Remarks by Bill Gates

MIT LCS Laboratory Anniversary Speech

April 13, 1999

BILL GATES: Well, good morning. It’s always unusual for a Harvard dropout to be here at MIT, addressing a distinguished crowd like this. But, of course, I got started when a friend of mine, Paul Allen, recognized that the impact of Moore’s Law on chip technology would lead to computers that were very different than anything that we had been playing around with over the years. We have been beneficiaries of time-sharing that, of course, got a start here at MIT, but the idea that these machines would become incredibly inexpensive, almost personal in nature, was something that excited us a great deal. That was back in 1975, and the structure of the computer industry was very different than it is today. The way that people thought about computing was very different than they think about it today.

What I want to talk about is where we’re going to go with software, and how we’re going to get there, because not only are there interesting problems to tackle with software, the actual art of creating software has changed surprisingly little since 1975 - or pick any date that you want to. Go back to the first higher level language, think about how debugging, how specification was done, and we have hardly changed at all in terms of how it’s done. And yet, some of the problems I’m going to put forward require a scale of software that certainly requires a breakthrough in the techniques for how we build software.

The PC’s success loop is an interesting kind of phenomenon, a phenomenon where the more volume of machines that were sold, the more software came out, and therefore that volume allowed for great price reductions.

That success loop led to over 100 million of these machines being sold every year. And there’s a very large software market out there.

Back in 1975, before the PC, there effectively were not independent software companies. There were a few that did some high-end application and databases, but 90 percent of the industry was hardware oriented, because that’s what was expensive. Today it’s reversed. The software industry, broadly defined, is much bigger than the hardware industry, and that’s a trend that will certainly continue in a very big way.

Well, what are we going to be doing with these computers? We’ve actually solved a couple of problems fairly well, creating documents, keeping databases of customers. Of course we’ll continue to improve with those things. But the future we want to tackle is more about communication. It’s making it so that when you come to a meeting like this you have a computing device, you can take notes, you can look back at what I or other people may have said in the past, and make some comparison there. You can always stay in touch with the information you’re interested in.

Today, computing is not much associated with learning or entertainment. That’s been talked about for a long time. In fact, virtually all these things we talk about were anticipated by Vannevar Bush back in 1945. You read what he wrote, he talked about keeping track of everything somebody does, dealing with digital photography, having the entire encyclopedia there at your fingertips. And so in terms of visionary work here at MIT, over 50 years ago there was a vision laid down that we’re probably about 20 percent of the way toward achieving.

During this time there’s been a lot of turmoil. New companies coming in, great innovation. This ad you probably never saw, it’s from one of the 50 companies that made a dedicated word processor. And they ask the profound question: why CompuCorp? Well, why indeed? All 50 of these companies went out of business, and yet there was a period where people thought about having a computer that was dedicated to that one task, dedicated to word processing. And at the time people made fun of the idea - why wouldn’t you have a machine that was dedicated to that task? How could a general purpose machine compete with something that really was designed for just word processing? And the answer is the magic of software. As the PC came along, even though it was inferior in many ways to these machines, the breadth of what it could do and the phenomenon of the software industry writing for the PC meant that these machines became as antiquated as an adding machine. Perhaps more antiquated. You still have adding machines, but you have very few dedicated word processors.

Today we’re taking on a whole new class of software problems, because of the Internet, because of its role as the most revolutionary communications device of all time. And it’s fantastic that MIT, through Tim Berners-Lee, through the W3C and through many activities here, is taking a very important leadership role.

The Internet is not like anything that came before, because this ability to match people with common interest, this ability to let people engage in a dialogue that’s any number of people to any number of people, either synchronous or asynchronous, is critically important.

