Remarks by Bill Gates, Chairman and Chief Software Architect, Microsoft Corporation
University of Waterloo
Waterloo, Ontario, Canada
October 13, 2005
BILL GATES: Well, it's great to be here. As you heard from some of your alums, Waterloo has contributed an amazing amount to Microsoft. But what I really want to focus on today is the future. How the software industry is going to make more breakthroughs in these next 10 years than it's made in the last 30, and how that software is really going to transform not just what we think about as the computer industry, but the way that everything is done. The way people design products, the way they communicate, the way they collaborate, the way they understand their business activities, and even extending out into education and the things we do as consumers, film, music, creativity all reshaped because the software is becoming a lot better.
Now, this dream of software as a tool of empowerment certainly goes back to the very beginning of Microsoft. That's a long time ago. We're celebrating our 30th anniversary this year. That means Microsoft has been in business longer than most of you are old. So, a long time. Big computers back then were very large and expensive, and there were very few of them. And there was no software industry. And it was actually a wild idea that the technology of the microprocessor would change the character of computing, and that the software would be the key element to bring that to life. Even Intel, which is an unbelievable company that was driving forward to actually create the engine that would make this possible, didn't see how it would be used. And so, myself and my fellow student, Paul Allen, had an idea that we could come in and provide the software. And, in fact, instead of the hardware and software being designed together by one company, that we'd allow all the different companies that wanted to make the hardware to match up, and have our software then provide a common interface, so we could really get a software industry going.
The idea of the software industry is really a pretty simple thing. You need to have enough volume of machines out there so that somebody can invest literally tens of millions in building a piece of software, and yet sell it for only a few hundred dollars and still make a profit doing that. And that means you've got to have millions of machines, particularly if you want people to be able to do very specialized software for different types of businesses, different types of activities. So, that did not exist, the computer industry rewrote its software all the time, there weren't this breadth of applications. And, in fact, it wasn't an industry that employs a lot of people, or existed in most countries.
The idea of the PC as an individual tool changed that, and we started a cycle, which was the more software we got, the more attractive the machine was, that sold more. The more we sold, the more the price could come down because these things are very volume sensitive, and as the price came down that further made it make sense to write more variety of software. And so that cycle got going. In fact, today for every dollar at the sort of platform level that Microsoft or anyone else does, there's over $10 that are done at the applications and services level, the variety there is pretty unbelievable, and that exists in every country around the world.
At the hardware level, the improvement that we were holding for and that reduction in price has really come true. The miracle has been delivered. It starts with this idea of exponential improvement. Gordon Moore, the chairman of Intel, made the prediction that twice as many transistors could be put on a chip every two years. He knew that would end at some point, but it's continued from the time he made that statement in the '60s up to now. And it looks as though that will be able to continue at least another 15 years or so. That's why if we compare the original machine that I did the basic interpreter for to a powerful personal computer today we see a lot of difference.
That early machine has 4k bytes of memory, not 4 megabytes or 4 gigabytes, but 4k bytes of memory. And a real early feat that we did was actually write a floating point interpreter that ran 3,100 bytes, so that in that 4k machine you'd still have room for the programs that you typed in, the variable table, your arrays and all those values in that nice little 4k machine. I won't say that you could do some huge type of software program there, but that was exactly what the hardware made available.
And so going from 4k to 4 megs to 4 gigs, that's a factor of a million change. In fact, these exponential improvements lead to that kind of dramatic difference. The difference is there in almost every aspect. It's not just the memory capacity, it's the disk capacity also, growing more than 5 million, networking bandwidth is almost infinite, because we just couldn't connect the machines together back then, and now broadband is not only available to businesses, but even to consumers, as well.
So the constant improvements in the platform allow us to be more ambitious in terms of what software does. We need to graphics chips to deliver better images, because certainly we want to create 3D virtual spaces that are either imaginary, entertainment, or reflect the real world. Where you can look at a product, what will it be like, or you can look at an image of downtown and see the traffic, and see what different things are going on, what's nearby, see that not in the sort of text-oriented way we do it today, but rather in an immersive, fully representational environment. So the graphics chips are going to let us do that.
We need screens with higher resolutions. There are, again, breakthroughs in LCDs in that we can do reading off the screen that's as good, or in the future even better than reading off of paper. That' letting us have portable devices, devices that are small that we can carry around, devices that are a variety of sizes. All these devices will be software driven. They'll all connect to the single network, the Internet, the Internet will take over not just for data, but for voice, for video. TV won't be something separate, it will simply be a straightforward data application that happens to run on the Internet.
As we think about the devices, in some ways they'll almost just appear in the environment, because we'll have lots of different cameras, and microphones, and displays that can project onto the floor or walls, or kitchen desktops. You won't think of the computer being in a single place. The environment will see that you're there. You can give spoken commands. It will be quite responsive, and so storage, and computing and the screen and the input will be disaggregated. Software will give us that type of flexibility.
