Vint Cerf at the Residence Lecture Series

Uploaded by Google on 30.07.2007


MALE SPEAKER: Good afternoon, everyone.
Welcome to the Residence Lecture Series.
Oh, I turned on my hearing aid.
It sounds good in here.
We have an extra treat today because we not only have a
distinguished speaker we're are all looking forward to
hearing, but, as a representative from Google, he
and others have arranged for this talk to be taped.
So this is going to be a Google tape on
record after this.
That's what's going on in the back of the room.
The speaker will be introduced today by one of our Residence
Lecture Series committee members, Bill Anderson, who
has known our speaker Dr. Cerf.
BILL ANDERSON: I'm Bill Anderson.
On behalf of the Residence Lecture Committee, I am
delighted to have this opportunity to introduce to
you Vinton G. Cerf who will speak on Tracking the Internet
into the 21st Century.
The internet is used by millions of people around the
world every day.
If it was owned by somebody and they charged $0.01 each
time you clicked on, it would be by far the richest company
in the world, probably the richest country in the world.
If one of your grandchildren asks you, what is the
internet, how does it work, who manages it, who owns it,
how many of us can give the correct answers?
Vint Cerf will now give us all the answers.

He holds a bachelor's degree in science
and mathematics from--
guess where--
Stanford, as well as a master of science degree and PhD in
computer science from UCLA.
So in a sense, he's one of us Californians especially as his
personal tastes include fine wines and gourmet cooking.
He's currently vice president and chief internet--
listen to this now--
vice president and chief internet
evangelist for Google.
He is universally recognized as the
codesigner of the internet.
In this country, he has received the US National Medal
of Technology, the Turing Award, sometimes called the
Nobel Prize of computer science, and in 2005, the
Presidential Medal of Freedom, the highest civilian award
given by the United States to a citizen.
He has also received numerous awards.

He serves as chairman of the board of the Internet
Corporation for Assigned Names and Numbers.
You may have wondered who approves all the addresses you
and the website use and how it is controlled, so perhaps he
will tell us.
Vint has received numerous awards and commendations in
connection with his work on the internet from all over the
world including countries, such as China, Switzerland,
Sweden, Spain, Italy, the Netherlands, and so on.
I'm not going to list all his many honorary degrees, but I'm
sure you can look them up on Google if nowhere else.
So without further ado, please welcome Vinton Cerf, co-father
of the internet.

VINT CERF: Thank you very much, Bill.
And I have to tell you, I can't tell you what a special
pleasure it is to be here.
I'm seeing some faces that I haven't seen in a while, some
familiar ones.
And I expect I'm going to meet some new ones.
I'm staying here for dinner, which brings up something
called theorem number 208.
The theorem reads, if you feed them, they'll come.
And I'm being fed, so here I am.
The other thing I wanted to acknowledge is two people
especially who are in the room tonight.
Don Nielson, if you'll wave your hand, please.
Don was one of the principals involved in the packet radio
system, which I'll tell you about.
It was one of the three networks that fed into the
design of the internet.
And the other person is Bill Perry, who you all know is
sitting in the front row, just back from Japan, bless your
heart, sir.
Am I audible if I'm at this level?
VINT CERF: Then we're going to go this way until
my voice gives out.
Anyway, thank you, Bill, for doing this tonight.
The reason that Bill Perry is special, apart from many other
good reasons, is that he was in the Defense Department,
responsible for research and engineering during the time
that the internet was undergoing its rapid evolution
in the research world.
At the Defense Advanced Research Project, he could see
where I worked.
And from time to time, Bill was kind enough to stop in and
ask, what the heck are you guys doing now?
So we now know a little bit more about what we were doing
30 years later.
It turned into something bigger than we expected.
Let me explain a little bit about the title.
When I came to work at Google, they asked me, what
title do you want?
And I said, well, how about Archduke?
I thought that sounded pretty good.
And then somebody reminded me that the last
archduke was Ferdinand.
He was assassinated in 1914, and it started World War I.
And so we thought, maybe this would be a bad idea.
And they said, considering what you've been doing for the
last 30 years, why don't we call you our chief internet
So on the first day of work, I thought I should wear
something ecclesiastical.
In my closets are an outfit that I got from the University
of the Balearic Islands.
It's their formal academic robes.
And it was the most ecclesiastical outfit I owned.
So I showed up for work on my first day at Google wearing
this outfit.
And Eric Schmidt, the CEO, grabbed his camera and took
this picture.
So I'm not dressed in my formal robes, just in my usual
three-piece suit tonight.