When you think about the fundamental nature of capitalism, it’s about matching buyers and sellers, and every communication tool that’s been used to date to drive that mechanism has had a lot of overhead, a lot of difficulty in making sure buyers knew about the sellers. Well, the Internet is unbelievable at doing that, and so it’s actually taking capitalism and defining it in a more efficient form. In fact, businesses everywhere are having to ask themselves, what is their strategy in this new era. The pricing, the role of the middleman, it’s all changing, and the one thing we can say for sure is that it’s changing for the good on behalf of the consumer. Consumers are going to get better products at better prices, because this is all taking place.

Companies are finding now that software, to some degree, defines their company. If you go to a company’s Web site and see what products it offers, and that Web site actually remembers what you bought, what your interests are, and talks to you about different things there, that’s a fundamental basis of competition for that company versus another. In the case of Amazon, which was born as an Internet company, that’s obvious, but even for traditional companies, there the software on their Web site will really be how people think of that company, and how they compete.

So software is taking on a new role, even inside businesses. Even the CEO, not just the CIO, is having to think about, well, who are our partners, how are we getting this done? They’re having to go out and look at their competitors Web sites and think about the flow of information inside their company in a very different way.

So it’s a period of great change and great excitement. Microsoft is very involved in building some of the software that are really just building blocks that make it very easy to build these very high-end sites.

One of our key partners is Dell, and it’s been fun going around the world, working with them to set up these sites. I’ve got here the screen of their German sites, and their French sites. And when we were actually putting the French site together, there was a question, we got it done ahead of schedule so Dell hadn’t run a single ad telling anybody the site was there. And so we thought, well how long is it going to take before somebody finds the site and buys a computer. And the answer was fifteen minutes. (Laughter.)

And it just makes you wonder, was there some guy out there who every fifteen minutes had been typing in that URL and getting nothing and then finally it was there and he could buy his computer? Now, was that a coincidence? Well, we went to Germany, same situation, not a single piece of promotion. And again, it was fifteen minutes before somebody bought the first computer.

And so this is the Internet in its most profound sense, where it’s not just replacing paperwork of a buyer and seller that would have done business anyway, but it’s allowing people to find products, to find configurations that just wouldn’t have been possible before the medium came along.

We are seeing the early stages of this. And everybody here in this room is very involved actually taking the Web for granted, living what I call the Web lifestyle, where for news or sports or organizing a trip, you just use it and it’s there, and most of the people you want to communicate with are connected up. Even on the most optimistic basis, though, the percentage of the population in North America that’s using this is about 29 percent. And that includes people who just do one piece of e-mail a month. If you really take a tight definition, it would be less than 5 percent truly engaged in this Web lifestyle. And many of those you’d find on university campuses or in high technology companies. And so it’s just starting to spread out beyond that core group. That core group is incredibly influential in showing us where the future is going, how people are going to be using these things.

Well, once you get outside North America, the numbers drop down very dramatically. Even in Europe, where you’d expect the numbers to be similar, there’s about a factor of four or five difference in terms of level of use. And that’s a significant fact, but a lot of those countries are saying, how do we change that? Because the efficiencies of their businesses, the way that their universities are equipping people, they’re not taking advantage of the full potential of what’s there.

And certainly if you look at the rest of the world, the percentages are very, very small.

So there’s a lot we’ve got to do, and we’ve got to make it simpler, we’ve got to make it cheaper, we’ve got to make it more relevant. And to kind of stay in touch with how people are thinking about this, we often have somebody go out and just interview people on the street and ask them what do they think of the Internet, why aren’t they living this Web lifestyle. And so I’ve got a little video that shows that, that I’d like us to take a look at.

(Video presentation.)

(Applause.)

BILL GATES: So we have a long ways to go. (Laughter.) A lot of improved software needed to achieve the vision.

As part of this, a lot of things that have been physical goods will become digital. It’s been interesting over the last five years, talking with the music industry, saying, "Hey, your product can be delivered in a digital form, in a form that’s easier for the user to select, easier to transport." And it’s only in the last year that this has become really evident to all the people in that business.