You'll have some type of computer in your wristwatch, so you can just glance at information. You'll have the device that you can take out of your pocket. Of course, today's phones have improved very radically. We can think of them now as cameras, as well as phones. In the future they'll be kind of a digital wallet, they'll show us that map, not just a 2-D map, but a 3-D environment map. You'll be able to take the camera and, say, take a photo of sign in a foreign country and have that translated for you. You take a photo of a receipt that you get, say, at a restaurant, and it will see what it is, file that away, ask for reimbursement, get your expenses right without your having to do a single thing. Take a photo of a bar code or a product and your phone will tell you where you might get that at a lower price, what the competitive products are, what you might want to buy that's different. All of that will be kind of common sense. And that will be because we're bringing the richness of software into all of these devices.
Moving up from the pocket device, we have the next level, which is the Tablet. That's the vision that Microsoft has had for a long time. When we think about a Tablet we think about something that will replace textbooks. You'll be able to call up on the screen through the Internet the textual information, and interact with it, have videos, audios, have it customized by the teacher in a way that today's printed textbooks don't allow. And yet this Tablet will come down in price so that eventually it costs less than textbooks used to cost. So it's one little light thing that you carry around when you're in the classroom, and there is software that lets the teacher have some control over what you're doing, over what you're looking at.
As you leave the classroom it's your device of empowerment to pursue things, and look into anything you want. It's a device where the reading capability is so good that of course the idea of how you look at the daily newspaper, or those what would have been weekly printed magazines of course you're doing that off the screen because you can annotate, you can share it with friends, you can search. It's just far richer than it would have been if that had to be a paper-type piece of information.
Already in some areas, like the encyclopedia everyone would acknowledge that the digital form, either something like Encarta on a CD, or something like Wikipedia out of the Internet itself, is way richer than print-based encyclopedias. When I was young I read the World Book. I started with A, then B, then C, then D, and went all the way to the end, and it's not a very good way to learn things, because you're kind of jumping around between centuries and subjects, and you're trying to remember how things related to each other. The equivalent right now where you can do that on the Internet would let you do it in a far more ordered fashion. Let you test your knowledge, let you find things that are more in depth. So the way we can take people's curiosity and satisfy it is far, far better.
The Tablet concept will also change the desktop computer. You won't have a phone that's separate from your PC. Your desktop surface you'll be able to write things on and have that be recognized. So the input will be voice, it will be ink, it will be the keyboard, all really brought together. And the machine will have a sense of your schedule for the day, the context you're in.
When people want to contact you it won't be a phone number, multiple e-mail addresses. It will simply be one identifier, and based on who it is trying to do the contacting, you'll have software that runs on your behalf that you've set up to decide should you be interrupted right where you are, is it important enough, should a message be kept. What do you say to that person about your status for future availability, depending on who they are.
So you'll have an agent working on your behalf, and we'll laugh at the days when we had phone numbers, or we had multiple e-mail accounts, and we were being interrupted when we didn't want to be, or we didn't get the information that we needed that was urgent for us that the system didn't understand to bring that to us. No matter which device you're working on, that information should absolutely be made available to us.
So it's a period of dramatic change, not in the overnight sense during the dot com craziness, where people had said banking would change overnight, retailing would change overnight, but in a more gradual way. The dreams that existed then were valid, it was just the timeframe was a little wrong, because it takes time to get people to change behavior, it takes time to get these software systems to be simple enough, secure enough, rich enough, connecting together so that all the information is there, so that these things can be delivered on. And yet every one of those dreams we'll be able to pull together within the next 10 years, so a dramatic change in every sector.
One place we can see this very vividly, and I'm sure all of you are essentially avant garde practicers of the so-called digital lifestyle, is that already music, photos, some videos, the accessibility to have a device in your pocket, the accessibility to call these things up when you want, already many people are just taking that for granted. And with what we call Internet TV it goes to a point where even that, the medium that requires the most bits, the most bandwidth will be utterly different and ideas like channels will be a thing of the past. The idea that when you watch a news show everybody would see the same subjects in the same depth, we'll look back on that as a very inefficient thing. If you care about certain sports you want to see a lot more of those. If you care about the weather in certain places you want that in depth. If it's certain topics in politics or international affairs you care about, great, that should brought to you. Even the advertising should be targeted so that it's more interesting to you, and more effective for the advertiser, as well.
So the digital work style, digital lifestyle are moving ahead are moving ahead at full bore. And the one thing that will determine how quickly this happens is the software, the software presenting these capabilities in a fairly straightforward way.
Now, for Microsoft it's key that we get great people to help us drive this. In fact, everybody in the company gets involved in recruiting from time to time. And I made a little video of one of my recruiting experiences. So let's go ahead and take a look at that.
( Video Segment.)
BILL GATES: We had a lot of fun making that; he's a good guy. (Laughter.)
Well, let me quickly give you a glimpse of some of these things that I've been talking about. This is an example of a next generation phone that comes out. You can see the thinness is getting a lot better, the ability to have keyboard capabilities, and it's got that camera built-in, and so really the magic now is going to be next generations of software: speech recognition, image recognition, the model of the user and their communications needs and wants.
