And the collaboration that's going on today, whether it's between
Microsoft and Stanford in areas of software advances or between my
foundation and Stanford on global health things, really are
I spent the afternoon meeting with faculty, talking about the
progress on those things, and sharing our ideas about where we go
from here. Really it's exciting because my optimism about technology
is deeply underscored when I meet these brilliant researchers, and
see that they are going to get the resources, and take on these
John mentioned that the middle of this year is a change for me, that
I'll switch to being full time at the foundation, and part time at
Microsoft, and that could be traumatic for me. I was 17 years old
when I started working full time on Microsoft work, and I've done it
basically every day of work since then. So, who knows what it will
be like to make the change? I'm looking forward to it, and some
friends said that they'd like to volunteer to help make a little
video so that I'd understand what my last day will be like, and how
things will change. So, let's take a look at the video they helped
BILL GATES: Well, we certainly had a lot of fun making that, but the
transition is going very well. You saw Ray Ozzie and Craig Mundie in
the video, who are taking over a lot of the things I've been doing.
And I'll still be very involved in some things I've had a lot of
passion about, including natural user interface, some things about
how we structure knowledge, and really take on the big frontiers in
Let me talk now about what I think software will do in the decades
ahead. Certainly if you go back to the start of Microsoft, nobody
thought of software as being important at all. There was no software
industry. The little software that there was, was simply bundled
along with the mainframe for the very expensive computers. And
computing itself was almost thought of as a threatening scary thing,
where governments and large companies would use it to track
information about you and to print bills that were never right.
People talked about stapling the punch cards that came with your
bill, if you've ever heard of a punch card, and messing up these
And so it's a real mind shift change to say that computing was going
to be about individuals, that it was going to be about empowerment,
and that even more important than the hardware would be the software
that was available, and that a gigantic industry would group up
Now, that dream required some heroic assumptions. We had to believe
that the cost of the hardware would come down. We had to believe
that the volume would go up. And only then would the economics of
being able to spend tens of millions of dollars to write a software
package, and yet being able to sell it for say $100 or less,
actually make sense.
And so we undertook the idea of reaching out to other people,
getting them to start software companies, and making sure that the
personal computer became that high volume platform. In fact, today
the software industry is gigantic, and the range of solutions and
creativity in that industry is absolutely phenomenal.
That's really changed the way we think about computing. Today, we
think about computing as affecting almost everything. Ten years ago,
I talked about the start of the first digital decade. That's about
the time where the Internet was just showing up, and nobody was
doing their photography in digital form or banking online or
organizing their trips or looking at stock results. Well, today, 10
years later, many of those activities, certainly in the rich
countries, we almost take for granted. The idea of a printed phone
book or a CD or a record almost seems antiquated.
My daughter doesn't know what a record is. I keep meaning to go find
one and show her, but they're hard to find nowadays. Soon enough
things like the phone book or a print-based encyclopedia will be
So, we're now at the start of what we call the second digital
decade, and I think the changes, the impact of this second digital
decade will be far more dramatic than the first; in fact, as
dramatic as all the things that software has done in this entire
30-year period since the personal computer came along.
Part of that is because of the foundation we have. We have over a
billion personal computers out there, and several billion people
who've had a chance to use those. We have several billion people who
use cell phones. We have somewhat less, about 300 million people
connected up to the broadband Internet, but that's a number that
keeps growing quite dramatically.
We reached an interesting milestone just recently where China now
has more broadband users than the United States. You can be sure
that the United States won't catch up, because China has a lot of
people who are going to be connecting up. In fact, of all the IT
related markets, the personal computer itself and software are the
only ones where the U.S. market is still much bigger than the market
in China. So, it's a very, very global business in terms of where
the talent is, where the innovation is, where the markets are, and
different ways of using these tools to have an impact.
I think it's fantastic that the Internet has made the world a
smaller place. The growth of the personal computer from a device
that you create documents and you edit them to one where you can do
a little bit of e-mail to one where you could get a little bit of
content to now where almost everything should be digital by default,
that is a big mind shift change.
For any industry it has huge implications. Even for education, which
we think of as being about the same for the last several hundred
years, if somebody said to you, the best math teacher lived in 1890,
you could say, well, maybe that's true. You couldn't say that about
the best person who understands physics because there's been an
accretion of knowledge, and people are building on each others'
In some areas the ability to watch people who practice very well, to
see their results that are numerically analyzed, to understand what
those techniques are, it's been difficult to create that learning
cycle. And now that we have things digitally, that we can store
videos digitally, that we can look at test scores and correlate
things, and other teachers can see what those teachers are doing and
try out those best practices and find out how well they're working,
then even in that area we get a fairly substantial change.