In the area of books, it’s going to take longer. We have to have some breakthroughs in terms of the screen hardware itself. We have to have some breakthroughs in terms of the software that lets you read the material. But it’s clear that within the next five years, those advances will actually make it so that people will read even a magazine, a thesis, a long document off of an LCD screen.

As magazines have started to put their material out on the Internet, very few have seen a substitution effect; that is, people canceling the paper subscription to read the magazine online. I have a high-speed connection, like many of you do, and many of the trade journals that I would have read in paper form I am starting to ask, why should that arrive at all? I don’t look at it except up on the screen.

And so the business shift of music, of documents, those will all be taking place in the relatively near future, delivered across the Internet to a new kind of device that not only lets you read, but lets you take notes, annotate and traverse the entire world of information.

And so people often talk about this as a digital world, where everything in the physical world will be modeled, where you’ll be able to navigate in a 3D interface. This idea of virtual navigation has been talked about a lot, but the quality of the graphics, the speed of the download meant that today the Web is basically a 2D experience. I think that will change, but there’s some threshold in terms of the quality of the graphics update, the resolution, that we certainly haven’t achieved yet. And so there’s a lot of work on both the hardware and software side to make it easy to do those things and build those sites in a way that people find attractive.

As we model the physical world in the digital world, there are some concepts like trust that we need to think about in a fairly deep way. In fact, in many cases the digital world can do a better job than the physical world, the main problem being that if there is a flaw, the ability to exploit that flaw in incredibly large ways is very different than in the physical world. So a very tough and interesting problem in which the public key crypto technology that was pioneered here will certainly play a central role.

Vannevar Bush actually talked explicitly about this digital life record, the fact that everything you say, everything you see, all the notes you take, the people you’ve met, what you’ve learned about them, that all of that will be easy to capture. And that leads to the question of how would you organize it, how would you find the best picture you took of your kids when they were a certain age, how would you want to go back and retrieve information; particularly if some of that information, the computer’s ability to recognize it, is probabilistic, and so it’s imperfect in nature, and so you want to make the search as accurate as possible. You’ll have a wide variety of things, and you want to call it up in different forms.

This also raises a fairly deep privacy question. You know, would somebody really want this to happen? What are the plusses and minuses of that? We see that in a small way with electronic mail today. It’s mostly plus, because it lets you establish a record of exactly what went on and how it took place. But these are tough questions that are not yet answered.

Now, as people get all this information, where is it going to be? Today, the user is explicitly involved in moving the information around. And if you have a computer at work, a computer at home, you copy the files around. If you’ve got a palm sized device that you carry around, you think about, okay, have I done the replication, did the replication work, a lot of trouble just moving the information.

And what will happen is that all your information will simply be in the clouds. It will be on the Internet, and all of these devices of various types, as they connect up, they’ll do intelligent replication to know what information they should go out and retrieve. It doesn’t mean that as you actively edit the information it’s always necessarily being replicated back, but as you leave a device the changes you’ve made go back to the center, so they’re available anywhere.

Today we have some things that work this way, some of these file systems like the Andrew File System, we have sites on the Web like Hotmail that are accessible anywhere, some of the synchronization software, but it’s really in a very primitive state compared to what we want.

In the future what we want is this: when you buy a new PC, you hook it up and boom, your files are there, your applications are there, the music that you’re interested in is there. Even in the home environment, it’s silly to think about having to move things between different devices, and making the user get involved in that. The idea of data moving anywhere is a very important concept for advanced storage.

We also need to move up in terms of the semantic level we deal with in this information. When you have two companies out on the Web that want to do business with each other, there’s a problem, because one company has a way it likes to buy, and the other company has a way it likes to sell, and how do you match those things up, how do you allow that transaction to take place.

Well, building on XML as a standard, there’s now a lot of great work going on to build schemas and to have standard ways of interchanging information about places, products, purchase orders. You know, even simple things like calendars are personal attributes, your zip code, things like that. We’re just now getting the Web to be able to share those, so you don’t have to enter them in again and again. And yet we want to have that information in a form where you always have control over how it’s used.