I've got another quick thing here on a PC that just gives you an idea of steps forward in visualization. I've got a set of photos, I can pick any set of those. I can say that I want to select a subset of these things, I can go in and out in terms of what's available here. And there's different views that we can have. For example, if we want to have an album view, it will take and put them there, I can organize them however I want. Or I can take and have a view which I'll call a mantle view where it puts them in a little bit of 3-D. I can play that to show how it would look on my, say, living room display, and anything that's interesting I can zoom in on that. Of course, I'd have data about when this was taken, who I was.
And so very quickly the idea of organizing all the photos you're going to have, which will be a big number, throughout your life you'll take a lot, you'll want to do different cuts on those and present it, having it dropped in and be easy to view, nicely accessible, that's going to be a very, very important thing.
Next let me shift to this little device here. This is the Xbox 360, which will be coming out next month. It's our second generation of videogames, and this is the generation that's really about high definition. We'll have our 360, Sony will have the PlayStation 3, and we'll have a generation of games that are far better than what's come before.
But it won't just be the games and entertainment, these will also be devices that will enhance this digital entertainment idea that we've been talking about, so that the power is here that all of our digital media you can get at, getting at, say, video on the Internet or even things that you carry around.
A good example is let's say I've got a portable music player and I want to have that music in my living room, I ought to be able to just take and use the standard USB connector here, plug this in, works with any kind of music player, even a little iPod can connect up here. (Laughter.)
As I plug in, I see the interface I have on the Xbox here. There's different screens that let you control the games that you play in, what you do with those. So here we have our system profile. Here's this media screen and so now with this one let me step down and say, oh, what do I want here, we actually have music, select that. And as I go in, you can see it's recognized that particular device I had, which is an iRiver device. I can select that and then just immediately it's going out, looking at all the different music that's there. We can just go in here and it's connecting to the device, pulling up the complete directory, and then I just go ahead and play whatever songs I have.
But it's not just music. Let's say I have a camera that I've gone and taken some photos on that just happens to be a Canon camera. Again, I've got this nice standard connector so I can take the USB and plug that in. And then as I go back up here, you probably saw we had pictures as another thing I can connect to, so here's pictures. And you can see again it's recognized that my device is out there, so I'll take that, select it, go over, see all my photos come up, and then I've just played a slideshow.
So pretty quickly without knowing a lot of complex commands you've got the power of this device with your music and you can play that music while you're playing a game, you can have this slideshow, you can have special effects coming out of the music, and so it's more than simply a game machine that you might have thought about videogames just being in the past.
With that, let's switch to this machine, turn the volume down low. This has got one of the games that will ship when the 360 comes out. This is called Project Gotham. Some of you may have seen it, it's a racing game that was available for the previous Xbox. But you'll see here it's dramatically changed because of this graphics power that's been made available.
So I'm not that great at this, I'll pick easy -- (laughter) -- and then it's going out and loading it in. It's got rich models of all the different cars. We're actually going to be racing here in downtown Las Vegas. We'll have lots of different cities, lots of different levels, you can race against different people. We actually have a lot of artificial intelligence in here in terms of how it's done.
So here we go, my guy is in fourth. It's actually the corners that get hard here. (Applause.) These spectators are good; whenever I race into the wall, they just fly everywhere, very realistic. I can also turn the driving over to the computer. All the detail is very, very different.
Also it's of course connected up to broadband and so if I want to find friends to play with, if I want to get into contests, if I want to be a spectator, all of that makes it a far deeper experience than you would have ever had in the past.
And, in fact, the whole player thing opens up these genres, so we'll have role-playing games, virtual reality. We'll also have very lightweight games, we have Xbox Arcade where you can come in and in just a few minutes be playing a fairly simple game. We'll have the old classics brought down with a little better graphics, and so it's easy to get into, it's not just the games that take a long time to learn how to use.
We even connect the Xbox up to the PC so if anywhere in your home you've got a Media Center you can project out on through the Xbox onto that living room device everything that's going on with the PC; so really taking the idea of that home network and broadband and bringing TV, gaming, PC, music, photos together.
The last demo I'll show is a little bit more futuristic. Here I've got a setup that let's say it's a table in an airport lounge, and I'm on a flight and I've got just my phone with me, I didn't happen to take my Tablet this time. Well, the phone has got a limited screen size, and so if I want to do a lot of rich interaction, it's not quite that productive.
So here I'll just take my phone, put it down and actually what's going on is there's a camera here and a little infrared light that lights this up and so it's recognizing that there's a phone and then talking to it and saying, OK, that's my phone. Then to open it up on this full table and let me get at the information, it forces me to use my fingerprint here and actually authorize and say exactly who I am. And now I have a bigger service area. I can annotate with ink, I can read large documents, that kind of thing.
Somebody gave me a business card that I want to follow up on, so I just put that down on the table and again the camera scans that, sees the text that's there. I'll flip this over because I actually wrote some notes on the back, that gets recognized. Then I'll take it and say, okay, I'd like it to be up in this phone, so it shows me that it's taking that data and putting it in my contact list. Obviously it will go and synch that up to the contacts I have, so it's on all my devices; the software is working cross-device, not just on one device.