Also, this is a completely unexpected extra
slide in the deck.
My wife and I spent a week in southern India, a
place called Kerala.
And there's a big lake called Lake Vembanad.
And on this lake are houseboats that
look like this one.
And I madly took pictures of these things because they look
like hobbit holes from The Lord of the Rings.
So I took a bunch of these pictures, and I sent them to
Peter Jackson whom I had met a few months before in New
Zealand and said, here, if you're wondering where all the
hobbits went, they're in southern India on Lake
Vembanad in one of these houseboats.
So if you get nothing else from tonight's lecture than
the fact that the hobbits are in southern India, you've got
something of value.
All right.
So let's talk a little bit about statistics.
If this were 1997, 10 years ago, I would be crowing about
the fact that there are 22.5 million computers on the
internet and 50 million users.
Holy cow, that's a big number.
Well, it was a big number in 1997.
But fast forward 10 years.
There are over 1 billion users of the internet and over 400
million servers, things that do email, and web services,
and the like, not counting things like laptops or
desktops or personal digital assistants like this
BlackBerry that are occasionally connected to the
net, not necessarily on all the time.
So there might be as many as a billion devices on the net
total from time to time.
Now another phenomenon, which has occurred over the last 10
years or so, has been a dramatic growth in mobile
telephone communication.
There are now estimated to be 2 and 1/2 billion mobiles in
use around the world.
The reason that's relevant to any talk about the internet is
that many of these devices are now internet enabled.
The consequence of that, which I'll talk about a little more
later, is that those of us who are offering services on the
net, like Google, have to learn how to present those
services through a relatively limited
interface on a mobile device.
So that's becoming very important.
Also, for many people in the world, they're first exposure
to the internet will be through a mobile, rather than
through a laptop or a desktop because the mobiles are less
expensive and more widely available.
I thought you would find it interesting to look at the
predecessor to the internet, the ARPANET, in 1969.
There were four nodes installed by December of 1969.
I was a graduate student at UCLA in Len Kleinrock's
laboratory, responsible for the software for this computer
called the Sigma 7, connected to the very first node
installed on the ARPANET.
The Sigma 7 is in a museum now.
And there are a few people who suggest that I should be there
along with it.
And in fact, if you go down the street in the Computer
History Museum-- and I know several of you are well
familiar with this, Gene Amdahl in particular--
there are lots of computers that I remember from my early
programming days.
But it's a little stunning to realize they're sitting in a
museum now.
In any case, that was the network before it became the
internet, before known as ARPANET.
Now this is the part I love.
Some people think that the internet happened because a
lot of local area networks got together and hooked themselves
into a giant network of networks.
Well, that's not what happened.
What happened is the Defense Department started to
experiment with a technology called packet switching, which
is what the ARPANET was based on.
This was a way of sending data among computers on dedicated
telephone circuits as if they were
little electronic postcards.
So if you think for a minute about what you know about
postcards, you actually know quite a bit about how the
internet works.
Because a postcard has a To address and a From address,
and it's got a little bit of content.
Well, an internet packet is just that way.
It has a source address, and a destination
address, and some content.
Now when you put a postcard in the postal system, there's no
guarantee that it comes out the other end.
This is what's known as a best-effort system.
And the same is true of the internet.
When you put an internet packet in, there's no
guarantee that it comes out the other end.
When you put two postcards in the postbox, there's no
guarantee that they'll come out in the same order that you
put them in.
This is also true of internet packets.
So you don't guarantee that they'll just stay in order.
So we have this interesting problem that you have a
network which potentially loses things, it potentially
gets things out of order.
And in fact, it even does something that the post office
doesn't do.
Sometimes when you put a packet into the internet, two
of them come out on the other end because they may go in
different paths.
And actually, both of them get delivered.
Well, as far as I know, the post office doesn't do that.
But here we have this relatively odd design where
the data is flowing through like millions of little
internet postcards.
And it doesn't look very reliable at
the internet layer.
So we put new layers of procedure above that.
We call these protocols.
And the one above the Internet Protocol, or the IP protocol,
is called TCP, for Transmission Control Protocol.
That's the thing that retransmits if something gets
lost, puts things back in order, and throws away
And the way it works is pretty simple.
Imagine for just a moment that you have a book, and you want
to send it to your friends.
So I'm going to send Bill a book.
But I'm only allowed to send it through a postal service
that carries postcards.
So my problem is how do I send a book through a postcard
transport system.
Well, the first thing I'll have to do is tear pages out
of the book and cut them up, so they fit on postcards.
Then I recognize that, because I cut the pages up, not every
postcard has a page number.
And since I know they might get out of order, I decide I
better number the postcards, one, two, three, four, five,
six, so Bill can get them and put them back in order if they
arrive out of order.
Then I remembered that some of them are going to get lost or
might get lost, so I keep copies in case I have to send
duplicates to recover from the loss.
And then I wonder, well, how do I know whether I should
send any duplicates?
Well, I could ask Bill every once in a while to send me a
postcard saying, I got all the postcards up to number 420,
please start sending anything that was missing.
Then I realize that that postcard, in the middle of
sending, might get lost. So now what?
And the answer is you look at your watch, and you say, how
long has it been since I heard anything from Bill?
And if I haven't heard anything soon enough, I'll
start sending duplicates to him until I hear from him that
he's gotten all of them.
That's basically how the TCP protocol works on top of the
internet protocol.
And now you know how the internet works.
I've left out from details like how does the routing
work, and what are domain names, and what's
the World Wide Web.
But the basics are there.
The second thing that goes on in this network, they have--
sorry, I left out the important story here.
So what is this all about?
This is the famous, anonymous SRI International
packet radio van.
When the ARPANET was successful, DARPA decided to
try out this packet switching idea in other modalities,
radio in particular, ground mobile radio and satellite.
So the ground mobile system, the packet radio system, was
actually built here in the San Francisco Bay Area.
Don Nielson was the head of that project and oversaw the
construction of this test experimental network.
And among the things he needed to do was to drive up and down
the Bayshore freeway and measure how well the packet
switching system was working in the presence of radio
noise, deep fades from moving into certain parts where there
are radio shadow and things of that sort.
So the story goes that the engineers drove this van up
and down the Bayshore and then pulled off to the side in
order to do some measurements.
And of course, the driver is one of the engineers, so he
got out and joined the rest of his colleagues in the back.
And a police car came along and noticed there was no one
in the cab of this rather nondescript-looking vehicle.
So he went around and knocked on the back.
And they opened the door.
And he looks inside, and he sees these really
geeky-looking, bearded folks with a lot of electronics, and
cathode ray tubes, and radio gear, and everything else.
And he said, who are you?
And they said, according to the story, we work for the
And he says, which government?