So building up these rich schemes and very powerful tools, and perhaps even new languages that can allow for schema mismatch is an important research topic. You’ve got to have tolerance for ambiguity. You’re not going to have one standard schema for all these things. The world is just too rich and too complex. You can see that when you think about ERP applications; an application like SAP has over 10,000 tables to represent the objects that businesses deal with. Whenever that product is brought into a new business, there’s an immense amount of customization that has to take place; new tables, new logics that are built because each business wants to think of its information in a different way. And yet parts of that information you need to exchange across business boundaries.

So the new emphasis that’s going on with schema, how schema relates to natural language is a very important software area.

Now, the hardware guys, again, are doing their jobs in a fantastic way. There are not going to be any limits to us as we tackle these new frontiers in terms of either processing speed or storage capability. You can get now about a 50 gig disk for about $3,000. When people buy a personal computer today, they’re actually buying a disk that they can type their entire life without filling up. And it’s not even economic for companies, if you want to make a cheaper PC, to just do a 1 gig disk, because the economics are such that you might as well give the guy 4 gigs of storage, even in a very low-cost machine. So we’re going to have things with terabytes and with thousands of terabytes, or millions of terabytes that let us deal with all this information.

Classic databases don’t need this much storage. You know, you can take every transaction that’s ever been done on the stock market. It’s nowhere near a terabyte of information. In fact, when Microsoft wanted to play around with this idea of, okay, can our databases, can our backup software deal with a terabyte, we had a hard time coming up with where were we going to get this information. Well, the answer, of course, is multimedia. It’s video, it’s audio, it’s all the imagery that’s out there. So we actually worked with the people who do satellite photography, both the Americans and the Russians who had a lot of data that they wanted to get up and sell on the Internet, and built something that has several terabytes.

Here you can see with these satellite images, what it looks like when you zoom in and look at an airport, so you can see the planes and the resolution that’s available here, actually that’s available inside the government is dramatically better even than what are in public databases.

It’s been fascinating to see people using that - we said, let’s just put that out there on the Web and let people zoom around and see the different parts of the Earth’s surface. And we’re getting about 50 million hits a day. It’s been interesting to really see how people do the queries, how they look at that database, and we’re filling out the entire database, and once we’re done it will actually be close to about 20 terabytes of information.

Well, of course, the form factor in computing will be changing quite a bit. There's nothing preordained that the computer has to be large in size. That’s just an inconvenience. It’s an artifact of the hardware. Eventually we’ll characterize computers really by only two characteristics: the size of the screen and whether it’s a portable device connected up through wireless networking or it’s a connected device that gets the entire bandwidth you have by having a fixed connection. And so every size between something that just sends an image to your retina by being part of your glasses to something that is a wristwatch, or the pocket-sized device, all the way up to your entire desktop where you might have a screen that’s three feet by three feet in size, every one of those will be a choice.

And there are some interesting software problems about how do we create a user interface that’s scaled across those different sizes of devices. We don’t want to ask people to create Web sites for every possible screen size. And even today where we have modest variability in screen sizes, it’s a little bit of a pain that many sites don’t take advantage of a large screen and some sites work very poorly when put onto a smaller screen.

We get involved in that, because we have a product called WebTV that takes Web sites and puts them up on an NTSC display, where the resolution of the graphics is substantially less than you get on a computer display. And there’s a lot of interesting tricks that allow you to do that better than most people expected, but even so the fact that the HTML is really tied to that screen layout and can’t be a little bit more abstract than that is one of the interesting problems that I think is important to solve, so that we don’t get a fracturing of these Web site standards. In fact, for a while it looked like the wireless phone people were going to go off and do their own stacks and their own display protocols and things like that, but I think that’s probably in the process of being prevented, which brings to the fore this issue of how can we get those specifications to be scalable in nature.

All of these devices will have browsers built in. All of them will be connected to the Internet, either through a wired or wireless type connection.

