Tip:
Highlight text to annotate it
X
GARY TAN: A very good afternoon to you, ladies and
gentlemen, and a very warm welcome to
NUS School of Computing.
I'm Gary, and I'll be chairing this afternoon's session.
As part of our school's 10th anniversary celebration,
together with Google, we are proud to bring to you our
special speaker, Dr. Vinton Gray Cerf, Vice President and
Chief Evangelist of Google.
Dr. Cerf holds a B.S. from Stanford and
a Ph.D. from UCLA.
And in the '70s, as you know, co-designed the TCP/IP
protocols with Robert Eliott Kahn.
Incidentally, as [? a ?] piece of trivia, that was roughly
the same time when our school was starting out as a
fledgling department in the [? Bukit Timah ?]
campus some 33 years ago.
For his work on the internet, he has won numerous awards,
including the ACM Turing Award and the US Presidential Medal
of Freedom, the highest civilian decoration bestowed
by the President of the United States.
Now, trying to tell you more about our speaker would be a
disservice to you, as I'm sure you're all dying
to hear from him.
Besides, our speaker really needs no introduction.
So, without further delay, I can call upon Dr. Vinton Cerf,
father of the internet, to share with us the tracking of
the internet into the 21st century.
Dr. Cerf, please.
VINTON CERF: Thank you.
Thank you.
Thank you.
Thank you very much.
I appreciate everyone taking an afternoon off so we can
have a chat on Friday afternoon.
At least it gives you an excuse to escape from classes
for a few minutes, anyway.
It's a real joy to be back on a university campus.
It's something I enjoy the most, because it's a chance to
interact with people who are thinking about hard problems
and looking for solutions to them.
And I'm going to bring up some hard problems in the course of
this discussion.
And I hope I'm going to trigger some Ph.D.-level
research as a consequence of the chat we have today.
So let me start out by simply giving you a bit of
statistical information about the current
state of the internet.
I've been giving these talks for quite a while, and in the
recent past, let's say in the last couple of years, it's
been apparent that Asia is now the dominant user population
on the internet.
It's not a big surprise.
China and India are part of Asia, and they certainly have
very big populations.
They are increasingly part of the internet environment, and
you'll notice that the percentage penetration, on the
average, is still fairly modest.
The average around the world is 20% penetrated.
The average in Asia is a little bit below that.
But just imagine, as the penetration of internet in
Asia reaches the same levels as it is in other
parts of the world--
40, 50, 60, 70%--
there's no question that our Asian colleagues will be the
dominant participants in the internet.
They will be the dominant source of content.
Their languages, your cultures, your styles of
interaction will be, in fact, the most significant part of
the internet.
Now, the next largest component on the internet is
Europe, almost 350 million users.
I've given up trying to make any guesses about Europe
because they keep adding countries.
So the definition of Europe is now--
I don't know anymore what's going to be Europe in 10
years' time.
North America, the US and Canada, are now number three
on the list, whereas, once upon a time, they were the
largest single segment on the network.
And the rest are as you see them.
Probably the biggest challenge in the internet universe is
getting Africa up and online.
Every country in Africa has some internet capability.
But, as you look across the continent, there are a billion
people, and only about 40 million of them are online.
The telecom infrastructure in Africa--
in African countries--
varies pretty dramatically, on the average [? still ?]
immature.
And the economies of Africa vary widely.
And so the affordability of internet service also varies
substantially.
So this is still an area of real challenge.
I'm going to show you some projections of global
population and regional population.
The projections come from this UN report.
I'm sure you don't need to write all this stuff down.
I'll be sure that these slides are available.
And I'll ask the university to make sure that they are known
to the students and staff, to get access to this.
So this is where my data is coming from.
If you have disagreement with the methodology for making the
projections, you need to complain to the United Nations
and not to me.
The projections go from 1950--
looking back historically--
to 2300, so it's a pretty dramatic estimate.
This is an eye chart for people in
the back of the room.
But what you can see is that the estimates of global
population--
high, medium, and low estimates--
are between 10.6 at the high end, 7.4 at the low end, with
a kind of a scary image showing that it might drop
off, and 8.9, which is a slow increase up to the year 2300.
I've chosen for my calculations of the year 2035,
to stick with the kind of middle chart, just for
simplicity.
This picture is rather interesting.
The parts at the bottom here represent the population of
what is currently called "the developed world." You'd be
included in that.