It notifies me I've got a fairly urgent message about a press release that I need to approve, asks me to verify that I really do authorize this to go out because that's a very important thing. So again I simply use my fingerprint and that goes off.
So I could interact, view mail, browse the Web, things like this just with this device. Then when I'm done, I pick it up and it will recognize that it's gone. And so I'm logged off and all my information is simply on this device, so I'm not leaving anything behind.
This may seem pretty far out in a way, but, in fact, the componentry involved here is very inexpensive. These digital cameras are getting down to be very, very inexpensive; the actual projection display, the technology there is more and more based on single chip digital light projectors, so we can expect that to be cheap. So not just in the business environment will you have these intelligent type tables but even in the home as well. And so we can think of a play table and a business table and this type of recognition becoming actually kind of pervasive in the environment.
So where will software be used? Well, it's almost hard to think about where it won't be used. In fact, if we think about the sciences and the frontier in all the different sciences, every one of them is dealing with lots and lots of data. And how does the world deal with the fact that when you want to find patterns in data that humans aren't very good at that? Well, in fact, we take state of the art software, data mining software and machine learning type software and we apply that to the problem.
A good example is astronomy where you have different observatories, different locations, different wavelengths, different timeframes. If you want to say something about astronomy, you really need to check with all those databases. And so being able to think of that as one virtual database, that's a tough software problem. One of our top researchers, Jim Gray, brought the astronomers together, got them to do that, and it's made a huge difference in terms of that field moving forward very quickly.
The same thing is happening in all the sciences, biology is probably the most interesting, although the complexity there means that it's a much tougher challenge. You have, of course, lots of complexity in terms of privacy, you have people who have proprietary data there, but the big advances will come because the genomic data, the proteonomic data, all of those things will be combined into databases that people will sit down at workstations and navigate those things, look for patterns and try out different ideas.
And so software is driving the sciences forward. In the same way that math has always been a common tool, now software is a necessary part of that toolbox and so breakthroughs in software and people who understand software are very necessary for that interdisciplinary activity.
So when we think about computers and software, we often think about the richer countries and how they're doing a good job of taking advantage of these things, but we can also see this as a tool for equality, that as soon as a student in any country has a connection to the Internet, they don't need that huge library with extensive books. If they're connected up, they're on an equal footing, they can get the latest research and latest information.
Now, to really make that true, we have to have machine translation. That's another one of those Holy Grails of computer science where as we've made progress, our respect for human intelligence just goes up and up. And yet the progress is good enough now that in many domains, including text about science and computers, automatic translation is nearly as good as human translation.
We test this out by taking some of our online support articles in things like Japanese, Korean, and Spanish, and do half of them by hand, half by the computer, and then look at the satisfaction, the pattern of usage and see now that we've driven that to the point where it's equally acceptable to have those articles be machine translated. So that means a student, no matter what their language is, could go out and browse most of the Internet.
Now, when it comes to things like poetry, translating that automatically, that's still well out into the future, plenty more to be done there.
But the Internet is a leveler, it's a leveler of making information available, so it's certainly been a factor in driving towards more democracy, not letting political restriction limit the flow of information. It's also been a great enabler in terms of education, letting people try out their ideas, pursue their curiosity. Even in areas where people have disability, the blind now through software that we and others have done can get at and get all of that Internet material, not just be restricted to the Braille books that were very limited and not very available. Likewise for people with disabilities, the idea of the computer and special software allowing them to get involved in jobs that simply wouldn't have been available to them.
And so software is the place where the action is. It's a fascinating area. It is an area that will continue to generate jobs. The growth in Microsoft's employment here in North America is continuing to be very strong, also very strong on a global basis. We know that for the next decade there will be a shortage of great software engineers, and so we'll be scouring everywhere we can to find those people. And these are jobs where you get to come in and very quickly work on tough problems, very quickly find out that when you ship software to millions of people, as soon as it gets out there they're going to tell you what they like, what they don't like, you're going to have that feedback and an opportunity to take the software to the next level.
More and more we'll have different kinds of cycle times. We'll have things we improve literally automatically on an almost daily basis, things that improve monthly, things like, say, the browser or the Media Player that are more like the ones here, and things like the file system or the scheduler that are on more of a two or three year type cycle. By layering the software, having very provable component boundaries, we can make the efficiency, the ability to release separately, the depth of the security far better. And that's requiring very state of the art techniques in computer science to define interfaces and do proofs around those interfaces. In fact, these are things that were talked about over 20 years ago, but now because they're necessary, now because of the advances we're really doing those things.
They're very interesting problems in computer science that are now very practical. For example, the chips that are coming along, the clock speeds won't be going up as much, so we'll have lots of parallel execution, and so there we need new programming languages, we need new ways of looking at dataflow so that we can take advantage of that.
So this is the golden age of software. We can say that all the 30 years of work up until now is just to get the platform, just to get that we have the connectivity of the Internet, we have the great hardware that allows us to be more ambitious, and I'd say the thing that will be the most fun for me is over these next 10 or even 20 years seeing how your generation comes into the software industry and takes your creativity, your open-mindedness and takes software to a whole new level, so that's going to be very, very exciting.