Education will probably split into different things. For example,
how many universities should have to give lectures on subjects like
physics? Well, the answer is very few, because whoever does that
well can put it out on the Internet, make it available for free --
and there's certainly a trend towards doing this -- and everybody
the world over, assuming it's localized into different languages,
which again using labor across the Internet again should be a very
straightforward thing that even volunteer labor in most places would
be able to achieve, then you have the best teachers of all the
college curriculum available on a worldwide basis.
I personally go up and watch courses about physics or chemistry or
anything that I want to know about -- I have to admit they're mostly
MIT courses at this point, but I look forward to seeing more
Stanford courses up there -- and I feel so privileged to be able to
Not a week goes by that my children aren't asking me some question
that I go up to the Internet and say, okay, what is it about stars
or different animals that I can take a story back and actually be a
dad who knows the answers to these questions, and encourages their
curiosity. None of that would have been possible before.
The ambition level we can have for different realms of activity
should be much higher, and it's because of what software can
achieve. If we think of somebody who works in an office, today they
are really information starved: their ability to navigate, to
understand customer trends and quality and costs and opinion, even
to survey information and look at how that's changing over time,
look at key indicators that make sense to them, to collaborate with
people at a distance. You can just talk to these what we call
information workers about how valuable their time in meetings is
spent, how hard they find it to get data, and you understand they
are not yet fully empowered.
The way that even just communications works, where they think about
phone numbers and busy signals, and shall I send e-mail or instant
messaging, the way that when you're at a distance you can't really
meet and collaborate in a rich way, it is very, very antiquated.
It's way better than it was say 10 or 20 years ago, but it's nowhere
near to what it can be.
When you start to take products like cars and planes or any physical
product, do the design digitally, share those plans around, let
people try out simulation models, what might happen with that
product over a period of time, you're shortening design cycles.
So, you take the fact that there are more educated people on a
global basis, that they're connected, and that the power of software
will give them better tools, not just to work together but also to
model and understand the nature of the product work they're doing,
then innovation will accelerate, and it will accelerate on this
foundation of the advances in computing and software.
Why can we be so sure about this? I mean, after all, when Gordon
Moore first predicted that the number of transistors would double
every two years or so, it was just a prediction.
Well, we can see that that prediction in transistors will remain
true for the next 10 years. We can see that the storage capacity
will also grow at exponential rates, that optic fiber bandwidth will
go up at exponential rates.
If we look for bottlenecks in this, we only see them in a few
places. We see a bottleneck in terms of clock speed of the
microprocessor. So, finally we have to deal with programming
computers to work in a parallel fashion. It's one of those great
problems that when I was in computer science we thought, hey, maybe
we're about to solve this. Well, now we really have to solve it. The
brute force of clock speed scaling is not likely to come and bail us
out the way that it has in the past.
There are some issues of modularity and proving programs correct
both for just reliability and security things that are also now very
required as we're taking all of society and connecting it up
digitally, with financial records and product orders and private
medical data, all of these things being stored digitally.
So, the basic foundation of how we understand the way that software
works to say does this software maintain privacy, does this software
control this information properly, there are some fundamental
advances in computer science that we need to drive.
Most of the things we want -- cheap screens that for US$30 or $40
you can take every wall in your room and project up something at
very high resolution; cameras that when combined with software can
recognize the kind of gestures that you make, and who's in the room
doing what -- these things will be very inexpensive.
And, in fact, we're on the verge of a big change of how you interact
with all these devices. It's been the mouse and keyboard
overwhelmingly. It was just the keyboard, but then the mouse became
mainstream, actually invented not far from here in the '60s by
Douglas Engelbart, but then with graphics interface that came and it
was standard. That is the way we interact: You sit down at a chair,
it's really just one person.
You're starting to see the beginnings of a change to a broad range
of interaction techniques I call natural user interface. You see it
in the 3D controller that the Wii has.
You see it in the touch that the iPhone has.
You see it in products like Microsoft Surface where we have cameras
that can look at any gesture, any object that's appearing, and
seeing what you're doing.
You see it in RoundTable that sees who's in the room and decides
who's speaking by taking these multiple camera feeds.
You see it in products like the TellMe software that runs in mobile
phones where instead of trying to use that keyboard, you simply say
what you're interested in, whether it's a directory lookup or a
software interaction, and it recognizes that speech.