And it's roughly oscillating around the 1.2 billion range.
The other populations vary pretty dramatically
up to the year 2050.
You see a massive increase in the number of people living in
what are currently called "less developed" countries,
and so that will include some parts of Asia and
Africa, and the like.
Once the year 2050 arrives, the 7.7 billion people living
in those regions varies not very much.
But, in fact, you can even see it kind of dropping off.
Now this is on a global scale.
Let's look and see what the regional picture is like.
Here's the European picture, and you see a very dramatic
change in the populations.
From 1950 to 2000, the population goes all the way up
to nearly 800 million.
And then, by 2050, there's a massive drop.
And the most significant drop is in the lower bar here,
which represents Eastern Europe.
Some of you will know that Eastern Europe is already
experiencing declining population.
Their fertility rate is below replacement rate, which is
pretty amazing.
That has all kinds of economic impact on the countries in
Eastern Europe.
Because, if they have declining populations, they
have a declining workforce, which means that their
economies are going to be under stress, unless they
bring in working populations from other parts of the world.
Now why am I showing you all this stuff when we're supposed
to be talking about the internet?
And the reason I wanted to show this is that, if you're
trying to make projections about what the internet is
going to be like in 20 or 30 years time, you have to know
something about who's using it.
Because it's the people who use it who will produce the
information and the contents.
The people who use it will determine what applications
and services are of interest to them.
And the people who are using it may also be the inventors
of new applications on the service.
So understanding which populations are part of the
internet is important if you're trying to make any
projections into the future.
If we look at Asia, we can see, again, a very rapid
growth between 2000 and 2050 and some declining showing up,
which by around 2150 stabilizes and begins to grow
a little bit more.
Once again, the interesting major variation in population
comes here, in this segment, and that's China.
And so we're eventually seeing the effects of the Chinese
attempt to manage population growth.
What's not being as heavily managed is India.
And the Indian population, in fact, goes up quite rapidly up
until 2050, and then you see some significant declines
between 2100 and 2150, for a variety of different reasons.
Now, the reason that these variations occur, the reason
that you're not just seeing continued increased growth, is
an interesting interplay between
fertility and longevity.
How long do people live?
What's the mortality rate?
And what is the birth rate?
And those three things interact with each other in
fairly complex ways, which is why you see charts like this,
that are not necessarily smooth.
So let's imagine that it's year 2035, and we ask
ourselves what's the world like?
Well, one almost certain issue is that the global warming
effect will have been ongoing all the way until that time
for another 27 years and, of course, continuing.
Oil will become less and less available.
Fresh water will become less and less available.
Perhaps more critically, the location of populations around
the world will have to change in some dramatic ways, because
much of the world's population lies along the coastlines
where the oceans are.
And, if you listen to or watched any of the Al Gore
projections, you know that a rising sea level will cause
populations along the coastal areas to have to move inland
because of the rising sea levels.
Rising sea levels have other serious implications.
For example, if you have sanitation plants and the like
that are trying to clean up the water, maybe even recycle
the water, they can be overwhelmed by
a rising sea level.
The available freshwater aquifers can be overwhelmed
and, in fact, polluted by salt entering into
the freshwater systems.
That can have a big impact on fresh water for drinking, for
cooking, and for growing food.
And so there could be many very, very significant side
effects of this global warming and rising sea level.
The other thing which is obviously going to happen--
and it's already happening--
is that fresh water will be less and less available.
If you're dependent on, for example, the snowpack in the
Himalayas, and the glaciers melt, then the fresh water is
not going to become available through the snow melting.
And as a result, you have large populations which don't
have enough fresh water.
Barcelona is a city in Spain that already has this problem.
It is now importing tankers full of fresh water for the
people living in Barcelona.
And in some of the central parts of Spain, they're seeing
desertification, more deserts being formed.
The Gobi Desert is moving at the rate of a couple of miles
a year, another major issue.
Desalinization may be a source of water.
So we can imagine in 2035 that there are major desalinization
plants taking water out the oceans and
getting rid of the salt.
I had a unique experience about three weeks ago.
I was in Tokyo, along with my colleague Bob Kahn, as the
recipient of the Japan Prize.
And one of the features of the Japan Prize ceremonies is that
you get to have dinner with the Emperor
and Empress of Japan.
