invite IBM in to help make it happen, but that's a long story.

It was a group of pioneers working on their own. With this revolution in electronic communications, it's so mainstream, you can't get away from it, even if you wanted to. You can't turn on your TV set without seeing lots and lots of URLs. You can't read magazines without encountering that.

And there's almost a gold rush type phenomenon that we're seeing here. That's a fantastic thing, because it means the level of investments are very high. The innovation is quite rapid.

We actually went out and interviewed a musician down in Bermuda. And we talked to a lot of people, but we thought he was the most articulate about what the state of the art is and what's really going on with the internet phenomenon. So let's go ahead and look at a movie of what this musician had to say.

[VIDEO PLAYBACK]

MUSICIAN: Ha, ha, ha, (SINGING) It's digital, man. It's really clean. If you have a satellite dish on the lawn, and the computer is your best friend, you spend your free time cruising the infobahn, you're on the net hours on end. If you think you understand McClewan and all the deals that Villa is pursuing, it's cyberspace is your kind of place and you're technophile at heart, you're state of the art. Oh, yeah, you're state of the art.

[CHANTING]

[NON-ENGLISH SPEECH]

MUSICIAN: (SINGING) Ha, ha, ha. You've got the latest in technology and an ISPN connection, you use acronyms without apology and have a digital obsession, if you think the best home pages are really cool because they're outrageous, if you like to hack another guy's univac and you think that makes you smart, you're state of the art. Oh, yeah, you're state of the art.

[GUITAR MUSIC PLAYING]

[PHONE RINGING]

ANNOUNCER: Have you ever worn a fax machine on your head? You will. And the company that will make money every time you use it.

MUSICIAN: (SINGING) Ha, ha, ha. You take that gillion dollar gamble, grab a peice of the future land, join the frenzied content scramble, buy up the technical know how, take advantage of deregulation with a superhighway demonstration, and you spell convert, are you willing to merge and put the horse behind the cart, you're state of the art. Oh, yeah, you're state of the art.

ANNOUNCER: With the web of wealth system, you can start today with an online service, then run an interactive television system from the comfort of your own living room. You'll learn how to make deals, do lunch, and take 15% of the gross. So get John Avarice's amazing web of wealth system by calling this toll free number today.

AVARICE: Thank you. Thank you. Friends, before I discovered my Web of Wealth System, I was working two menial jobs. I couldn't even afford to buy my wife a used car. But today, I'm on wife number four. And they all drive fabulous cars, just like this. How-- thank you-- how did I do it?

By realizing that venture capitalists have to turn ordinary people like you and me into millionaires, or they don't get rich either. Plus, just mention the internet in your business plan, and they won't care if you're on parole. You will be funded, I guarantee it.

Why? Because as this chart proves, soon there will be more internet than people. Folks, nobody understands the internet, but everyone's afraid of being left out. With my Web of Wealth, you can be a part of this explosive growth.

BARRY: Yeah, I'd really like to be on that superhighway to fame and riches. So could you sign me up today to the Web of Wealth?

AVARICE: My point to you is that everyone has some worthless content languishing around in their garage, some weird little fetish tucked away in some closet. Please, put it on the net. And you, too, in days will be a global sensation.

Folks, everyone is cruising the net, from busy executives eager to expand their mind, to people with so much time on their hands they just don't know what to do. It's how they stay informed about what's hot and what's not.

Folks let's hear from some of the thousands of people who have used my Web of Wealth System.

PAUL: I started my first software company with a high school buddy. It turned out okay. But since then, I've discovered the Web of Wealth and Hollywood's beat a path to my door. Thank you, John Avarice.

BILL: Believe me, it's so easy to get into the online service business. If I can do it, so can you.

ANNOUNCER: Order John Avarice's Web of Wealth System today and get into the digital age. Operators are standing by now. You'll be banking. You'll be ordering pizza. And you'll be sending mail with a $12,000 computer instead of $0.32 stamps.

STEVEN: Before we discovered the Web of Wealth System, my friends and I were just struggling filmmakers.

DAVID: But now we've got our very own media company.

JEFFREY: And thanks to the Web of Wealth, we're no Mickey Mouse operation.

MUSICIAN: (SINGING) Oh, yeah, bud, we're making a place in history. The future is after and so are we with the avant garde held in high regard with a big head start. Oh, yeah, cause we are state of the art. State of the art. State of the art. Oh, yeah, man. State of the art. oh, yeah.