We now have the power to perform natural user interface.
A form factor that I'm a big believer in, that I'm excited to make
sure we keep investing in, to drive it so it's attractive to the
mainstream, is the tablet device. This is where you can read off the
screen, that it's light, cheap, long battery life; eventually a
replacement for paper-based textbooks.
My daughter goes to a school where they use that Tablet PC, and they
use the pen, and they're very adept at it, and it's amazing to see
how they kind of learn in a different way, because they have that
There's still a lot of work to be done to get that down to the say
several hundred dollars and the lightness and battery life that we
need, but that is absolutely coming. It's a fundamental tool that
will change the consumption of learning material, and even in the
office place will be the device that you have as you go off to
We now talk not just about computers on the desk but computers in
the desk, because we can recognize what you're putting there, and
let you touch and expand things. Your desk will just be a horizontal
surface display, your whiteboard will just be a vertical surface
display. So, the ability there to take business information or
project schedules and touch and manipulate and see those things, and
then have a portion of it that's a videoconference with another
person where you're working together and interacting, that will just
be commonplace. When that's cheap, people will go to that, and we
need a whole new generation of software that can interact and use
In the consumer space experiences like TV, which are very passive
today, very channel oriented, will change to be very personalized
and very interactive. The dichotomy of broadcast video that you have
say through your cable or satellite provider, and the video on the
Internet, those will be brought together, so that if I have a child
who's in a sports even and somebody with an HD camera just happens
to go and film that, when I go back to my TV menu, based on my
interests, that will show up as one of the top big choices that I
might be interested in. There's nothing that will divide those two
worlds. The advertising will be targeted, the shows will be
interactive. Something like watching the Olympics and picking which
sports you're interested in, or the election and seeing the
background and the breakdown of what's going on with different
votes; we are so used to a very limited TV experience, that this
revolution that is literally on the verge of happening, people don't
really appreciate how dramatic that's going to be.
Today, there's a few million people that are getting their popular
mainstream video through pure Internet feeds that can be
individualized, and so that infrastructure is starting to get out
there. As we get it to the tens of millions and hundreds of
millions, then all the content programmers will realize that the
dividing line between what's kind of a set-top box in this
environment and what's been a videogame, there is no dividing line;
it's just a spectrum of content.
So, many of the things that will be available on TV in terms of
watching together and chatting with your friends who are at a
distance, or trying different things out, some of those you see more
in the videogame world today than in the TV world, but we think
they'll be very broadly adopted.
Things like organizing the memories of your children as they grow
up, and having the images and the homework and the exchanges with
them, and being able to go back and view that in rich and fun ways,
that can happen very automatically.
Today, we're still very device-centric, and we rely on the user to
move information between their phones, and their phones and their
PCs, and their PCs and their PCs. Well, as we get this sort of
unlimited power in the cloud, both in terms of computation and
storage, the ability to move that data automatically so that if you
buy a new phone your information just shows up, if you borrow a PC
your data is there but only available to you, that will become
So, the willingness to work with multiple form factors, even in the
car where it's more voice oriented, or in the living room where it's
more distance, 10-foot oriented with gestures and a simple remote
control, or using your phone to control things, those experiences
will not be bifurcated like they are today.
Now, we also need to revolutionize how we write software, where we
can define things at a much higher level. That really hasn't changed
much in these last 30 years. We're still writing declarative code
that can take something like two banks whose products are 90 percent
identical, and you can end up literally with a million lines of code
that are different between these two banks. And yet if you describe
say in English their products, you'd only find like 40 pages of
And so you say, what is that explosion of complexity that is
expensive, it's fragile, it's hard to prove it's correct? Well, it's
a failure of abstraction. We have not changed that level of
abstraction. And finally we have the computing power and some of
these ideas that can create runtime environments that particularly
in domains that you focus on like the business domain that so much
software is written to, we can make some huge breakthroughs.
This is part of the reason why one of the best investments I think
any company makes is in its research group and in the way that
research group connects up with universities. It's something that
Microsoft looked forward to doing, and about 15 years ago we were
successful enough we were able to start down the research path.
People like Nathan Myhrvold, Rick Rashid came in and built something
really phenomenal that not just in terms of the research it does but
in terms of the way it lets us understand the brilliant ideas at
places like Stanford, it has made a huge difference for us.