And what I didn't know, until I got there, is that the
Emperor is, in fact, a scientist.
He's a taxonomist, and he studies particular species of
fish in the general area around Japan.
And we got to talking about desalinization, and he
mentioned that there is a problem with desalinization.
What do you do with all the salt that you
get out of the ocean?
And so one answer might be, well, you pile it up in a big
mountain somewhere.
Another answer might be that you turn it into building
material, which is apparently technically feasible.
And I said, well, what about just dumping it back into the
ocean again?
And he pointed out that increasing the salinity of the
ocean might actually have a pretty bad influence on the
species of fish that could survive in the ocean if it
gets overly salty.
However, there was another crazy idea, which didn't occur
to me til after this conversation.
Those of you who may know something about the way the
ocean currents work may know that the reason Europe has
moderately, let's say, clement winter weather is that there
is an ocean current which brings
warmth to Northern Europe.
If the glaciers melt in the northern part of the globe,
and put a lot of fresh water in, that current goes away.
And that will translate into very severe
winter months in Europe.
So one answer to the desalinization question is,
dump the salt in the North Atlantic in order to preserve
the salinity of the ocean and maintain the currents.
Now, whether that's actually technically feasible, I don't
have any idea.
But the reason I bring all this stuff up is that the
world that we are going to live in 2035 is going to be
dramatically different from the world we're living in
today for a whole lot of physical reasons.
What can we do, though, with IT--
information technology--
to try to deal with some of the problems?
Well, one example is that we can substitute IT for
transportation.
Instead of expanding crucial fossil oil resources, we can
use telecommunications systems with increasing fidelity to
have meetings that we would otherwise have to meet
face-to-face.
So, conceivably, I could have this meeting with you using a
virtual event projected on the wall and interchanges with
each other over the net.
And, in fact, increasingly that's a common tactic.
There are technologies that allow for very, very natural
interactions.
A company called Cisco, which I'm sure you all are familiar
with, makes a service--
or a system-- that they call the TelePresence System.
It's a wall-size display like this, where the projections of
the people are life-size.
You're sitting at a table.
And you look across the table at this high-resolution
screen, and you see people the same size that
they are in real life.
The sound system is designed to produce sound from the
image that you're seeing.
So, when someone speaks over here, you hear them coming
from over there.
When you look at each other eye-to-eye in the system, it
really looks like you're looking eye-to-eye, as opposed
to the one where the camera is at an angle, and there's a
person projected on the wall, and you're never really
looking at each other.
So it's a very different experience.
So I found it quite natural to use that system for
communication and for a small meeting of
six or seven people.
Now, the only problem is that system costs about $300,000.
So that's a little more than I planned
to spend in my basement.
On the other hand, technologies have a tendency
to get cheaper and cheaper as we build more and more of them
and learn more, so my projection is that by 2035 it
will be quite common to substitute this kind of
virtual interaction with physical presence.
The second thing which strikes me as possible is to use IT in
order to minimize the use of natural resources.
We're already starting to see that happening.
Companies that do overnight delivery, like FedEx and UPS
and so on, are using IT algorithms to change the
routes of the trucks that are either picking up or dropping
off packages in real time, based on the demands that
people are making, in order to minimize the fuel consumption.
Another thing which is happening, to go back to water
shortages--
I guess I mentioned that.
The water tankers that are coming into Barcelona are
already examples of that.
Energy production and consumption is a
big issue for IT.
We experience this at Google because we build large-scale
data centers, and they consume a substantial amount of power.
They also give off a lot of heat.
We may have to get rid of that heat, as well.
The one single thing that would make a big difference to
Google, in terms of its ability to do what it does,
would be for people to invent computers that run at very
high speed at 1/10 the power that they currently do.
So if there are physicists among you, or hardware
electronic engineers who are interested in doing something
really significant, go figure out how to build a computer
that doesn't use as much power as the ones today.
Now, there is a tendency already to limit the amount of
power consumed by computers.
Those of you who've been following Moore's Law will
notice that the normal projection shows increasing
clock speed for the computers.
That's stopped.
Moore's Law has now leveled out in a very interesting way.
The clock speed is relatively constant.
The thing that's still increasing is the total amount
of computer cycles available per chip.
But the way that's being accomplished is to put
multiple cores on the same chip.
Well, this has an interesting challenge for people who run
algorithms that were relying on having faster clock speeds
to make the serial algorithm run faster.