Thank you. (Applause.)
TOM COLEMAN: I'm Tom Coleman, the Dean of the Faculty of Mathematics, and it's my pleasure to deal with the next session of this show, which is the Q&A; session.
Before we get started, I just want to say, Bill, that as you know, this is a very mathematical university and just in case some of the questions get a bit rough we have a little aid for you here. (Laughter.) This is a t-shirt from the math society with a few formulae, just in case.
BILL GATES: Excellent. Thank you. (Applause.)
TOM COLEMAN: OK, so we'd like to do it the following way. If you have a question, of course, on the bottom floor here raise your hand, I'll try to pick you out, and I'll repeat the question, maybe paraphrase it. On the top floor there are mikes set up, so just queue up there, I won't repeat those questions. I really can't see up there; oh yeah, there you are.
Okay, so why don't we begin? Raise your hand if you have a question for Bill. Yes, right in the front row here.
QUESTION: (Off mike).
TOM COLEMAN: So the question is the balance between theory and practice with regard to looking for a job in the future of software.
BILL GATES: Well, the key thing about a university education is to have a fairly deep understanding of the kind of tradeoffs that you want, understand algorithms, understand data structures. And it's not so important whether you know that in the context of compilers or graphics or operating systems, but rather that in some areas in a fairly deep way you've seen how programming works and the importance of very, very good design.
You know, the field of computer science today is very broad, and so some of those things will be very attractive to you, will be fun areas for you to exercise those skills, but as you move out and are doing work, the specifics will be available out there, and so the particular language, the particular way that you've learned it will be less important.
You know, when somebody comes to Microsoft we typically say to them, OK, what's the most complex piece of software you've worked on, whatever it is, get a sense of were they interested in the different choices that were available there and as they evolved over time did they see new ways that that could be done.
And so I'd put a strong emphasis on the basic theory of algorithms and how those things go on, but with at least one area that you get by hands on.
QUESTION: (Off mike).
TOM COLEMAN: So the question is, how do you see Microsoft changing the way you do business with such changes in all the software and tools that are coming out.
BILL GATES: Well, there are some things that have been true about Microsoft since it was founded 30 years ago that won't change. We believe in software, we believe in a company that hires the best people, does long term research around software, and that software will run on a variety of devices, from that wristwatch to the future set-top box to something in the car. As long as we stay focused on what we know, which is software and software platforms, there will be incredible opportunities.
Some things will change. In fact, instead of buying packaged software that you install one time and then you wait a few years until you might get another version of that, much more you'll be connecting up to a service that's delivered to you through the Internet. Now, underneath that service is the same kind of complex software and capabilities, but even something like Microsoft Office, we want to make constant improvement, we want to have you have templates that connect up and are relevant to the work that you do. And so you'll think of it more as a subscription or something that's funded through advertising than buying that piece of packaged software.
And for us we'll be, to the degree we're allowed, taking information about what features you're using, where you're frustrated, what messages you're getting and using that to be able to make the continuous improvements to the software.
So more of an ongoing relationship through that network connection, and in many cases eliminating the need to have a server that you run yourself or have a lot of IT professionals helping you out, because we simply take a lot of that complexity and we encapsulate that in the Web Services that we provide.
QUESTION: (Off mike).
TOM COLEMAN: So the question is advice that Bill would have to art students with regard to the new technologies coming out.
BILL GATES: Well, the pace of technological improvement is accelerating, and even for people in the field, tracking all the things that are going on is fairly tough.
I take two weeks a year, which I call Think Weeks, where I just go off and read about, OK, what's the latest in nanotechnology, what real progress is being made there, what's going on in the hardware field with satellites, lumileds, fabrication techniques, because all of these things come together and have a big influence.
I certainly think it's fantastic to have people who are not in the field watch what's going on, because there are so many issues that will come out of the advances in technology, issues of is there global harmonization in laws around privacy or e-commerce. To the degree that bad people are coming on to these networks and trying to take advantage of them in an improper way, what sort of rules and regulations should there be, how do we strike the balance properly for preventing those things and yet letting the network be fully available.
So there will be tons of policy issues that we want everybody to engage in and that those need to be written up and described in a way that doesn't assume that people actually understand the technology itself.
So this is the change agent, there's nothing else like this out there. So the fact you're using it as a tool, that you're aware of it, I think that's great for every type of intellectual endeavor.
QUESTION: (Off mike).
BILL GATES: Yeah, this is a topic that really deserves to be made more front and center and I think it is a little bit getting into the broad discussion. A recent book that's come out is one by Ray Kurzweil, who's a real pioneer and innovator. It's called "The Singularity Is Near". And even though I think almost everybody who reads it will find something that they disagree with or they think is overoptimistic or whatever, it's very important that books like that be looked at.