[END PLAYBACK]

GATES: All right, so you've got all this excitement, the biggest gold rush of all time taking place. What's going to hold it back?

Well as I say, there's social adoption, getting used to using this to gather information and buy things. But we also have a technology barrier. There's one element in this picture that precede at sort of Moore's law type rates. And that's the local loop connecting up, primarily in homes, but also businesses.

And here we can look at sort of three generations of capability. We have today's POTS phone network, where we can use 28.8 modems and dial up. And of course, text is very fast. And pictures are just barely okay. It takes anywhere from 2 to 10 seconds to get a web page down that has fairly high resolution images.

Now across POTS, people think they might be able to squeeze out another factor of 2, particularly if they go asymmetric in use of it. But that's going to be about it. And yet, for quite some time now that will be how a percentage of households are connected up.

Originally, there was a view-- well, five years ago-- there was a view that the companies would skip over any sort of midband approaches and go right to broadband. In fact, there was a view that they'd skip over the PC and they'd connect up TVs for video on demand. And cable and phone companies were trying to outdo each other with big promises of how many millions of people they would have hooked up by now.

Well, if you go out there and look at how many people are hooked up, it's a few thousand. It's a number that rounds to zero. And so something happened.

Well as they were making those promises, they weren't really thinking about what the revenue opportunity was. People saw that that simply by doing digital compression and using digital broadcast, your take rate on pay per view, where you only do, say, the top 10 or 20 movies, you get 75% of the revenue of the video on demand system. So your extra money to fund from that application to fund a true ATN fiber network is very modest indeed.

And new applications, whether for medical or travel, there was no way to bootstrap them. That is without users, there weren't the applications. Without the applications, there weren't the users. Time Warner made a valiant effort with their Orlando trial putting $20,000 machines into people's homes to just try it out. But it didn't spark into something important.

Now throughout all of this, clearly the PC was getting stronger, particularly in terms of the home market. And then the internet came along and woke everybody up to the fact that there's an evolutionary path. That is we can keep increasing the bandwidth, get better audio support, get better video support, and move to have higher and higher bandwidths that eventually will reach every home.

Now some of those intermediate steps are grouped under what I call midband here; ISDN, PC cable modems, ADSL. Any of these approaches let the still images come in incredibly good speeds and let you start playing with video. None of them are good enough that you'd sit and watch a two hour movie through a midband connection. So that awaits sort of the ultimate holy grail, which is the broadband.

And broadband will happen. We were spending a lot of time developing software for interactive TV. And a lot of the work we're now playing to the internet in security, a lot of the new user interface work we have comes out of an interactive TV work. In fact, the idea of being able to partition media servers so you can deliver audio and video using low cost technology, you don't have to have essentially a supercomputer server to do that, that all comes out of that interactive TV work and is directly applicable to the internet.

So now the world at large, I think, has a more realistic view of how bandwidth rolls out. And to be honest, I think there's a lot of households, that as long as it's narrow band, it's just not fast enough or compelling enough. The information they need, they can get from the newspaper or calling someone up. And it's only when we get into the realm of rich interaction, that it will start to be as pervasive as we want it to be.

Now, there will be quite a few physical device types hooked up, the PC. And we'll see more and more differentiation between desktop and portable. And certainly there will be a way of using your TV to connect to the web.

Pages are not authored to work with that type of display today. That's fairly straightforward. We need to have a well-defined subset of broad HTML, in particular, in terms of what kind of add on capabilities, what variety of media objects we're going to support. So that if we have to run that software-- make it run in a finite amount of RAM, this can work. So set-top boxes, building the electronics in the TV, or using game machines, which are out there and you can plug in a modem and a browser, those would be how the TV gets pulled in.

Certainly, a voice handset makes sense. Connect it up. It's a great trade-off in terms of the size of the device and simplicity. And that's often what you want. And it sure would be nice to have the connectivity for voice with the way that things are tariffed there.

We also are big believers in a machine that you carry around in your pocket that connects through digital wireless. In a sense, these personal digital assistants, things like Newton or the Shark machines or Sion, are sort of progenitors of this class. But I think there is a radical change in how you think about it when you have the digital connectivity, when you can just walk through a hotel lobby and check it out by having the wireless connection, where you can board an airplane without physical tickets and select to see through using that digital connection, when the GPS feature is cheap enough that you're seeing maps and it's populated with all the nearby things that you might be interested in. And certainly there are many, many other locations we'll see PCs, including in the automobile.