Officer, we were only going 50 kilobits per second.
So the radio van, this wonderful vehicle, was used
for religious conversion also.
We would take four-star generals on rides around the
Bay Area showing them how the packet radio system would
actually survive all kinds of problems, radio interference
and the like.
And they would go in in disbelief, and they would come
out believing that this technology actually held some
promise for the Defense Department.
So with a tip of the hat to Don Nielson and his folks.
This van, by the way, is at the Computer History Museum
now in Mountain View.
Then the other piece was the packet satellite system.
And this one was interesting because it was looking at the
problem of communication among ships at sea or from ship to
shore, distributing data around in a global network.
In this case, what was interesting is that there were
local ground stations, and they were all using the same
satellite and the same satellite channel.
So they were actually competing with each other for
access to common resources in the satellite.
And technologies were developed in order to allow
that to work.
So the three of those networks were the basis on which Bob
Kahn and I designed the internet.
When he came to Stanford in the springtime of 1973 and
said, look, we've got this ARPANET, and it works well.
We have the packet radio network, which is lossy,
things can get out of order.
We have this packet satellite system, which operates in
packet switch mode, but it's at different speeds, and the
packets sizes are different.
So we have these different networks, all of them behaving
like packet switch systems.
The question is how do we interconnect them together in
order to make it look as if it's one big, uniform network.
That was the internet problem.
So over a period of about six months, Bob and I designed the
basic network architecture and published a paper in 1974
called "The Protocol for Packet Network
A copy of the IEEE magazine that has that paper in it is
apparently for sale in some antiquarian bookstore in New
York for $7,500.
So of course, I immediately looked in my files to see if I
had any other copies of it.
New retirement plan.
Well, I wanted to mention one other important date.
And that was November 22, 1977.
For the first time, we have all three of the networks
interconnected and operating at the same time.
And it was important milestone from my point of view.
I was in Washington at ARPA.
And as program manager for this thing, it was the first
time we could show that all three of these networks would
actually interwork successfully and essentially
But I have to tell you, it was a pretty amazing
Bayshore Freeway and its packet radio van are going up
and down transmitting data through the packet radio
network to a gateway to the wireline ARPANET.
The packets went through the ARPANET, through an internal
satellite inside of the ARPANET, which just looked
like it was another link in that network, all the way to
London, where it emerged from the extension of the ARPANET,
through another gateway into the packet satellite system,
back down to Etam, West Virginia, to another satellite
ground station, all the way across the internet to a
computer at Marina del Rey in California, just
north of the airport.
Now as the crow flies, that's about 440 miles from San
Francisco to Los Angeles.
But the packet actually went 88,000 miles when you go all
the way up to satellite level, go down and up, and down, and
back across the Atlantic and the continent.
So these packets actually made it all the way through to
Marina del Rey and then all the way back, of course, to
the packet radio van.
So this was a big milestone for me.
And I remember making this picture and hanging it up in
my office as a milestone for a--
let's see, George Heilmeier was running the Defense
Department's ARPA agency at the time.
And I remember showing this to him back then, proud of the
fact that we actually made it work.
Now if you fast forward to 1999, this is what the
internet looked like.
And about all you can tell is that it got bigger, and it got
more complex, and it got more colorful.
And I think that's as good a definition of the internet as
you could come up with right now.
It's gotten much bigger than it was when it started.
In fact, it's very interesting to see where the users are on
this system.
If you look at the distribution, you notice that
Asia is now the single largest region of use of the internet.
Almost 400 million people in India, China, Indonesia,
Japan, Malaysia, and so on.
That number is going to get bigger partly just because of
56% of the world's population is in Asia.
So a lot more people will be using the internet from that
part of the world, that implies different languages on
the network than the ones that are commonly available now.
You will see different scripts, things that are
written in Latin, but they're in Hindi, and Arabic, or Urdu,
or Kanji, or Hangul, and so on.
All the languages of Asia will show up.
Europe is actually the next largest group of population
with 314 million users.
Of course, Europe is funny because it keeps redefining
itself by adding new countries.
So it's a little harder to figure out what's going to
happen in Europe in the future.
But North America is now only third.
Whereas 10 years ago, if I had been here speaking to you, we
would have been in the top and largest group using the
network 10 years ago.
So these statistics tell us that the internet is now
vastly more diverse than it was certainly from the
beginning and certainly even in the last 10 years.
Now I had a t-shirt made that says, "I P on everything." I
don't know whether you can see that now.
This is actually a very important principle.
When Bob Kahn and I were doing the design, we knew in 1973
that we did not know what new transmission and switching
technologies would be invented.
But we wanted the internet to be able to use them.
So we wanted to future proof the design of the system.
So we said, OK, it's important that these little electronic
postcards, the internet packets, be able to be carried
on anything, an optical fiber, a radio channel, a twisted
pair copper, anything that could carry a bag of bits from
point A to point B with some probability greater than 0, is
all that we're asking of the underlying system.
And all the other hard work, to keep things in order, to
retransmit, is what happens at the edges of the net.
So I had this t-shirt made that says, "I P on everything"
because that's basically what we were trying to do.
And for the last 35 years or so, it's been very successful.
Every new transmission and switching technology that's
come along has been swept into the internet as another
underlying transmission or switching system.
But there is another interesting implication here.
This means that the packets don't
care how they're carried.
They don't care if it's over a satellite link
or an optical fiber.
But the packets don't know what they're carrying.
All these little packets know is that they have some bits in
them, but they don't know whether it's video, or audio,
or a piece of email, or a piece of a web page.
All these packets know is that they're supposed to take the
bits from here to there and dump them off at
the edge of the net.
This led to the notion of an end-to-end principle which
basically says, only the computers at the edges of the
internet actually know what data is
flowing through the system.
They're the only ones who know what
applications are being supported.
So the net itself is completely application
independent and ignorant.
Some people call this the stupid network.
And the whole idea was that it didn't have to know what kinds
of applications were being invented.
The consequence of that is a cornucopia of innovation.
Because we didn't have to get permission from some internet
service provider to put up a new application, you didn't
have to change the network to put up a new application, it
was possible for people to try new ideas out without any
permission at all.
And so all of the applications on the internet that have
evolved over the last 10 or 15 years have simply happened
because people could try things out.
Google didn't have to get permission from anyone to try
out its applications, nor did Amazon, or Yahoo, or Ebay, or
Skype, or any of the other popular applications and
services on the net.
They just did whatever they wanted to do.
And if people were excited about it and interested, they
had access to it because the network was transparent.
Well, it's pretty clear that radio has become a very
important part of network because
it allows for mobility.
Certainly, in this building, it wouldn't surprise me if
there is wireless internet service available and you can
move from room to room--
VINT CERF: You don't have-- no wireless network?

This must be repaired.
VINT CERF: This is not acceptable.
It doesn't cost very much to put in a wireless capability.