As Ray points out, most of this advanced technology comes in, in a very beneficial way. The fact that we can do cochlear implants now is taking people who would have been deaf their whole life and giving them back the ability to hear. We're at the very early stage of doing retinal implants and so that people who have visual impairments will have some of their sight restored.
And you could think of almost any part of the human body, whether it's an artificial heart, artificial knee, things like that, you can say, yes, we want the technology to help out with that. If somebody's got a spinal injury, if we could reconnect using electronics and let them regain control over their muscles, that would be a wonderful piece of technology.
Now, exactly where you stop in the genetic area, is it okay to create a twin who is born later, is it okay to be able to do organ replacements to help out with medical conditions, you know, you start to get out into these extreme uses of the technology, Ray is of the camp that says let it come and if we try to restrict it that's not going to be successful anyway, you just drive it underground.
I do think it gets scary when we have what he calls strong AI, that is, machines that are smarter than we are. That's not really very near but it's the kind of thing that is interesting to talk about. The next 20 years are characterized by the machine as a tool that helps us and aids us, but the fact that some time out in the distant future Ray would say 2040, which would be in the early range for most prognosticators, you really do start to have questions about how far you want to go with strong AI.
QUESTION: For programmers that would often be working just with a machine, for example, with assembly, the machine would be very important in programming. And as you were saying, software is continually changing along with the machine and hardware. Do you see that it will become less important that you work with the actual machine and more important that you work with the software platform as time goes on, or do you think that working with the machine, as we saw with your cell phone – I'm not sure exactly what it was called up there – but that as hardware changes, do you think that knowing the hardware will continue to be as important as knowing the software?
BILL GATES: Well, for most software developers the idea of actually understanding the hardware level will become less important because the operating system will be much higher level and far more abstract.
I do think it's important though as students when you want to get a sense of what's hard, what's easy, to actually have a sense of how the hardware works, particularly as we're getting new architectures that force us to understand parallelization in a deeper fashion.
Applications will be written at a very abstract level; in fact, we want to model things like business processes in a way that you just have a few sheets of paper that can take something like credit approval and describe that instead of the thousands of lines of complex code that are hard to change, which is the way we do that now.
If you really want to push the state of the art, say, okay, how will vision advance, how will algorithms advance, you should even do a little bit of machine type coding, driver type coding, and not have those abstractions hidden.
There's a discussion about should people ever work in a non-garbage collected environment. A lot of today's language just give you this sort of memory comes and goes type paradigm, which is nice, it avoids a lot of programming errors, but it means in a way that there's some efficiency considerations that you may not have full appreciation for.
So for people who want to do Microsoft level software and have a full understanding of the software stack, yes, the low level stuff should be part of your involvement. For a lot of people who go out and simply do business applications, that will become less important, eventually to the point where they won't really understand it at all.
TOM COLEMAN: OK, so if you want to work for Microsoft you have to do some of that low-level –
BILL GATES: Yes.
TOM COLEMAN: OK.
QUESTION: (Off mike).
TOM COLEMAN: So the question is concerned with the proliferation of formats and different devices as well and software, and how to deal with that.
BILL GATES: Well, what we see is you get periods of time where you have lots of divergence and then a few things will emerge as either de facto standards or, in fact, formalized standards, and then you'll move on to the next frontier where there's uncertainty about which approach is best, and so then you'll have a variety will flourish.
If we look at, you know, just look at the operating system level, in the good old days Burrows, CDC, Univac, IBM, you know, everybody had completely different operating systems. Even Digital Equipment had the top 10 operating system, they had VMS, they had RTS, RTX, dozens and dozens of those systems. Today, for most applications it's either some variance of Windows or some variance of UNIX that's actually being adopted, so there you'd see some simplification.
In terms of video formats, yeah, there's a lot of use of MPEG 2, which is sort of the DVD level. Then in this next generation we have two or three formats that are still being sorted out where each of those is very appropriate, but then we move down and things are more standardized.
One of the great advances is in XML where we can abstractly define a schema and then have domain specific information adhere to that schema, and yet the generic approach is universal, and so databases understand XML, Microsoft Office can import various types of XML. And so that kind of heterogeneity, which we could never handle before, now we have the way to do that and so in a sense that's reducing the number of formats because they're all just flavors of XML.
In terms of preserving knowledge and how we back that up and make sure it's still accessible, some of the issues you raised are important. The old Web, we don't have that archived and we ought to be more explicit because it would be good to be able to go back and get those resources. And so we're very involved in standards organizations that are taking document formats, XML schemas and trying to have as much homogeneity as we possibly can.
QUESTION: (Off mike).
BILL GATES: What was the second one?
TOM COLEMAN: I'm going to have a hard time repeating that question. (Laughter.)
QUESTION: (Off mike).
TOM COLEMAN: Open content.
QUESTION: (Off mike).
BILL GATES: Forever we've had a mix of free software and free content and commercial software and commercial content. The big breakthrough on the commercial side has been this very high volume, low cost model that Microsoft really represents where we want when you buy a PC, the software is only a small part of that cost. If you look at an IT budget, it's a very small part of the cost, and yet the productivity benefit you get in terms of us constantly improving it, monitoring it far outweighs the small percentage that you invest in that.