The devices that are really taking off in popularity now are these palm-size devices. We still don’t have the wireless network. The battery life is not quite as good as it should be. But within the next five years, those will probably be as prolific as the larger sized devices are as well.

The large screen will still be the center of gravity, in a sense, because that’s where you edit documents, that’s where you create documents. And so when people say to me, what about the PC, well I define the PC to be anything that’s got a screen big enough that that I’d want to create or edit a document. And with that kind of definition, the shape may change, the hardware technology, the software may be completely redefined, but it’s got a very central role that’s complementary to your TV set, the computer in your car and all these other devices.

Well, this vision requires us to move to new levels of scalability. And even in the near term, with clustering approaches, we’ll be able to do dramatically better than we are able to today. In fact, Web sites actually lend themselves to a pretty natural clustering approach, where you can even take the IP connection and use that as a way of distributing the load across many different machines. We’ll be able to move up to 64-bit processors, much bigger SMP type machines, and then clustering those together. And so here, again, I don’t think the hardware will be the limitation.

Now, the most interesting software problems are the ones that we haven’t made any progress on. And it’s interesting to think how good is our prediction about these problems been. You know, how quickly did people think speech recognition would be a solved problem. Well, when I left Harvard, over 25 years ago now, people thought we were within five to ten years of that. And it’s interesting that it’s proven to be such a difficult problem in its broadest sense. Now, here at MIT and several other places, when you take subject domains and focus in on those, there has been great progress. And so the next decade I think we’ll see substantial solution to mimicking the senses: vision, speech recognition, speech synthesis and even navigation, creating small robots that actually have some utility because they can see and move around and do things that are very helpful.

One of the tests that goes way back, even before Bush, is the Turing test that asks, can the computer fool you? And we can think now of many varieties of the Turing test, the Turing test where you just have a text dialogue and a narrow subject domain, and in the simplest sense you could say that computers are getting close to solving that. When you get to broad subject domains and you get a richer interaction, with voice or vision, then it becomes a very interesting challenge. In fact, here again the kind of software, the kind of computer graphics we need are well beyond that of what’s available today.

Eventually people will want synthetic actors, the ability to take any sort of scene and actually pick who appears in those scenes and have it be done in a truly realistic fashion. That’s about a factor of a thousand more difficult than current technology allows us to do.

Another interesting challenge is how we deal with meetings. You know, what is it about sitting face to face that we need to capture to allow that to be done at a distance. If you look inside a corporation and you say what do all those office workers do, it’s kind of a good question. Well, the answer is: they go to meetings. And if you look at a university campus, a lot of it is meetings. And as people are asking questions about this, you know, it really requires a blending of the sciences, the social scientist, the software scientist, the people involved in hardware to think about how can we build something that would be as good or better. And I’m actually an optimist about this, although again we’re just at the beginning of it. I think for various types of meetings with lots of distributed participants we will actually be able to create something that in many ways is better, and allows people to be far more productive.

Just imagine if you could make meetings 10 percent more efficient. I mean, that would boost the economy for a good decade alone. So it's a software challenge certainly worth putting some effort into.

Well, as I said at the start, these problems are pretty clear. You know, being in the software business or software research, this is the most exciting time. But we’re going to have to change the way we create these systems to really do this well. Over the last 20 years, there were a number of solutions that people suggested would really make the breakthrough, make software simple. Well, we have made some progress, but people’s expectation of software, the scale of system, the level of reliability that they want out of these systems still greatly exceeds what’s possible. High level languages took us a certain distance. Object oriented programming in some ways you could say that’s just coming to the fore now, that’s helpful. It gives us some degree of reusability. But the kind of factoring you need to do has proven to be very difficult, and again require the very scarce programmers that are extremely good and therefore a real bottleneck in getting these things done.

There’s still a huge mismatch between how you manipulate data in the language, and how that data is represented. The in-memory data, the on-disk data, the way that you navigate through that data, advances in database technology, creating an object environment, where the objects move around automatically and they move from disk to memory, that’s going to be an important part of making the program specification be as simple as the problem that it’s trying to solve.