Now you can't rely on clock speed to do that.
You have to figure out whether you can change the algorithm
to run in parallel, or maybe create a pipeline for the
algorithm, so that all the delay goes up.
The total throughput increases, because you're
running this stuff through a pipeline process.
It's not always necessarily the case that the algorithms
that you were running in serial are easily
parallelized.
And so this is a challenge for us and others who are heavily
dependent on large-scale, high-speed computing.
We have to reinvent the algorithms to run in pipeline
or parallel mode.
And finally, energy production is going to be a
big issue in 2035.
We're going to be running out of the oil and coal resources.
Or even if we were not--
For example, we might still have plenty of coal available,
but every time we build another coal-fired power
plant, it generates a huge amount of carbon dioxide and
exacerbates the global warming.
So we're going to want to do something else.
And one possibility is nuclear.
There are countries like France that get almost 80% of
their power requirements from nuclear power.
But there's things like ethanol, wind, solar, and
geothermal.
All those are going to be more exploited by 2035.
So our world is going to look significantly different.
Oh, thank you.
Is this *** or just water?
I have often thought that it'd be more fun to have a lecture
where this is actually ***, and the lecture gets more and
more incoherent as time goes on.
So our world in 2035 is going to look pretty different from
the one we're in now.
Internet, by that time, will be 52 years old.
That's assuming you start counting from the year that it
was actually rolled out.
It was invented in 1973, but it actually got rolled out in
1983, in January, for the academic community in the
United States and for some of the military.
The world population, by that time, will be on
the order of 8 billion.
And assuming internet penetration at 70%, which is
about as high as it is now in North America and parts of
northern Europe, that's about 5.6 billion people online.
Some of the developed countries, including
Singapore, will be more than 100% penetrated by internet.
You might say, well, how can that possibly be?
And the arithmetic that gets done is like Luxembourg.
I was talking to Commissioner Viviane Reding, who handles
communications in the European Commission.
And she very proudly said the penetration of mobiles in
Luxembourg was 150%.
And I said, well, how can that be?
And she said every other Luxembourgian has two mobiles.
So that's why it's 150%.
We may very well have more than one device on the net,
per person in the internet, of the year 2035.
And in fact, I'm currently guessing--
I emphasize guessing--
there could be 60 billion devices on the internet, which
means, on the average, 10 devices on the net per person.
You might say, well, how can that be?
And the answer is a lot of those devices will be sensors.
Some of them will be things like controlling light
switches and other parts of building sensor systems,
heating and ventilation, and air conditioning, and so on,
control systems and the like.
Appliances around the house, appliances in the office,
appliances in the car, appliances that you carry
around on your body or in your purse.
So we could easily imagine having a very large number of
devices on the net, which by the way, motivates the utility
of IP version 6.
Because today's internet is based on a protocol that was
standardized in 1977, called IP version 4, which only has a
32-bit address space.
That means 4.3 billion unique terminations,
and that's not enough.
It was enough in 1977, because I thought it was an
experiment.
And so I picked 32 bits after a year of debate.
I picked 32 because nobody seemed to make up their mind,
and I said, look, it's four years into this experiment.
We don't know if it's going to work.
4.3 billion termination points ought to be enough for this
experiment.
The thing is, the experiment never ended.
And so here we are, in 2008, and we need
more address space.
IPv6 has 128 bits of address space.
That's 3.4 times 10 to the 38th.
By the way, it's a joy to be in a place where I can say
that, and people actually understand what it means.
The only other place where you can say 3.4 times 10 to the
38th is to talk to a legislature or a congress
that's accustomed to dealing with money, and they
understand big numbers.
Anyway, that's enough address space so that every electron
in the universe can have its own web page, except it's--
Actually, I used to say that.
And I don't anymore, because I got an email from
a guy in Cal Tech.
"Dear Dr. Cerf, you jerk.
There's 10 to the 88th electrons in the universe, and
you're off by 50 orders of magnitude." So I don't say
that anymore.
So anyway, there's plenty of address space if we can get
IPv6 deployed.
And by 2035, I'm pretty confident we will be running
IPv6 everywhere.
So the internet looks kind of like this.
In 2008, a half a billion devices on the net-- servers
on the net.
2035, 10 billion hosts on the net.
By the way, there isn't any Wikipedia yet for 2035.