Wherever the facilities committee is, perhaps
[? you should be ?] meeting after this one.
If you're looking for high speed,
optical fiber is terrific.
Digital subscriber [? loops ?] are terrific.
There's only one problem with the current broadband
offerings, and that is that they are asymmetric.
What I mean by this is that you can pull data in from the
net faster than you can push it out.
Because the way the high speed service works, it gives you
higher capacity from the network and less
capacity into it.
Now that would be OK if most of what you did was download
stuff off the net.
And for a long time, that's exactly what was happening.
But now, as people have the ability to produce their own
video, produce their own audio, produce their own web
pages and blogs, people want to push information into the
network just as much as they want to pull it out.
And so I think that this asymmetric kind of service is
going to be unsatisfactory for many of us.
And we'll be looking for high speed, symmetric capability to
and from the network.
One small, little packet of detail, and this is actually
important in historical terms, the internet that you're using
today was based on the design that was standardized in 1977.
It's called IP version 4.
We're going to run out of unique internet addresses,
things that tell you where there are unique terminations
in the net.
There were only 4.3 billion terminations made available in
this IP version 4.
Now, 1977 represented four years into the experiment.
And I thought at the time that the Defense Department
probably would not need more than 4.3 billion terminations
in order to demonstrate that packet switching and internet
made sense.
So I thought what would happen is we
demonstrate that it worked.
And then if it worked, we would invent
the production version.
Well, that didn't happen.
The internet just kept growing, and eventually, it
became commercially available in the late
1980s and early 1990s.
So the side effect of that is that we are going to run out
of the IP version 4 address space.
It's kind of like running out of [UNINTELLIGIBLE].
So there is a new version called IP version 6.
And if you're keeping track and wondering what happened to
IP version 5, that was an experiment in streaming audio
and video that didn't work out.
So we just abandoned that and went on to the
next version, 6.
Version 6 has 128 bits of address space.
And if you do the arithmetic, that means 340 trillion
trillion trillion addresses.
Now I used to go around saying, that's enough address
space so every electron can have it's own web
page if it wants to.
Until I got an email from some guy at Cal Tech, Dear Dr.
Cerf, you jerk, there's 10 to the 88 electrons in the
universe, and you're off by 50 orders of magnitude.
So I don't say that anymore.
But it's enough address space to last until after I'm dead.
And then it's somebody else's problem.
So that's where we're headed, IP version 6 starting in 2008.
There are some interesting social and economic effects of
this internet propagation.
One of them is that the consumers of information are
now becoming the producers.
It used to be that in mass media, there were a large
number of receivers and a small number of transmitters.
Whether it was book publication, newspapers,
magazines, radio, television, all of the mass media tended
to be a small number of producers,
large numbers of consumers.
Internet inverts that.
Now we have people blogging.
We have people uploading theirs videos to YouTube, and
Google Video, and elsewhere.
We have people creating their own web pages.
So we have a very democratizing environment for
the production of and sharing of information unlike anything
we ever had before.
We have the ability for groups of people to discover and
interact with each other.
So the side effect of that is quite challenging for
businesses because the business models of mass media
that were based on the idea that this small number of
producers set the pace and determined the content is
turned on its head.
And so for a number of people, that became a major change.
In fact, let me just pick one other thing on
this slide to emphasize.
A few months ago, I went to a computer store, and I bought 1
terrabyte, 1 trillion bytes, of disk memory to
take home to use.
And I paid about $1,000 for that.
Now already, I had several people tell me that in the
past that I had overpaid, that they could get a terrabyte of
disk memory for $300.
So obviously, I went to the wrong store.
But I remembered buying some disk memory in 1979.
I bought 10 million bytes of disk memory in a thing the
size of a shoebox, and it cost me $1,000 in 1979.
So I got curious, and I said, well, what would have happened
in 1979 if I tried to buy a trillion bytes of memory at
1979 prices?
And if you do the calculation, it would have cost $100
million in 1979.
Now I have to admit to you, I didn't have
$100 million in 1979.
I still don't have $100 million.
But I can guarantee you that if I'd had $100 million in
1979, my wife would not let me spend it on disk memory.
She would have had a better idea.
So what's happened is that the cost of digital things has
just dropped dramatically.
The cost of storage has dropped.
The cost of processing has dropped.
The cost of transmission has dropped.
And that is having a dramatic effect on some people's
business models.
For example, suppose that you thought you'd like to build a
store that sold CDs and DVDs.
And you wanted to have a big inventory, lots of choice.
How about a million DVDs and CDs?
Well, first, you'd have to have a store that's big enough
to hold all that physical inventory.
Then you'd realize that you're only going to get a market
that's a function of distance away from that store.
If you wanted to build a market, you have
to build more stores.
Transmitting this data online is less expensive, it's
faster, and, in fact, when you need to replicate it, you can
do it on the spot.
So if you have a very popular movie and you needed multiple
copies of that movie, you can do it
instantly at the theatre.
You don't have to wait for somebody to print another 35
millimeter canister and send it to you physically.
So these economics are really changing the way people work
on the net.
I mentioned mobiles before.
I don't want to overstate this one topic.
But I did want to mention that they are no longer just
I'm sure many of you have these.
This is a BlackBerry.
It does have some challenges.
You notice the size of the screen?
It's about the size of a 1928 television set.
And it has variable data rates depending on where you happen
to be, inside of a building, or whatever the local
facilities are.
And the keyboard is suitable for people that are three
inches tall.
So it's a challenge to offer products and services through
these devices.
But it's programmable, and that's what's important.
All you need to do is to download some new software in
order to make this do some new application.
Moreover, some people are starting to use these as
devices for making payments.
In places in the world where the mobile is the only
communications device available, some people are
using minutes of the mobile system as a means of currency,
as a way of paying for things.
You can transfer minutes from one account to another, and
those minutes are worth money.
And so the mobile industry is actually becoming an
infrastructure for payments, by creating a kind of a
So this has turned out to be a pretty exciting outcome.
Because it means that people who don't have bank accounts,
don't have checking accounts, don't have credit cards, but
they have mobiles, suddenly have an instrument that allows
them to grow the local economy in ways that
they couldn't before.
So we're very excited at Google about the potential of
mobile devices to bring new information and new
opportunities to people around the world.
The other thing that's interesting about these mobile
devices is that they are
essentially information portals.
So what we noticed is that when you carry your
information portal around with you, that you often have need
of information about what's going on around you.
Where is the nearest Thai restaurant?
How do I find the nearest ATM machine?
How do I get to a particular building on
the Stanford campus?
Those kinds of questions can be
answered with these devices.
Especially, you can have databases that are, what we'll
call, geographically indexed, that is to say, the
information in the database is indexed by where that
information is located.
So if I have a latitude and longitude or the street
address of something, I can create this
geographically-indexed database, make it accessible
through the mobile, and, therefore, help people with
information that is important to them based on where they
are right now.
And so the value of geographically-indexed
information has skyrocketed as a consequence of that.
Well, I have to tell you that I have been astonished at the
kinds of devices that have been
connected to the internet.
I certainly never anticipated that the refrigerator would
become an internet appliance.
Or for that matter, a picture frame.
I remember the first time somebody ran into my office
and said, Vint, Vint, did you see the internet-enabled
picture frame?
And I remember thinking, boy, that sounds about as useful as
an electric fork.
In fact, it's a very clever device.
There are a number of them made now.
And they can plug into the wall to get power.
And you plug into the telephone system or into an
ethernet connection, and this little picture frame wakes up
every 24 hours, logs on to the internet, goes off to a
website, and says, what do you want me to do now?
And it will say, well, download these pictures, and
display them, and erase these pictures, put up this new
information, and then just cycle through.
It's really nice for the grandparents.
They don't have to boot up Windows and figure out how to
log in to the internet.
They just turn the little picture frame on,
and it runs by itself.
So we have half a dozen of these around the US.
And all of us have digital cameras.
So we upload the pictures of the grandchildren and the
nieces and nephews.
And you get up in the morning and now look at this little
picture, and you can see what's
going on in your family.
Now those of you who have any interest at all in security
will appreciate that if this little picture frame is
downloading images from a website, and if the website
gets hacked, the grandparents may see pictures on the
picture frame that they hope are not of grandchildren.
So suddenly, security becomes important not only at work,
but at home as well.
Well, the guy that takes the cakes is one in the middle.
This guy invented an internet-enabled surfboard.
And he's out in San Diego somewhere.
And I guess he was sitting on the water, waiting for the
next wave, thinking, gee, if I had a laptop in my surfboard,
I could be surfing the internet while I'm waiting to
surf the Pacific.
So he built a laptop into his surfboard, and now he sells
that as a product with the wireless internet access on
the rescue shack back on the shore.
So I actually believe that there are going to be billions
of devices on the internet.
I sound like Carl Sagan.
Billions and billions of devices on the network.
Many of them, appliances that we don't normally think of as
being programmable at all, let alone networked.
So the consequence of that is really pretty interesting.
Some of these devices are already there.
When you go to a hotel, you'll find a web TV with a little
wireless keyboard.
Sometimes, well, the programmable digital
assistants like these are now part of the net.
The surfboard.
There are washing machines, picture frames.
I am fascinated by the idea of an internet-enabled
Now some of you probably use these toll road devices that
you stick on the windshield of the car.
They use what are called radio frequency identification
devices or RFIDs.
You hit them with a pulse of radio energy, and
they radiate back.
The account number that's supposed to be debited as you
go through the toll gate.
But you can imagine having products that you put into the
refrigerator with a little RFID chip in there.
So now the refrigerator, if it has a detector, could figure
out what it has inside.
So you can imagine coming home and seeing your
internet-enabled refrigerator with a nice liquid crystal,
touch-sensitive panel surfing the network, looking for
recipes that it can make with what it knows it has inside
the refrigerator.
When you come home, you see it [UNINTELLIGIBLE] like this.
And you can extrapolate this.
Imagine going on vacation.
And you get an email.
And it's from your refrigerator.
It says, I don't know how much milk is left, but you put it
in there three weeks ago, and it's going to walk out on its
own if you don't [INAUDIBLE].
Or maybe you're at the grocery store, and
your mobile goes off.
It's an SMS message from your refrigerator.
Don't forget the marinara sauce.
I have everything else I need for spaghetti dinner tonight.
Now I'm sorry to tell you that the Japanese have spoiled the
whole thing.
They've invented an
internet-enabled bathroom scale.
Step on the scale, and it figures out which family
member you are based on your weight.
And it sends that information to the doctor and becomes part
of your medical record.
Well, that's probably OK, right?
Except for one problem.
The refrigerator is on the same network.