One of the toughest areas, and one that we’ve been in our research group looking for as many people as we can find to help tackle is the area of testing and verification. We have a database of the kind of mistakes that programmers make that’s bigger than any company around, I think. And yet the willingness of people to say, okay, we’re going to go through that and by looking at that we’re going to figure out exactly how tools can be better, that hasn’t been an area of hot research. And yet I think the impact of that can be pretty dramatic.

We’ve got to have programming environments where we can specify things and get great concurrency. Concurrency at many scales. You know, Moore’s law tells us about transistors on a chip. It doesn’t tell us that we’ll be able to map those transistors into linear performance improvements. And if we can’t take advantage of multiple threads, multiple processors running in parallel, then even with all those transistors, the performance gains will start to be very, very small indeed. So concurrency is largely an unsolved problem. Certainly here is the case where we’re humbled by nature. Most interesting software problems were solved by evolution in a very profound way, and what could we do, if you imagine having 100 cycles in parallel, which is probably given the rate of transmission of information in the brain, what goes on in a typical recognition event, what could we do at 100 cycles. Well, basically nothing, by comparison. So there’s something there waiting to be discovered.

Declarative programming, we have some great examples. Bob Franks and Dan Bricklin, with the spreadsheet, sort of brought declarative programming to the masses. SQL in a certain sense has done that for a certain type of database problem. And yet as we’ve tried to go beyond that general-purpose approaches like PROLOG, they have not been successful.

I think it’s very important that we do that, because when people are talking to computers about what they want it to do, sort of specifying the agents and the information they like, that’s got to be a form of declarative programming. You don’t want people to have to create explicit logic in the classic sense.

Intelligent environments are going to require us to have data models that go way beyond what we have today, and data models that deal with uncertainty, because, you know, the world is a world of probabilistic inputs, including things like the speech recognition systems that we have to have.

If we take these software problems and we say, well what does it create as a system, these have got to be systems that work. People are relying on the Internet in a very deep way. And every new thing that’s put up there is going to be incredibly mission critical. So these systems have to work, they’ve got to adapt to a changing environment. You’ve got to be able to update them without actually bringing the system down. And you have to be able to simulate things in a rich way. And so there’s a lot that we have to do in terms of how would you specify those systems, and how would you test those systems.

Some people have suggested that genetic programming is the answer here. Well, I think that’s a great thing. It’s a fascinating approach. It leads to images like a software developer sort of being a shepherd to herd this virtual flock of programs and try and get them to do the right thing. But in a sense what genetic programming does is it puts the burden not on the writing of programs, but on the testing of programs. And yet today that is the weakest part of software technology. That is where we have the least advance. And so the people working in that field, I hope they can make a breakthrough, because I think it would be a profound contribution for all types of development.

The final software problem that a lot of great work has been done on here at MIT by Marvin Minsky and many, many others is the problem of learning. In fact, when I dropped out of Harvard, I sort of thought, well there are two fun things I could do, I could work on AI, be a graduate student and enjoy that quite a bit, or I could go start a software company. And I thought, well if I start a software company I could miss out on the breakthroughs that will probably take place imminently in that area. Well, great people are working on it. There is progress, but in terms of being able to objectively measure and say, what have we really solved. When we play chess, do we use a learning algorithm or brute force. We use brute force. In all the interesting problems that were proposed to help us make progress on this, in fact have been solved in a very specific manner rather than a general manner that those problems suggested.

If we don’t make much progress on this over the next twenty or thirty years, it’s possible that we’ll understand how to do this essentially be cheating; that is, by understanding enough about genetics to see how evolution did it, and then simply taking those algorithms and re-mapping them from a carbon-based world into the silicon-based world. But if somebody asks me what kind of idea would allow them to create a software company that would be far more successful than Microsoft, I think the answer’s very simple: just invent a way of doing learning software and the opportunity is boundless.