There's a lot of industry competition. And it's part of what makes our industry so much fun. There are multiple companies with operating systems. There are multiple companies with browsers. Microsoft is a distant, distant number two in that competition, having just entered it fairly recently.

There's a lot of competition over defining what is openness. Openness to me means that anything can be cloned. There's no patents. There's no intellectual property that stands in the way of somebody creating something that's compatible, but better.

And the beauty of that is it forces you to keep prices extremely low and listen to the customer feedback about how you can do better versions. All the popular products that Microsoft has ever done, DOS, Windows, they've been cloned, but we've been able to move fast enough to stay ahead here.

We've been on the other side of the coin in businesses like spreadsheets and word processing, where we started with very low market share, but the companies involved not only didn't they have anything proprietary that they could protect, but they didn't move their application along. So whether 123 or Word Perfect, their market share fell over time. Perhaps netware is a similar situation. And if we do extremely well, then we'd add browsers to the list of such competitions.

Now part of that is that we have a vision of integration, of actually having the local information we access on the PC be as easy to get to as remote information. And in fact, you're things like help files, why shouldn't those just be web pages? There's no reason you should have to think about help any differently from any other text that comes up on the screen.

And likewise directories themselves, you should be able to annotate. And you can think of today's directories as just a degenerate case of a web page. It's a web page where each file is essentially a link, which sort of hearkens back to the kind of gopher approach. Well, we want to take and let you put arbitrary HTML, including active controls, into those pages so it's easy to find your way around.

There's a major thing that's always going on of PCs versus expensive computers. As the chip technology in PCs has gotten better and better, as the software got richer, it's taken over more and more of the computer marketplace. And it's fair to say that within the next two or three years, the performance and capabilities will be such that you can reach up virtually to the top of the computing hierarchy.

And this is greatly reinforced by the positive feedback phenomena of getting more-- the more applications you get, the more volume you get, the more volume you get, the more attractive it is to build those applications. More recently, some of these people like-- well, some in particular-- have taken a tack of not trying to sell a more expensive computer, but selling something less. And we don't know what it is yet. When they actually exists-- this is a so-called network computer-- we'll be able to see what the trade-off involved is.

If you're talking about a high quality screen that you're going to sit close to and something that runs a browser, it's tough to eliminate much that's in a PC, because a browser is the most demanding PC application. It uses more RAM. It grows faster than anything, including word processing, spreadsheets. All the classic things that tend to push the limits are out the window. It's the browser.

And browser growth, if you just take Netscape as an example, it's doubling in size every nine months. And at some point, there's got to be a limit there in terms of what can be done. But still it's very relevant to additions that are being put in. So it certainly isn't going to stop anytime soon.

There's so much software involved in making the web fulfill this stream of information at your fingertips, there's programming tools. But I'd be the first to say that most people who create web pages won't be programming. Even if they want to put in little animations and buttons and rich interactive behavior, they'll just have a palette of controls. And then they say, OK, I want my users to vote on this, they'll get a voting control and then automatically be connected up to the right database capability there.

So the actual programming tools where JAVA has now come in a first class language that's going to be a choice just as much as C has ever been or COBOL or BASIC or any of the other popular languages. That's one dimension here.

But there's another dimension of helping people manage large webs of content. When you get thousands and thousands of links, it's very hard to make sure they're all correct. And when you start to reorder your taxonomy, how do you make sure that links that people are still holding that they get redirected to the right place.

You also have tricky issues of versioning. If I want to do some work preparing a new website, I don't want to duplicate everything I have today. I just essentially want a version off of that.

Versioning also comes up in terms of preservation. Now, one thing about written material is usually we can go back and dig around and find a historical record of what a product brochure looked like or what a company was offering. With the web, there's a tendency now to just completely wipe out the historical electronic presence. And that certainly won't do over time. We have to have an approach that allows for archiving, so you can always go back and see what was being said at any point in time.

So managing this type of content brings in some new issues. If we go back to the people who were in a sense the conceptual pioneers of hypertext, people like Doug Engelbart at Stanford Research or Ted Nelson with his Xanadu project, they had a much richer concept of what hypertext would be all about in terms of annotation and tracking and even commercial mechanisms that might make sense there than we've been able to implement today. So lots of work to go on that has to go on in an integrated fashion.