Every one of our products is dramatically better because of that
work. New things we do like ink recognition in tablet or all this
visual recognition stuff that is just coming to the mainstream was
totally developed there, our machine translation stuff. The
breakthrough work that gives us the belief that we can take what
search is today and so something that's dramatically better than
that, that optimism comes because we have great people in our
research group who are doing very advanced things.
We spend a bit over $6 billion a year on R&D, but it's really this
long term piece that ranges from graphics techniques to quantum
computing to natural user interface that really define what the
future is going to be.
We've now spread that activity across the globe. When people come to
me and say, hey, would you put a research center in a certain place,
I used to say, well, if you have a billion people, we'll put a
research center there, because we have one in China, one in India,
one in the United States -- now we have three in the United States,
so obviously I'm breaking my criteria that you have to have a
billion people. So, it will be a little harder to exactly say the
criteria, but it really has to do with where the top universities
That research activity is risk-oriented, and it's actually fairly
surprising to me how little research is funded by businesses. Even
here in the United States, if you take what Bell Labs and Xerox PARC
did, which are some of the foundational work that Microsoft
benefited from immensely, the entire personal computer industry had
a huge boost by that work. And unfortunately those companies didn't
get an economic advantage; the way they managed the research and
thought about it actually set an example that may have set back the
willingness of companies to make these investments. I think now you
see a range of companies like GE coming around and saying that this
is an important thing, but that's a huge challenge. Those are the
kinds of jobs and breakthroughs that really are going to change the
Now, when we think about the sciences broadly, the role of software
is becoming more important. In the past you could say, well, what
was the language of science? You could say mathematics, and it was
very important for physicists, chemists, biologists to have some
understanding of particular parts of mathematics to express their
ideas, to write down formulas, and to make predictions.
Today, the amount of data in most of these sciences is large enough
that we can say that computer software and databases and pattern
matching that come out of software breakthroughs are really
important for what is going on in the sciences, particularly in
biology, but I'd say almost as strongly for astronomy where the
amount of data and taking a theory about the density of things, the
creation of things, it's not just one telescope, it's not just being
there at midnight and seeing something cool and writing it up and
getting the Nobel Prize; rather it's deep analysis across massive
amounts of data.
So, we are sort of the handmaiden of those advances, and making sure
that we're reaching out and collaborating with the sciences, and
understanding from them how do they want to process that genomic
data, how do they want to take and get insights into it, that's very
We're doing our best to reach out to scientists, so getting
ourselves out of just pure computer science, which is very
important, lots of tough problems there, but to play a role in this
more interdisciplinary activity that's happening in a very deep way
in the top universities. In fact, in my discussions with faculty
this afternoon I was really pleased to see how Stanford is really
trying to push the limits of getting departments to work together,
and particularly bringing in computer science.
One area of complexity that I'm sure fascinates all of us is
studying the brain. There's a lot of great research going on in
that. One of the people we're working with and providing software to
is [Jeff] Lichtman at Harvard. So, I wanted to take just a quick
look at a short video about how what he's doing, and then show how
software fits into that.
BILL GATES: Yeah, I've got the HD View running right here, so you
can get a little sense of it.
Processing lots of image data now it turns out with the right
algorithms we can do this very well to let you scan in and out, and
even apply a lot of recognition algorithms to understand. Here what
we really want is a database of all the neuron connections inside
the brain, and eventually understanding exactly what's being
So, if we look here, this is the layer diagram, and I can go in and
look at individual layers at any time, try and understand exactly
what's changing as we go through that. A lot of data, but processed
very quickly. Then here's where we take an algorithm that's trying
to understand exactly what the patterns are, and then map that, as
was being said, into those structures.
Now, obviously this is just the beginning of this type of capability
to really get the model and understand what the meaning of the
messages are all the way up to the highest level, that's going to
take a lot of time, but that's a very software driven activity.
One thing that's amazing is in the computer industry and sciences
broadly is how much students have really been at the heat of a lot
of breakthroughs. John mentioned a lot of the great companies that
Stanford alum or dropouts have started, and there are other examples
as well. So, it's very interesting that at a young age people are
very open-minded about new approaches.
We announced a new program today to actually let students have all
the same tool software, things like Visual Studio or Expression, the
same software for free that professional developers use, really
trying to broaden that out, both to not only the computer science
department where we've already had grant programs, but to the other
departments and even down to a younger age level, so that this
access to the very best tools is there from the beginning. Some of
these people will go on and start companies, some will just be a lot
better in whatever activities they engage in.