So in conclusion I’d say we really are just at the start of this. You know, despite the hype, the numbers about personal computing and the Internet, we’re going to look back on how we’re doing things today and think of it as very primitive, even in ten years time. Institutions like MIT are really at the center of helping drive this forward. We feel privileged to have the collaboration. And I think of software as being the most interesting part of this, because software tools are what are going to change not only how we do business, but also how we entertain ourselves and also how we learn.

Thank you.

(Applause.)

SPEAKER : Now, if you’re leaving, you’re leaving and going to miss a big announcement. Before anyone goes anywhere, we’re about to have a special announcement, and this announcement will be made by the President of MIT, Chuck Vest. I’d like to invite Chuck and Michael to come up to the stage, and Bill also. Could you please join us on the stage.

CHUCK VEST: It’s a tremendous privilege to be standing here between these two great leaders this morning. And this is the moment you all have been waiting for, because on the count of three we’re going to recreate that Riverdance scene. Are you ready? (Laughter, applause.) Seriously, it is an enormous pleasure and privilege to have the opportunity to announce to the LCS family that Bill and Melinda Gates, through the Gates Foundation, have made a gift of $20 million to help us create a new home in the middle of our campus, where LCS belongs. Bill, thank you very much.

(Applause.)

Now, this has been one of the worst kept secrets in history -- (laughter) -- but I want to read to you from the wonderful letter that Bill Gates sent to Michael Dertouzos over a year ago now. Among other things he said, "I do not plan to be in the Boston area for some time. I hope that we might wait until April of 1999 to make a formal announcement of this exciting project. For this reason I am hoping that everyone involved will endeavor to be circumspect with regard to this donation." (Laughter.)

Well, Bill, we tried our best, but as many of you know, the Wall Street Journal managed to preempt us last week. We apologize for that, but at least we did make it until April 1999.

Last Friday, the executive committee of the MIT Corporation formally voted to name this wonderful new facility the William H. Gates building. We are most grateful to Bill and Melinda for this tremendous gift, and of course to Michael Dertouzos for his leadership and his ability to infect Bill with the enthusiasm and vision for this new home for the Laboratory for Computer Science.

This building, the Gates building, will not only bring our LCS colleagues back to the heart of campus, but it will create new opportunities for the interactions and innovations that will once again redefine information sciences.

Bill, you’re an innovator of the first order, and you understand the importance of creating the environment and support that make it possible for people to do their very best. In this case, the people of LCS and their best, of course, is nothing more or less than shaping the future of information, science and technology.

Thank you again. And, it’s tradition at MIT that whenever we say something wonderful and positive about someone, as I have attempted to do about you, that we then ask them if they’d like to come forward and defend themselves. But you’re probably getting tired of that, but we would very much love to have you say a few more words. And thank you so much.

(Applause.)

BILL GATES: Well, I feel very privileged to be able to contribute in this way. You know, when people talk about the strengths of the United States as a leader in software and taking advantage of the Internet, you know the reason why this country is so much at the front of that is because of incredible institutions like MIT. In fact, MIT would probably be on the top of most people’s lists when they think about something like that.

I, personally, and Microsoft, have been incredible beneficiaries of the kind of work that goes on here and other places like this. Microsoft employs over 100 great people from MIT. We’d like to employ even more, as we tackle some of the problems I’ve talked about. And so when Mike sat down with me, and he brought it up, and he said, "Well, do you want to think about this?" I said, "No, I don’t want to think about it. That’s a great idea, it’s a great request, something I’d like to be involved in." So it’s great to be here, and I hope the facility gets done as soon as possible.

Thanks.

(Applause.)

CHUCK VEST: May the generations of students that will occupy the William H. Gates building share and reflect your vision and your generosity, through their contributions to computer science and the world.

On behalf of all members of LCS, past, present and future, thank you, Bill and Melinda gates.

(Applause.)