I'm inventing one that has these statistics in it.
5.7 billion users, 6 billion broadband connections, homes
and residences combined.
10 billion mobiles and maybe 12 billion PCs.
Of course by 2035, it could be that the PC will be replaced
by something else.
And so this is just a wild guess.
The infrastructure of the internet-- the information
infrastructure in the internet--
in 2035, will be both wired and wireless.
They'll both be higher speed.
We'll be seeing speeds that exceed a billion bits per
second for most of the broadband wired users, and
they could be in the tens to hundreds to even gigabits per
second for the wireless versions, as well.
There will be lots and lots of interactions in the net that
go from business to business.
There'll be standard transactions that take place.
Sometimes people refer to that as Web 2.0.
I think that's a kind of marketing buzzword.
But there will be enough standards so that businesses
will be able to automatically interact with each other,
placing orders, generating inventory requests, demanding
payment, making accounts payable, and so on.
Creating bills of lading for shipment and the like, all
automatically, because of the standards that companies will
use to interact with each other.
I think cloud computing is going to be very common.
Today it's a new term, but it's
referring to an old concept.
Back in the 1960s, people saw computers as very, very big
expensive things.
And so they imagined that, in the long-term future, the way
people would use computers would be to have gigantic
factories full of computers that people would access over
the telecom network somehow remotely.
Well, then, of course, came minicomputers and
microcomputers, and laptops, and desktops, and personal
digital assistants.
And so computing power migrated out to the
edges of the net.
And people didn't think about having big computing utilities
anymore, until companies like Google came along and said the
only way we can do what we do is to build these gigantic
computer centers.
So, suddenly we've come full circle in 40 years' time to
having gigantic computer utilities which, in fact, run
applications on behalf of people who are remote.
There's a lot of logic to this kind of architecture.
Because when you think about the maintenance of software,
we learned that it's not easy to maintain the software in
individual and separate laptops and desktops.
If, in fact, most of them have standardized browser-style
interfaces, and they're interacting with software
that's maintained in central computing centers, that
software can actually be maintained more easily, and
more reliably, and more uniformly than trying to
maintain copies of software in every
single laptop and desktop.
So I think cloud computing is likely to be quite popular in
the year 2035, and perhaps even before.
The other thing that's important is that, if you need
more computing power than your laptop can supply you--
Even though laptops of 2035 will be enormously more
powerful than the ones we have today, the cloud computing
arrangement allows you to expand the computation into a
much larger amount of compute power, and then return that
compute power to the pool to be available for other people
to have this large computation.
Moreover, the databases that you're working with may be so
big that it would never make any sense to transport them
back to the laptop.
It makes more sense to transport the computation to
the data, and allow it to take place in the cloud, and then
get the answers back.
I'm expecting to see a lot more collaboration in virtual
environments.
Some of you may be members of Second Life.
And it's kind of a crude virtual environment today.
By the year 2035, those environments will probably be
holographic in their character.
You'll be able to interact with people very comfortably.
Be able to initiate video conferences and things of that
sort, sharing information in databases with each other in a
very natural way.
And even more interestingly, we can create virtual
laboratories online.
You walk into a virtual laboratory, and you start
interacting with a virtual telescope.
But, in fact, that virtual telescope is simply a stand-in
for a real telescope somewhere else on the internet.
And when you ask that telescope in the virtual space
to take an image, a real telescope gets pointed to
where you need to go, and you get the image
and bring it back.
So the idea of creating these virtual laboratories not only
has a possibility for educational experiences, but
it also makes it possible to control multiple instruments
that are attached to the internet.
So I expect that to be quite common.
I'm also believing right now that today's naming and
addressing environment will be dramatically
different in 2035.
Today we identify host computers and endpoints with
IP addresses.
And they are topologically significant.
So the IP address tells you something about where
something is in the net.
And if you attach to the internet someplace else, you
need a new IP address because it's telling where you are.
Then we have domain names.
And, in theory, the domain names will work no matter
where you are on the net.
You just have a different map from the domain name into the
IP address.
But what happens with that paradigm is that the basic
identification space is of a physical device.
This computer is hanging off the
internet at this IP address.
And the domain name is really just a stand-in for a computer
that has lots of processes running and has lots of files
contained inside of it.
If we had unique identifiers for the files, and we had
unique identifiers for the processes that are running in
the machines, and those unique identifiers were independent
of which machine the processes or files were on, then we
could freely move processes and objects
around in the internet.