So when you come home, you see diet recipes.
Or maybe it just refuses to open.

We don't have time to go through all the examples of
what happens when internet enables everything.
But because Burton Richter is here tonight--
I'm honored by his presence--
I have to tell you about my idea for dealing with the
Nobel Prize for people who are in the computer science world.
As you may know, there is no Nobel Prize for mathematics.
And some of you may think, well, what about economics,
and the Nash theorem, and so on?
Yes, but that's not Mr. Nobel's money.
That was money from another source even though it's
awarded by the committee.

And the reason, I am told, that there is no no Nobel
Prize for mathematics or any branch of mathematics, which
unfortunately has to include computer science which rules
me and my colleagues out, is, the story goes, that Mr.
Nobel's wife ran away with a very good mathematician.
And he forbade the committee to spend any money on
Whether that's true or not, I don't know, but it makes a
great story.
So here I am thinking, well, what can we do about the poor
people in the math and computer science world.
And I thought, OK, how about this?
You remember that one of the peculiar things about quantum
physics is that these particles can sometimes be in
more than one state at the same time.
And my recollection is that Erwin Schroedinger was trying
to explain this odd phenomenon by describing a thought
He said, math is kind of a box, and you put a cat in the
box, and you put in a capsule of cyanide inside of which is
some radioactive substance.
And if an alpha particle is emitted, it will break the
cyanide capsule, and the cat will die.
Now no cats were harmed by the way.
This is a thought experiment.
We haven't hurt anything.
So if you seal this whole system up and you ask, well,
what's the state of the cat, the answer is you don't know.
It could be either alive or dead.
And you have to treat the closed box as a system that is
in both stages at the same time until you open up the box
to find out the answer.
Well, I believe that a bottle of wine must be a giant
quantum particle.
Because if you think about it, until you open it up, it's
either absolutely horrible, or absolutely wonderful, or
something in between.
And you don't know until you pull the cork.
So I'm going to submit my theory of quantum wine bottles
to the committee, and I hope that they'll recognize
millions of [UNINTELLIGIBLE].
Now in the meantime, what we can do is put a memory stick
inside the cork, one of these little random
access memory sticks.
And then we could record when the wine was
put into the bottle.
And maybe we can put in the temperature and the humidity
at daily intervals.
So if you're thinking about opening a bottle of wine, you
can interrogate the cork and find out, on July 4, 1997, the
temperature rose to 120 degrees Fahrenheit because the
air conditioning failed.
That's the bottle that you give to a friend.