So in the content area I think in areas like music, probably most of the content will always be commercial where the bands want to buy food or whatever the bands do with their outcome, people who write long books, I think most of them will want to be paid. Again, hopefully it's a very small fee measured against high volume.
In the world of software it's a very fruitful balance. Some of the early UNIX stuff was done as free, then people did additions to that. They created companies that created jobs. I know Waterloo is very proud of the number of startups that have come out of the work done at the university. Usually the work done at the university will be some type of free capability, and the fact that you can then build on that and make it even better gives customers a choice, so they could back and use the free but the commercial will generally be the thing that has the services, the support, the richness, the depth, the evolution that goes with it.
And so the great thing about the Internet is these models can absolutely coexist. In a sense it's just different incentive systems. In the free world you have a certain prestige for contributing, in the commercial world you get to send your kids to school, you get to buy a house, and I think those incentives will still be fairly valid out in the future, so we love the mixed model.
QUESTION: I was wondering what predictions you can make about Apple's recent adoption of the x86 platform or soon to be.
BILL GATES: Well, I think Apple was forced into using the x86 platform because that had a roadmap that would let them do low power portables. And the device they were on before, the Power PC, didn't have a low-power roadmap that met up with that.
It really doesn't matter what microprocessor we use. After all, so much of the software is written in high-level language. We even do dynamic compilations, things like .NET byte codes when they show up in the environment can actually be late compiled and then bound into the environment.
There is a certain irony that the Xbox 360 is based on the Power PC so the architecture that Apple is abandoning is actually one that we're picking up and using and, of course, there will be tens of millions of those out there.
And so we work across the Intel architecture, both Intel and AMD, we use Power PC, we do a lot of ARM work. When you take these small devices where power is super important, there even the x86, today's roadmap doesn't have it getting down into the kind of low power that ARM has done. So ARM is really very predominant in those phone type devices, both StrongARM from Intel, but ARM from lots of other people.
So I don't think their switching their processor is a particularly big deal. They're one of many companies that we work with. We do Microsoft Office for the Mac, it's the best selling software on the Mac, and obviously we compete with them as well. So Apple is a great industry contributor.
QUESTION: Hi there. I believe that you have to be passionate at what you do to be successful. And today we saw just a few of Microsoft's products and services that you provide, there are many, many more. And I'm just wondering at a company where there are so many products, so many services and after 30 years of working at that company, where does your passion lie?
BILL GATES: Well, one thing that's great for me is that the PC that Paul Allen and I dreamed about 30 years ago still does not exist. (Laughter.) And today's machines are not rich enough, now powerful enough, not simple enough, not secure enough compared to what we thought about then. I mean, after all, most people are still taking their notes on paper, most people when they have sales figures print them out and sit there and can't dive into the numbers and understand them very well, we still have phone numbers, we still have TV channels. We still have physical media; in fact, the last great battle over a physical format is being fought now on how high definition movies will be distributed but that will be the last physical format because in the future everything will come across the network, both wired and wireless. And so by picking something that was very ambitious, I never got to the finish line and had to say what's next.
As well, the opportunity to work with very smart people across these different areas, the people in Microsoft Research, the product groups, that makes my job probably the most interesting in the world. I get to see how the pieces can fit together, and so the Xbox team will come in and say how excited they are about the graphics, and I'll say, well, should we connect that up to Media Center, and they'll say, "Ah, maybe" and I'll encourage these groups to work together.
So making sure that it sums up in a very coherent way, making sure we take the advanced research and get that out into the products very quickly, which many companies have failed at that, and that's why we see a systematic under-investment in research in our industry and probably across all industries, we want to be a model that that can be done very, very well.
So I'd say my job today is as fun as it's ever been. I am complimenting that somewhat with more time on my foundation, although that's still part time at this point. Again, that's about technology and the optimism of breakthroughs bringing smart people together and saying that we can get vaccines and new medicines to take on some of the diseases that are still killing millions not in the rich world but in the world at large.
QUESTION: (Off mike).
TOM COLEMAN: The question is, what personal characteristic do you think most influenced your success.
BILL GATES: Well, I certainly didn't expect to have the kind of success that I've had, and so certainly there's an element of being lucky enough to be at the industry at the very beginning where there weren't other companies and to know Paul Allen and share this vision and be willing to take the risk of starting the company when most people thought the ideas we were pursuing were pretty crazy ideas.
I do think that some early habits I had in terms of liking to read lots and lots of books, liking to have a broad set of curiosities, so not just about computers but also about the fields that touch on it, enjoying hiring in smart people and working with them, being competitive, thinking, hey, how come this product can't improve, even though it's doing really well. I think we've done well at not looking back on our success, always being willing to obsolete our products and drive forward.
At various stages of the company I've had to learn new things. In the first five years I didn't let any line of code get out of the company that I hadn't reviewed, and most people's code I didn't like as well as mine, so I'd mostly just rewrite it. (Laughter.) And I knew it all by heart and I didn't waste bytes. Now I don't get to review the code, and not only do we waste bytes, we waste megabytes. And so stepping up to work more as a manager, eventually work essentially as a manager of managers, more about vision and the dreams we have, really making sure the IQs add up, that's still a challenge I come in and think about every day, because we have so many bright people that if we can get it all to come together, then we can drive that forward.