And it wouldn't matter which machine they were on.
They have identities independent of their location.
I'm convinced that by 2035 our naming and addressing spaces
will be abstractions of these processes and objects.
And that they will be independent of where they
physically exist in the net, which means that we'll have
communications going back and forth between the processes
and the digital objects that are, in fact, mobile and can
be moved from one physical machine to another, which
increases their resiliency to various kinds of failure.
I think by the time 2035 rolls around, there will be
large-scale databases, some of them sponsored by state,
national, and maybe even international organizations.
I can give you an example of how powerful it can be to have
aggregated information available to us.
At Google, we speculated that it might be possible to detect
an epidemic, a health epidemic, if we knew something
about the kinds of questions that people are asking of the
internet contents, the web contents.
In the United States, there's an organization called the
Center for Disease Control.
And it is responsible for detecting and reacting to
things like an epidemic.
Now, the way that tends to work is that doctors will
report to the Center for Disease Control that a large
number of patients are coming in with particular symptoms or
have been diagnosed with a particular disease.
But it takes the Center for Disease Control some time
before they actually figure out that there's an epidemic,
because they have to wait until they get reports from
the doctor.
So we speculated that we might be able to tell the epidemic
was underway sooner by looking to see what kinds of health
questions people were asking on the internet.
So we tested our theory.
We looked at the dates that the Center for Disease Control
in the past had announced that a certain epidemic appeared to
be underway.
And then we started looking at our databases of the kinds of
queries and responses that had happened in the
time before that date.
And what we discovered is that, about 30 days before the
announcement of an epidemic, we could actually see people
asking questions about the symptoms for that particular
epidemic ahead of time.
And you almost can see why that would be the case.
You get sick.
And you have some symptom, like you're having trouble
breathing or something.
So, if you're on the internet, you go and look to see what
could cause me to have trouble breathing.
And you might discover a variety of
different possible causes.
But then you sort of hope it will go away.
OK.
So you wait for another week, and it doesn't go away.
Then you go to the doctor, and you say, doctor, I've had this
problem for a week.
And the doctor says, OK, try this medicine.
Here's your problem.
But the doctor doesn't know that there's an epidemic.
All he knows is that a few patients have shown up.
But after another week goes by, the doctor says, gosh, 100
people have come into my office this week with the same
set of symptoms.
I better report this to the Center for Disease Control.
So you can see how the reporting process is a lagging
indicator of this potential epidemic, whereas the querying
on the internet databases is, in fact, a leading indicator.
So we think, based on the evidence that we have so far,
that we should continually monitor the kind of health
queries that are going on in the net, in order to detect
the possibility of an epidemic somewhere in the world.
Well, by 2035--
I'm looking at this now from the US perspective--
we have an aging population.
We have an increasing ethnic diversity in the country, more
people of Hispanic origin, for example.
We see ourselves as in competition with India, China,
Europe, and other parts of the world.
Our GDP will probably be only 50% or less that of India by--
or I'm sorry, of Asia--
by the year 2035.
This is a big change for the United States because,
historically, it's always had the largest GDP of all the
countries in the world.
But by the time 2035 comes along, the GDPs of the rest of
the world, especially Asia, will have escalated
dramatically.
You have big populations already, and their earning
capacity will have increased substantially over the 27
years between now and then, and so our average GDP in the
US will be a lot less.
We can't produce scientists and engineers as quickly as
you can in Asia because you have a larger population from
which to draw scientists and engineers.
Now, I don't know whether these
numbers are exactly right.
I'm sure they're not.
The question is what are the real numbers?
And the US has to learn that it's not going to be
[INAUDIBLE]
on the basis of numbers anymore.
And so we have to learn how to live in an environment where
we're no longer the most significant
economy in the world.
Now, on the other hand, people in Asia are going to have to
discover how to manage themselves because you are the
largest economy in the world.
That's a change, too.
So all of us are going will face a
different kind of world.
The physical infrastructure of that time-frame will be very
much affected by the availability
of energy and water.
And we've already talked a lot about that.
Minimization of the use of water, minimization of the use
of energy, re-utilization of these resources, local
production of power, local production of
water, all very important.
So we talked about a lot of these things, and I'll just
leave them at that.
Let's turn now to the social and economic effects of the
internet in the--