So I don't have time to do the rest of these.
I'm going to skip over some stuff here because I realize
that I've potentially running over time.
Let me see.
So let me finish up with one last report for you.
It's important that you understand that what I am
about to explain is not a Google project.
So I don't want you leaving the room, calling your broker,
saying, Google's business model is to take over the
solar system.
That's not what we're doing.
This is a project that I started in 1998 with the Jet
Propulsion Laboratory.
It has to do with the essential expansion of the
operating of the internet across the solar system.
Now we instrumented planet Earth pretty thoroughly.
We have orbiting satellites.
We have sensors on the ground, on the water, under the
ground, under the water.
And more of them are needed in order to understand the
dynamics of climate change, for example.
We're also very interested in what's going on on Mars or
what has gone on on Mars in the past. And we're beginning
to instrument that planet.
Typically, the way this has worked is that we've been
using a 1964 design Deep Space Network with three big 70
meter dishes in Goldstone, California, in Madrid, Spain
and in Canberra, Australia.
So as the Earth is rotating, those big dishes are able to
see pretty far into the possible communications
equipment on board these sensory systems deep into the
solar system.
So typically, the Deep Space Net is forming a direct radio
link to these various space platforms. Some of them, of
course, make it all the way on to the ground in the form of
sensory systems. Or you'll remember the 1997 Pathfinder
that landed on Mars and sent back all the really great
pictures of Mars and other data.
And more recently, there were rovers on Mars, which are
bigger than the Pathfinders.
They landed in 2004 in January.
One on one end of the Gusev Plain and the other at
Meridiani Planum.
They've been sending data back now for three years.
The original project was only supposed to last 90 days.
And one of the reasons that it was thought that this system
would not last more than about 90 days is that the spacecraft
are using solar panels in order to recharge batteries,
so that they could last during the Martian night.
These are the things that look like the black wings.
What we thought was that over time, these solar panels would
become more and more dusty.
And the more dust there was, the less efficient the panels
would be at converting sunlight to electricity until
finally, one day, the batteries would not
sufficiently charged up, so that as the Martian sunlight
the devices wouldn't wake up.
Well, in fact, these things have stayed
cleaner than we expected.
Personally, I think there's somebody up
there dusting them.
But we haven't caught them on the cameras yet.
So a more likely explanation is that on Mars there are
little dust storms, like little dust devils.
And they're actually blowing the dust off
of the solar panels.
You can see evidence of this from the orbiting satellites,
for example, but then also from some of the imagery from
the ground.
So that's why these things have lasted for much longer
than we expected.
We have orbiting satellites around Mars.
And there are four of them, although one of them just died
last November, the Mars Global Surveyor, after 10 years of
operation finally failed.
But the other three are in operation.
And the most recent one, the Mars Reconnaissance Orbiter,
went into orbit just last year.
And it's been operating since then.
What's rather interesting is that the original design of
the system had data that was being transmitted from the
ground directly back to the Deep Space Network.
When they turned those radios on, it turned out after about
20 minutes, they overheated.
And they had to turn them off to avoid having them harm
And so the duty cycle for transmitting data was reduced
because of that.
And secondly, the data rate that was sustainable was
fairly low, 28,000 bits per second.
The engineers at JPL decided there might be an alternative
way of moving data back.
And they did program, Mars Odyssey, for example, one of
the other orbiters, so that the rover on the ground could
transmit data up to the orbiting satellite which would
hold on to it until it got around in its orbit, so it
could transmit the data back to Earth.
Now that's what's called store and forward, which, by the
way, is exactly how the internet works.
The internet is a store and forward system.
So this is one of the first times that in deep space
applications, we use that kind of a technique in order to
relay data back.
Before that, it had always been kind of a radio relay,
where you just change frequencies as
you go back to Earth.
Well, this is pretty exciting for those of us who want to
extend the operation of the internet
across the solar system.
Because now we have evidence of the utility of that style
of operation.
So what we end up with over the past almost 10 years now
is to develop a new set of protocols that will work
across the interplanetary distances.
We originally thought we could use the standard TCP/IP
protocols to do this.
That lasted about a week, and then we realized that there
were real problems.
The biggest problem is that the distances between the
planets is literally astronomical.
And it turns out that at the speed of light, it takes
something like four minutes for a radio signal to go from
Earth to Mars.
And of course, a similar time back.
And then at the farthest points where we're apart--
we're like 235 million miles apart where we're in our
respective orbits--
and that takes almost 20 minutes one way.
Now the internet protocols were not designed to deal with
40-minute round-trip times.
If you can imagine surfing the network and clicking the
mouse, and then waiting for 40 minutes for the
first bit to come back.
Now I know some of you probably have network
At least, that's not because of interplanetary distances.
So that was a problem.
Then the other problem we ran into is that there's this
thing called celestial motion.
You put something on the surface of Mars, and Mars
insists on rotating.
So after a while, you can't talk to it.
Well, that messed up the TCP protocols too.
So we ended up designing a whole new suite of protocols
that we called delaying and disruption-tolerant networking
protocols in order to deal with some of these problems
that show up in interplanetary communication.
We now got to the point where protocols are designed and
tested terrestrially.
And NASA is now considering making them part of the
standard architecture of the future.
In the meantime, we went back to the Defense Advanced
Research Projects Agency, and we said, we think you have a
problem terrestrially with delay and disruption in
military communications, in tactical communications, where
people may be jamming the network.
You may be hiding under a bridge in radio shadow and you
can't communicate.
So we started testing these delay and disruption-tolerant
protocols with the Marine Corps.
And the application was so successful in the test that
they ran off with our equipment to Iraq with it.
And I said, wait, wait, it's an experiment.
They said, no, it isn't.
And they took it with them.
So we now at least have evidence that these systems
that were designed originally for interplanetary
communication are actually useful in the terrestrial
environment as well.
So here's what we're hoping will happen.
We hope that, eventually, we'll send people out around
the solar system.
And we hope that they'll have comfortable places to stay
because it's a pretty hostile environment down there.
But in the long run, what we hope is that by standardizing
the communication protocols used in deep space, that every
time you launch a new mission somewhere in the solar system,
is there are any previous missions that are currently
operating and are using these standard interplanetary
protocols, that they will become assets to support the
new mission.
So over time, we hopefully will accumulate an
interplanetary backbone that will support exploration by
robotic means and by man means over time, all of them
supplying the same kind of standard opportunity for
communication that we get today on planet Earth.
Because of the standards of the internet, when you plug
your computer into the internet, you can talk with
those 400 million other machines that are on the
system because everything is standardized.
We want the same thing to be true for the
interplanetary space.
So that ends my formal contribution to this