One problem that corporations have is that administering all this software where you have database software and mail software, web software, file sharing software, you have all these things, have different interfaces, different administration, different security. That's a real mess. And so we've got to really pull it together and take public key encryption and these X.509 certificates and make that the foundation for a directory that can be used across all the different applications and, therefore, not only simplify the cost within a corporate network, but allow these inter-enterprise applications to work because the trust hierarchies can be set up so that information moves where it should and doesn't move where it shouldn't.

Now the internet is today viewed as sort of an environment that government's not involved. And it would be nice if we could leave it that way. But I have to say that I think it's very unlikely that institutions, like the Security Exchange Commission, are going to let people do offerings on the internet without looking at the honesty of the material that's being promoted. I think it's unlikely the Federal Trade Commission will let people scam buyers without looking into that. And laws related to libel or copyright and those things probably will have to be accommodated, but it's a very difficult accommodation because we haven't really defined modes behavior and we haven't dealt with the fact that this is a global network. And it's going to be a terrible problem if there's overly restrictive laws that really prevent people from taking advantage of it.

Certainly the telecommunications deregulation bill went too far in terms of trying to limit indecency on the internet. There's now a court suit that Microsoft and a bunch of other people are involved in suing the federal government to get that overturned, which I'm very optimistic about.

There are major issues if this thing is as important as we'd like to think it is of making sure it's pervasively available. So sort of duplicating the creation of the public library system, which took a period of over 100 years to make sure anybody could get to books. That's got to be done. And the investments in communications infrastructure has to be there.

And one of the hot issues right now that I'm going to go down to Washington DC for another round of lobbying on is that this country has restrictive export laws on encryption technology. And today, they only let us export 40-bit encryption, which, of course, can be broken by many people in this audience. And it doesn't take you more than a day, say, to do it.

It's a pretty ludicrous situation because it's not as though there's some US monopoly on understanding encryption technology. And the problem is that software vendors outside the US are now starting to provide packages that have very good encryption. The problem is that organizations like the National Security Agency justify their budget on being able to tap into people. And if mass market software has very, very good encryption, then it just makes their job dramatically harder. Most of the work they're able to do benefits from sloppiness. And by having it built into these electronic systems, you'll get rid of a lot of that sloppiness.

And so it's a tough issue. And believe me, it's very hard to lobby against these intelligence organizations. But we'll see what can be done there. So definitely a lot of issues that come up.

I was always very proud of the fact that for the first 10 years of Microsoft's history we didn't have to think about government and lobbying and anything like that. It doesn't seem like that much fun in some ways. But now, as a major company in the industry, there are a broad set of issues in-- making sure that immigration laws don't get overly restrictive-- that we have to spend time in and get involved in.

I want to quickly talk about some research areas. And you'll see as I talk about these that these are areas that there's great work going on all over the world. Some of them are ones that Microsoft has a research group with about, oh, 20 people nowadays, that works on some of these. And they're all predicated on computing cost, communications costs declining, so that we can almost think of them as free over time.

And one of the things I think that's wonderful about software is unlike a lot of companies where research tends to get very decoupled from the mainline company and almost be partitioned off as though it wasn't part of the same company, in the software world, the ability to take great research ideas and when they work out being able to put them in products and take advantage of them is quite incredible. And we've already-- even though when we started the research group, we didn't expect that. We're only three years into it, and we're already seeing things like our Word processor ship with work that the natural language group has done.

I want to do a little demo of a fun thing we put together. This is just one person in research, Dave Kurlander, who is thinking about computer dialogues and how boring chat is and what he could do about that. So he was thinking, well, maybe this 3D stuff is great. But there's no record, and it's hard for lots of people to watch into that. And so he actually came up with a fun idea that we call comic chat. And it draws on a very rich art form.

So I'm going to ask Craig [? Bilenson ?] to come up. And he's actually going to connect up and hopefully show us Comic Chat with some people across the nation on the West Coast.

[? BILENSON: ?] Great. Thanks, Bill. You mentioned a few things earlier, things like Internet Explorer 3.0, which yesterday went to beta 1. You can get that on-- I heard you-- on www.microsoft.com. You can check that out today. Microsoft is very excited about that release. Bill mentioned the ActiveX technologies that's included in Internet Explorer. You'll see things like support for Java, support for scripting languages like JavaScript and VBScript, support for Frames, animated GIFs, all that good stuff that you like in HTML, or that you don't like so much in HTML.

One of the things that I'm going to show you is an application called Comic Chat, which is actually what we call an active document. So this document, when you click on it, could come up in place of whatever container object you happen to be working in, something like a Microsoft Word, Microsoft Excel, or a particular web browser like Internet Explorer 3.0.