So I do think some level of curiosity and intelligence and the hard work mixed together with a fair bit of luck can lead to something pretty magical.
QUESTION: You mentioned there is a golden age of software to come, and so what technical skill set will you be looking for in the new generation? And based on that, what type of advice would you give to university students?
BILL GATES: Well, my simple advice is that you should look hard at having a career in software and preparing yourself for it. We really do need the best and brightest minds.
The particular curriculum you go after isn't that important to us. Once you get to Microsoft you can get immersed in the area that you pick. If you work in database, great; if you have some experience with it, that's helpful, but it's not absolutely necessary. If you've got the kind of mind that likes to think about why does it work this way, how fast is that, what resources does it use, was there another approach that could be used, enjoying taking your code or your ideas and having other people critique those things and constantly pushing yourself forward.
I think it's great that Waterloo thinks of math and computer science as very interrelated disciplines, and so good computer software people are people who like mathematics. It's not pure mathematics, it's not purely proof type things, but the discipline of mathematics, which was my early interest, absolutely informs the idea of understanding what can be done. What is the fastest algorithm to sort numbers? Well, that's kind of a mathematical proposition and yet that's the kind of thing that a software person should always have in their minds.
So a real understanding of the different choices that get made and then having fun, you know, you're only going to go into software if you pick projects where you can see that it's fun and you understand what a social level it is.
I guess that's one other thing I'd add is that very few of our programming jobs involve just shutting your door – we do give people offices – and just sitting there and coding. That's a good thing, but most of it is about working with other people and understanding what the customers want and working with those other groups. And so I think we haven't as an industry talked about how those social skills and team skills are very important as well. And those are some of the toughest things to measure, but once people do get into work, that is a key thing for a certain way of moving up and taking on more responsibility.
TOM COLEMAN: Bill, I loved that Waterloo math message, that was great. (Laughter.)
OK, we have time probably for one more question.
QUESTION: (Off mike).
TOM COLEMAN: The question is, what is it that inspires Microsoft.
BILL GATES: Well, hopefully the inspiration comes from a number of places. The first thing is that we have these tough problems like machine translation, speech recognition, ink recognition that I wrote those down 30 years ago, I'll write them down today, we've made good progress and literally those things will be used by hundreds of millions of people over these next 10 years. So that's kind of a technological viewpoint when you prove a program correct and you compress video to a higher level than you do today.
Another source of inspiration, hopefully a very strong one is what the customers are saying, where they're saying I have too many remote controls, I get too many error messages, I don't know if this thing is connected up to the Internet, why do I have to do this detective work to figure that out. In some things like fonts and printer drivers that were problems ten years ago, today are pretty much done, but there are plenty of new ambitions people have, so we're always kind of at that forefront. People don't like spam, they don't like phishing; we have to make those problems go away, just completely hide them from the user so they don't think about it.
We do have partners, you know, Intel will come to us and say things like, hey, the clock speed is not going up much past 5 gigahertz, and so, "Hey, Microsoft, solve parallel programming;" we say, "Oh great, that's really easy to do." (Laughter.) "We'd better start thinking about it." And so partners come to us with things that they'd like to see us do.
We also have a lot of smart competitors out there. When Google makes a great search engine, then we step back and say, hey, it's actually not very good. (Laughter.) You shouldn't have to click on all those links and have this treasure hunt, the computer ought to know what you're asking for and bring that to you, it ought to understand your background, your interests, things you've done in the past, and literally be about very direct answers coming to you. So that competitive sense, whether it's Sony in videogames or Oracle in databases, Google in search, that spurs us on as well.
And by having lots of people, the advanced researchers, lots of young people coming in, I each summer get some time with the summer interns at Microsoft because I love hearing what they're saying that they think we're missing out or that we ought to be doing more of, because they're exposed to the world in a bit of a different way. You know, instant messaging was not something that – this goes back quite a while – that I was personally using but that emerged as something super important as a complement to many communications approaches.
So we've got to be very open-minded and make sure that we're actually using the tools of software to get all these inputs and bring them together and then small teams pick exactly what they're going to do, what they're going to go after.
TOM COLEMAN: OK, well, sadly, we've finished our time and let's thank Bill again for coming here and answering those questions. (Applause.)
Bill, what a good note on which to end, the rejuvenation that comes from the young. I'm sorry that we don't have you for a whole day, only an hour and a half, but please come back next year same place and we'll pick up just where we left off. (Laughter.)
BILL GATES: OK.
TOM COLEMAN: Bill, this is a small token of appreciation. I think it's a day for sweaters. Now, this one is a little closer in color and logo to mine than it is to yours, and I only wish we'd given it to you yesterday so you could have worn it at the University of Michigan. (Laughter, applause.)
BILL GATES: Great. Thank you. (Applause.)