I promised that I would be available for questions if you
have them, but I won't be offended by anybody who would
like to leave now.
But if you have questions, theoretically, we have a
microphone that we can run around with, but I don't know
if it works.
What's the status?
It's dead?
Why does this remind me of Star Trek, where Dr. McCoy
says, "It's dead, Jim." Well, I can still try
to answer to questions.
But I may have to hop down and run around in the crowd
because I'm hearing impaired too.
But if there are questions, why don't we try?
I'll just jump down here.
That's what evangelism is about.

Yes, sir.
AUDIENCE: You're in the heart of
entrepreneurialism in this Valley.
Back in 1969 or so, did you ever consider making this a
propriety system and charging $0.01 per--
VINT CERF: As Bill suggested.
AUDIENCE: Exactly.
VINT CERF: Actually, this is a really good question.
The question is did Bob Kahn and I think about making this
a proprietary system and actually charging
people to use it.
The answer is yes, we considered that.
And we decided deliberately not to do that.
And the reason was very straightforward.
We knew that we had a problem in the Defense Department
because every time the Defense Department would buy
computers-- this is back now in the 1960s--
if you needed to make a network for those computers,
they had their own proprietary protocol.
So HP had one set of protocols.
Digital had another.
IBM had another.
They could talk within the same kind of computer network,
but you couldn't get them to interwork very well.
And we thought it would be a bad thing if the Defense
Department wanted to rely heavily on networking to be
confined to only one particular brand.
So what we wanted was a protocol that would be
non-proprietary and, moreover, would be globally accepted as
a standard.
So we set out deliberately to design something that had no
constraints with regard to intellectual property.
We didn't want any barriers or excuses for someone to say, I
don't want to implement that because I have my own
proprietary protocol.
Why should I license yours?
I have my own, thank you very much.
We didn't want that.
We wanted a standard.
Now we had no guarantee, of course, that this would become
a standard.
But we knew if we made it proprietary, it would just
increase the difficulty.
So we deliberately decided not to do that.
And I'm pretty sure that was the right decision.
I really doubt seriously [INAUDIBLE].

So I don't think we would be where we are now without that
Yes, sir.
Let me get this one, and then I'll get yours.
AUDIENCE: I keep losing my socks.
And I was wondering [INAUDIBLE].