This level of interest is very high. We have a contest every year we
call the Imagine Cup. Last year, it was about 100,000 students. This
year it will be about 150,000 students. The United States is the
third biggest country, where Brazil and India have a higher
enrollment, but the U.S. at 15,000 is very significant. And the
quality of these entrants are really unbelievable. In fact, we had a
thing where people won a programming contest, we would just
basically give them a job and some of the people who have come out
of that have been really phenomenal in terms of what can go on.
And what we're seeing is we're really getting to the point where
your level of education is what defines your opportunity. It's less
about where you grew up and simply having access to these tools, if
you're lucky enough to get access, then really the sky is the limit
in terms of what can be done.
So, this brings us to my final topic, which is the question of as we
have all these advances, how are benefits of those advances spread
in terms of the richest 2 billion on the planet, say the middle 2
billion, and the bottom 2 billion?
In fact, our record to date is that although there's benefits in
terms of improved medicine and food and electricity to a high
percentage of people, that the relative benefit has been
overwhelmingly to essentially the people who need it the least,
where the marginal benefit is lower than it is say in the poorest 2
billion, where literally for not spending a few hundred dollars a
child's life is lost.
Of the 12 million children that die every year, less than 1 percent
of them are in the rich countries, and yet if you look at medical
research that's related to those things, over 90 percent would
relate to the conditions that are in the richest countries.
So, you have this big disparity. Consider how much money should be
spent on baldness versus on malaria. Well, the ratio is about 50 to
1 for baldness. Malaria, of course, kills over a million a year.
I was pretty stunned when I found out about these statistics, and I
have to say it was after I dropped out of Harvard, actually quite a
bit, over 10 years, I read about a disease called rotavirus that was
killing a half a million children a year, and I thought, what the
heck is rotavirus; I'd absolutely never heard of it, this must be --
this article must be wrong. You can't have a disease that's killing
a half a million children and not have had courses. I flipped
through the course catalogue; I never saw anything, any of this
stuff. In fact, the one medicine there was for that disease, the one
vaccine was taken off the market because of things that really it
shouldn't have been taken off the market for, for the key target
market, which was the poor countries.
So, we have this disparity that as great as our system is, if
there's not a market need, it doesn't drive the innovation to the
particular requirements of the poorest.
And yet I think that's a very solvable thing, and, in fact, I think
there's an increasing awareness, a desire of people working at
companies, of companies, and of universities to have an impact
that's measured slightly in an additional way besides what the pure
market incentives are.
Our research group in India has a special group with a lot of social
scientists in it that goes out to the poorest and is looking and
talking to them, and very quickly you realize for that segment
there's no electricity, there's widespread illiteracy; you're not
going to give them a personal computer. I don't care if it's a 10
cent personal computer; the problem is very different than that.
So, some of the solutions they've come up with in terms of using
cell phones or even just using DVDs have been amazing. They take
these agriculture extension workers who go out and help farmers,
tell them what to do, and they come with a TV set and a DVD, and the
very best farmers have been filmed doing these things. Think of it
as like American Idol, except this is "Farmer Idol" and they really
want to be the ones to choose on the video. That technique has done
more for improving the productivity of those farmers, it's three
times as effective as just sending that person out, and yet they
don't need to be nearly as trained. So, some technologies like a DVD
player carried out to a village, when used in the right structure,
can have a very dramatic impact.
So, at every tier -- the bottom 2 billion, the middle 2 billion --
we have to think through what technology can work. For vaccines, you
have to keep them cold as they get out to these rural villages;
that's a very tough thing.
One of the things I'll be spending time on is reaching out to both
universities and companies, and encouraging them to get more
involved in this: food companies on micro nutrients and the ideas
they have about buying food and helping the small holder farmers who
represent the majority of the abject poor in the world; the pharma
companies in terms of doing more on these things.
I absolutely think universities have a big role to play here. One
element of it is that I don't think students should graduate without
having some sense, ideally both learning about it and having some
direct experience of it, of the average human condition in the
world, as opposed to the condition that we experience normally by
living here in one of the very richest countries in the world.
So, I think we can apply ourselves to this. I don't think it
requires a revolution, but it does require a focus, it requires some
value system that gets expressed, and some measurement, both in
terms of who's doing it well and who's not doing it well, that's
really going to drive more rapid change.
So, overall I hope you get a sense of my optimism about how
technology broadly and software in particular will become an
enabling element in the years ahead. I think it's a wonderful time
to be a student and to have gathered these skills, and so I'll be
very excited to see the great work that you can do.