I'm just going to-- [LAUGHTER] I keep hearing this hissing noise. I'm just going to introduce us to Comic Chat. I have a character here named Armando. And that's BillG-- your name is right there. I've always wanted to do this and pretend I was Bill G up on the internet.

GATES: A lot of people do that.

[? BILENSON: ?] I've heard that. What we have here is kind of a more fun version of a chat room with a record or a history of things that go on in that chat room. We can see all the people that are in the room. Here's myself as Bill G. There's a Mark ME and Tabscott.

In fact, I want to get these people's attention, so I'm going to go ring Mark. And he just saw a little dial on his computer. And hopefully we've got his attention now. And now that I've selected Mark, I can say something like, hi-- and you see my character waves-- how are you? And it also points to you.

Can everybody see this? Okay, why don't we make it a little bit. Okay, Hi, how are you?

You notice that as I type something like Hi, it started to wave. But then it said, how are you, so it pointed to the person that I selected.

We can scroll down a little bit. Someone doesn't believe that I'm Bill Gates. Right, I'm Bill Gates. And I'm really Ross Perot. Can we talk?

Someone asked me how I am. And I'm being rung back by Mark, whoever that is. So he wanted to get my attention as well.

So we can scroll down. Are you-- never mind. Elvis was here a while ago. This must be a very popular chat room. Say, I'm glad to be here. This is cool. And you notice that he's pointing to himself as I say that.

What are you up to, Bill? Pleased to meet you. All kinds of chat, things going on. I'm glad to be here. This is cool, pointing at himself.

So is it still raining in your neck of the woods? I can say things, cheer up, the sun is coming at. And when I do that, I can put the typical colon and parentheses, the little smiley face. And of course, when Comic Chat sees that, it makes my character smile. We just have to catch up to it. There we go. We're smiling while we put the little smiley face in our bubble there.

I can say something like laughing doesn't hurt either. And I can do the acronym of laugh out loud, put that in there. And my character is laughing hysterically as you can see just by using the laugh out loud piece.

There's all kinds of motions that you can draw out, though, which aren't automatic. If I want to say, hang on, let's show some emotions and send that across. I'm actually going to just make this a little bigger so you can see my character over here.

And you notice that we have this wheel on the bottom right. It's kind of like a color wheel in that you can bring out ranges of emotions. So someone can be kind of happy or very happy or very angry. So as I move around, it's very easy to bring out different impressions. Looks like he's doing a little dance, doesn't it?

Not only can you talk to these characters, you can do things like normal things you'd find in a comic book actually why we're calling it Comic Chat. So Bill likes to take risks every once in a while, so we can say runs with scissors. And instead of saying that, I could think it or I can even make it an action. And what happens there is it actually puts it at the top of the screen, much like you would see in a comic book. So there would be this narrative going by in the chat room as well as I'm describing things.

I can highlight everyone, so I can say something to everyone. And all the people show up in the room, so we can see everybody. It's easy to right-click on someone and I can ring them again. I can get information about them with their profile. And that should show up at the bottom. Hey, Bill what's up?

This man was raised by armadillos at a roadside park in Texas. He knows lots of armadillo songs. Want to hear one? We'll skip that.

So other things that you can take advantage of besides the emotions being the fact that you can save any of these, you can print these out, and it looks like a comic book, this will work with any normal IRC server. So I can go up to the internet. And the people who don't have this client, I'll actually see them as characters. And if they type something like the smiley face, their character will be smiling. And people who don't have this client will see my text completely normally. So it will work with anyone else who's using something like this.

Just to shock them a little bit, we do have the option of changing our character. So let's spruce Bill's character up a little bit. Anybody have any choices? Anna perhaps? I'm not exactly sure what Connor is. We'll have to look into that. Some aliens out there. And again, you can experiment with their different emotions and the different settings.

So something a little fun, something that came out of the research group. Again, you'll see that shortly, both for the web and for MSN. That's Comic Chat. Thank you.

[APPLAUSE]

GATES: I think it illustrates there's a lot of room for creativity in terms of how we're going to do collaboration. Just moving on the topic of where the frontiers are one of the big ones that we see is is parallelism. It's been kind of a holy grail in computer science for a long time. And as long as unit processor performance and memory subsystems scaled up, where most programs could run on a single computer, it wasn't that important. And so you'd have to say that progress has been fairly limited.