VINT CERF: Where is the missing sock?
Well, remember I mentioned the RFID chips that the
refrigerator could use to figure out what it had inside?
Well, what if your socks were internet enabled?
You can imagine.
You interrogate the sock drawer, and it sends back an
answer saying, well, there are 17 matched pairs of socks, but
sock number 144-L is missing.
So you send a multitask radio message around the house.
And you get back the response, this is sock 144-L, I'm under
the sofa in the living room.
You've solved the problem of the missing sock.
What a huge contribution to society.
That's the story.
Yes, sir.
You had a question.
AUDIENCE: Would you comment on internet neutrality?
Actually, the phrase that's used is network neutrality.
But you're quite right, it's in the context of internet.
Many of us have seen this internet be such a big engine
for innovation and have asked ourselves, what is it that
permitted that degree of innovation and invention.
I alluded earlier to the idea of innovation without
The idea that once you pay to be on the internet, at
whatever capacity you are willing to pay for and can
afford, then you're free to use the net any way you wish.
And of course, illegal activities
would be still illegal.
But the idea is that you should not be constrained by
the supplier who pays the internet service as to what
applications you offer or what applications you can reach.
As the broadband system comes up, there's been a significant
change in the atmosphere surrounding the internet.
When the internet was a dial-up system, there was no
problem changing your internet service provider.
You just dial a different number.
And so there were literally thousands of internet service
providers around the United States.
And you could switch easily from one to the other by just
changing the number you dial.
When broadband came along, the story was quite different.
Statistically, not everyone in the United States has access
to broadband.
10% of the population doesn't have any access at all.
And the data from the SEC, which are a little biased not
by any malice intended, but simply because of the way the
data is collected, these data suggest that, at most, 60% of
the population has a choice of either DSL or cable, which are
the two primary broadband services.
And another 30% have a choice of either one or the other,
but not both.
And the actual data suggests that there is even less
competition than that.
So the problem here is that we don't have the kind of
competition we had in the dial-up network.
As a consequence to the absence of having competition,
there is a very strong potential for the controller
of the underlying broadband access to make use of that
control by interfering with competing higher-level
So to give you an example, suppose that you are the
telephone company.
You put in digital subscriber loops required.
And you sign that you're going to sell internet
service this way.
And you're also going to offer streaming video as a service.
And then you notice that there may be competitors out there,
somewhere else in the internet, not necessarily just
in the US, but anywhere else, also trying to reach your
subscriber with a streaming video.
And you decide you don't want the competition.
So you make use of the fact that you have control over the
underlying transport medium to interfere with your
Or alternatively, you might say to the competitor, I'm
sorry, I won't be able to let you have access to my
subscriber unless you pay me a toll.
Now up until this time, it has never been the case that
there's been such a double payment required.
Typically, what has happened is this buyer of internet
service provides two-way passes to the internet to a
subscriber who pays.
And then the rest of the internet is connected, and you
use it however you want.
But in the past year or so, the first inklings of a
different business view emerged from the telco
companies and less obviously from the cable companies.
So the debate right now, which was hot last year and is
somewhat more tepid right now, was whether there should be
enacted in law some [UNINTELLIGIBLE] that says,
you must treat all of these customers equally and not take
advantage of the fact that you happen to be the supplier of
the underlying transmission capacity.
There used to be a time when there was something called
common carriage.
And there were common carriage obligations.
There were rules called Computer I, II, and III.
Those rules established basic and enhanced service in the
telecom environment.
The FCC did away with all of those distinctions a
couple years ago.
And the consequence of that is that internet is being treated
as an information service without any common carriage
requirement at all, leading some of us to believe that the
market power of those small number of providers of
broadband could be abused.
And we thought that we would codify a rule that says, you
shouldn't do that.
And if you do that and we catch you doing it, there will
be consequences.
The debate, of course, on the other side is why would we
build this new broadband facility if we couldn't build
business models that are favorable to us.
And my reaction right now is that if you would produce this
kind of inequity, you will suppress some of the
innovations on the network, which have been a consequence
of freedom to try anything out that you want to.
So we at Google are very strongly in favor of this
rather neutral environment, where anyone has an
opportunity to try something else.
And we're very concerned that we not introduce practices
that would inhibit that kind of innovation.
So that's my long sermon on the subject of net neutrality.
Let me get Bill, and then I'll get this gentleman over here.
Yes, sir.
BILL ANDERSON: If anything, can you tell something about
Google's big library project?
What state is it in?
Where is it going?
VINT CERF: Bill's question is about the library project.
We now have something like a dozen libraries at different
universities cooperating with us to allow us to scan the
books that are in their libraries.
In fact, there are more libraries offering us the
opportunity to do that than we can actually manage.
So we had to pace the growth of this program according to
the amount of capability we have to
scan all the materials.
We continue to have discussions with the
intellectual property community, book publishers in
particular, about their concerns, which are that in
the process of scanning these books that we might be putting
things up on the net that they
consider to be under copyright.
And so the debate continues there.
For certain volumes which were published before, something
like, 1894, there is no question that the copyright
has now expired.
For other materials, about 15% of published books are
definitely under copyright or still in print.
And we even know who holds the copyright.
But there is a large chunk of material which is possibly
still under copyright, but is not in print.
And it's been almost impossible for us to identify
the copyright holder.
In fact, there are people who make a business out of helping
me find the copyright holder for a particular
book or other work.
So we continue to take the view that we should be able to
scan all this material.
If there is uncertainty as to a copyright situation, we
wouldn't actually put the content up for view.
We would only allow people to search it and then tell them
that this book exists and that the contents match the search
that you just did.
You might be interested in getting this book.
It might be available at the following libraries, or you
might want to go to Amazon, or you might go to Ebay or some
other source to actually get the book.
What we want to do is help people know that there's
information locked up in books that they might want to know
about in addition to whatever
information was already online.
So we continue to press along on that.
The other thing that we're very interested in is things
that have never been published, maps, manuscripts,
and things of that kind.
Those would be invaluably useful for research if they
could be available in an online form.
So we're pursuing that as well.
Yes, sir.
AUDIENCE: When you were setting up the ARPANET, you
were primarily involved with file transfer protocols.
Today, we spend most of our time looking at web pages.
From a technical point of view, what is the World Wide
Web, and how did it come about?
VINT CERF: In principle, what's happening is when you
connect your computer to a website, it's transmitting a
file full of information which you display on your screen.
The information that's sent, in the early stages of the
World Wide Web-- which, by the way, was invented by a man
named Tim Berners-Lee who was at CERN, the Center for
European Nuclear Research in the late 1980s looking for
ways to help physicists share their information, especially
information that included charts, graphs, and
photographs into different texts.
So he was trying to find a way to represent the information
and help people find it by electronically pointing to it,
by hyperlinking.
That was his original motivation.
That, of course, has blossomed into a remarkable avalanche of
information flowing into the internet.
So today, there are literally tens of billions of web pages
that are out there on the net.
And they're accessible through browsers, which run in your
laptop and desktop.
Google, among others, like Yahoo, and MSN, and so on,
make a business out of indexing all of that content.
And in fact, one reasonable mind picture of the World Wide
Web is that in spite of the hyperlinking that points from
one website to another, without this indexing
capability, it would be as if you walk into the New York
Public Library after an earthquake, where all the
books fell on the floor, and there's no card catalog.
Or even if you have a card catalog, you couldn't find the
books because they'd fallen off the shelves.
The internet has a little bit of the character like that.
So the indexing is a huge help to find things in
this morass of data.
We are, I would say, only 1% into the real depth of
understanding how to manager and help people find relevant
It's all syntactic.
It's all stream engine.
We haven't even begun to really dig deep into
semantics, which is the thing that Tim Berners-Lee is now
working on.
He calls it the semantic web.
And he wants people to be able to search this mass of
information on the basis of concepts and conceptual ideas,
and not simply on the basis of this string of letters that
might match the string in some piece of text elsewhere.
So this is a marvelous time to be alive because there's all
these new problems that we have to solve.
And I'm so happy Google is next door to Stanford because
we got this flow of really smart students
coming into the company.
And I have to stop because I'm way over time.
BILL ANDERSON: Well, I think if we let the questions go on,
we might be here until 7:00.
And I think that [UNINTELLIGIBLE], and I think