But now that we're looking at database problems that are absolutely gigantic, deep analysis problems, we clearly need to be a lot stronger in partitioning, distributing problems across different computers. And there's huge commercial payback for being able to take a sort of arbitrary SQL and seeing what the patterns of usage are and being able to design a system that sets up the data the right way.

Testing is another area where I have to say I'm a little bit disappointed in the lack of progress. If you think at Microsoft in a typical development group, there's about 60% as many testers as there are engineers writing code. And yet the engineer spends well over a third of his time doing testing type work. You could say that we spend more time testing than we do writing the code. And if you go back through the history of large scale systems, that's the way they've been.

But what kind of new techniques are there in terms of analyzing where those things come from and having constructs that do automatic testing? Very, very little. So if you know a researcher out there who wants to work on that problem, boy, we'd love, love, to put a group together.

Certainly, security is a tough problem. You've probably all been reading about two companies in Japan just lost over $100 million because the pachinko electronic cards, the encryption on those was cracked. And so somebody actually fed in to the marketplace before they noticed-- this is pretty impressive-- $100 million worth of pachinko little credit slips. And it just shows you how much pachinko gets played in Japan. But the idea of having good security and yet being able to change those systems if they're broken and having the amount of damage that can be done during a period where, say, your private key has been accidentally released, that's a very tough problem.

I'm very much a big believer in structure unification. Today, we have all sorts of storage systems. We've got files. We've got pages. We've got documents. We've got records. We've got messages. All of those are stored in different ways. And I really believe that there's some sort of super file system that has the replication, security, rich properties, indexing that can actually unify a lot of those together. And certainly, the experience of using a computer would be dramatically better as a result of that.

Some of the great frontiers have to do with sensory things, generating real video type images, the most advanced graphics work, being able to do more and more of that in real time. There's a re-architecting of the graphic subsystem, so you don't have the memory bandwidth bottleneck that the frame buffer is creating that we've been very involved in investing in, and all sorts of high level tools for synthesis. One of the companies that is now part of Microsoft is Softimage, whose software is used to create things like the dinosaurs in Jurassic Park. And we want to take those tools and make them so they can be used by high school teachers and put neat things together.

Certainly video recognition is a great thing. As long as you're going to have a video conferencing camera, why don't you recognize when somebody sits down? Why don't you recognize when they want to scroll more of a document to read more? If they want to throw the document away or whatever other crazy gestures they want to make, the computer ought to use all those spare cycles to sit there and recognize that.

Certainly speech is the holy grail. 20 years ago, people were optimistic. Today they're optimistic. I saw a wonderful demo of the work that Victor Zue's has done recently. And you know, I think certainly, over the next 5 to 10 years, those things will be deployed and used in a powerful way.

Text to speech, we've actually made a lot of progress recently where we can actually take a speaker and listen to them and then do speech synthesis that will really sound like the speaker. It's amazing how bad speech synthesis has been. But it looks like just in the next few years that will improve.

And certainly linguistics, being able to parse documents and gather knowledge from documents is one that's very important.

I think software will also get involved in deep analysis, deep modeling analysis, whether it's using Bayesian imprints or some other type of engine underneath. As you browse the web, you go to it and you think, well, what should I do this weekend? Well, how can the computer help you with that?

You shouldn't have to type in a bunch of URLs to go do that. It should synthesize for you that information based on your past behavior. Well, that's a fairly deep set of analysis that we're just at the beginning of. And likewise, buying something, learning about something, modeling what you understand and what you don't, and rich collaboration require actually techniques that are our deep, deep AI problems.

In terms of really rich learning, I was saying to a smaller group we were meeting with earlier, I'm surprised how little progress there's been in the last 20 years. Maybe it's just because I'm over optimistic. But when I left working in computer science, I thought the thing I was going to miss was that within the next decade computers would start to be able to learn in a meaningful sense, not just have a pattern that's preprogrammed in that they can happen to match against, but really a general purpose learning.

Eventually if we don't figure this out, we'll sequence enough DNA and figure out how the brain does it, so we won't have to figure it out. We'll just use evolution's solution. I think we've got maybe 20 or 30 years before that approach overtakes somebody who's trying to do it de novo.

So this whole software area as said in the introduction, I kind of got into it through serendipity. It's a fun area to work in. And it's a fast moving area. And by surprise, it's actually a great business. Looking ahead, there's a lot more room for breakthroughs and a lot of great value in doing these kind of products well. Thank you.