Eric Schmidt at D5: All Things Digital


Uploaded by Google on 25.06.2007

Transcript:

WALT: --come the day before and participated in the
computer bowl that some you know we had, but has never
really sat down for an interview.
And we also invited Chad Hurley and Steve Chen, who you
already saw.
And then very late in the process, after the $1 billion
lawsuit was filed, Viacom suggested that maybe we would
want to have Philippe Deaumont, and we said, well
yeah, that's a good idea.
But the only slot we have is right between the YouTube guys
and the CEO of Google.
Do you really want that slot, because that's it.
And they said yes.
And so I'm now going to bring out the other bookend of the
Google assault on Viacom, Eric Schmidt.
Eric?
["I CAN SEE FOR MILES"]
ERIC SCHMIDT: Hi, Walt.
Thank you.
WALT: Thanks for coming.
ERIC SCHMIDT: I've never been introduced
as a bookend before.
WALT: Bookend.
Well, a flanker, a pincer, something.
I don't know what the right term would be.
And I suppose we should start off with the Viacom lawsuit.
I'm sure your lawyers have told you you're not supposed
talk much about it.
ERIC SCHMIDT: I can talk about it lots.
WALT: Good.
Great.
Well, talk about it lots.
I mean, doesn't he have a point?
He owns and has paid a lot of money for the production of a
bunch of this content.
We're apparently violating his copyright in the eyes of some
of his lawyers this morning, but he owns it, right?
So doesn't he have a right to decide who can distribute it
and get paid for distributing it and so forth?
ERIC SCHMIDT: Well, the Viacom lawsuit was
probably just a mistake.
WALT: Probably just a mistake.
ERIC SCHMIDT: Because the underlying dispute is about
the Digital Millennium Copyright Act, and whether
there's a shared responsibility
around takedown or not.
We read the law as pretty clear that there's a safe
haven for sites on the internet, and others, that
host content that is incorrectly or illegally
uploaded as long as they take it down promptly.
And indeed, we do that.
We're busy building tools that will automate as much of that
process as you possibly can, because it's crazy to have
people doing this stuff manually, and I think they
just made a mistake.
WALT: What should they have done?
ERIC SCHMIDT: They should have waited for the tools.
WALT: Well why should they wait?
Why should they believe that you're going to
come up with the tools?
ERIC SCHMIDT: Because we have lots of other
people using them.
WALT: Well, but why shouldn't he say, even one day of using
my material--
and we've talked to you before about this--
even one day is an outrage, because we paid for this.
I mean, do you believe in intellectual
property and copyright?
ERIC SCHMIDT: Absolutely.
And there are lawyers who have been to the school that says
that intellectual property is a full property right over
which you have absolute control.
There's another set of lawyers-- who apparently have
all been to the Supreme Court--
who said that it is, in fact, a shared responsibility.
And the important principle about copyright here is that
Google actually operates under the DMCA for lots of our
properties.
We happen to be sued by Viacom over this particular issue,
which is largely a business negotiation, I think.
And had they simply waited, the tools that are in
development now would have addressed most of their
complaints, I think.
WALT: But why should the onus be on them--
I know I'm rephrasing my question, but why should the
onus be on them to wait for you, a bunch of smart
engineers to come up with some tools?
Why shouldn't the onus be on you to say, until the tools
come out, we're not going to--
ERIC SCHMIDT: Because we follow the law.
The law does not require us to build tools.
We're building the tools to make their lives easier
because we think it's good for everybody.
The ultimate outcome of copyright is going to be a
complex mix of user authentication, producer
authentication, and user choice, and the more we can
automate those tools, the better.
From a legal perspective, which is what you're asking,
we met the terms of the law a year ago, we met it six months
ago, we meet it today.
WALT: OK.
Let me turn it around a little, because when Chad and
Steve were out here, there was question Gary Shapiro from the
Consumer Electronics Association asked, which
struck me as very interesting, which was you have this
community around you to give other communities and other
parts of your product portfolio.

And one approach to this is a technological approach, which
you're talking about with tools.
There are--
let's have a lawsuit with each individual instance, whether
it's Google or other companies around the web, and other old
media companies.
Let's have a separate business negotiation about everything.
But another approach is to say, we live in a country
where the copyright laws--
including the Digital Millennium Copyright Act--
which is supposed to be the modern one--
are entirely written from the point of view of
the copyright holders.
Yes, you have a safe harbor, because either you or--
probably not Google, actually, at the time-- but some
companies like you-- primarily, I think, AOL--
hired a bunch of lawyers and got a loophole.
You got a Get Out of Jail Free card somewhere in the bowels
of that law.
But these laws are written entirely from the point of
view of the copyright holders.
They all say nothing can be copied.
There's nothing in them that recognizes
the rights of consumers.
I'm not even talking about Google now, but consumers.
If you have a community at YouTube, why don't you
mobilize it to get political change, rather than just be
satisfied with either defeating
lawsuits or cutting deals?
ERIC SCHMIDT: There's a line of advocacy that we probably
should not cross as Google.
Google tries very hard to be an independent and
unaffiliated arbiter of what the best search results are,
and there's always this line that we have to be very
careful about.
My personal view may be different from the law, but
the fact of the matter is that in the United States we're
governed by the DMCA.
The various people who have suggested the DMCA be changed
have various vested interests.
As you know, DMCA was signed in 1996, under heavy lobbying
from Hollywood and other things.
What's interesting is it's being turned around from their
perspective in a negative way, but it is nevertheless-- from
my perspective, it's a pragmatic issue.
My own view is that the user content explosion is so
profound, and so obviously a big phenomenon, that it will
eventually cause the world to change.
It's also worth noting that the DMCA US-only, and that
laws differ by country.
And it's a real problem if you're trying to run a global
site, as we are, because the laws differ and you have to
deal with each--
obviously, you are subject to all the different laws.
WALT: Yeah, but just to continue on this theme for a
minute, I don't think it's in the 10 Commandments or the
Constitution that the terms of the DMCA are
not etched in there.
It's a law.
It can be changed.
It can be amended.
ERIC SCHMIDT: --by the Congress, not--
WALT: By the Congress.
I also-- the last time I checked--
Google actually had a right, and the people at Google, or
the community around YouTube all have a right to express
their opinions to Congress.
ERIC SCHMIDT: Absolutely.
WALT: And I'm not sitting here telling you what your policy
should be on this, but I would say, and lots of other people
would say, that there's some sort of imbalance in these
copyright laws.
ERIC SCHMIDT: One of the great things--
WALT: This is not to say, by the way, that there isn't such
a thing as copyright or there isn't such a thing as IP.
ERIC SCHMIDT: What's great about it is that's your view
and a lot of people agree with you, and the YouTube
communities form so naturally around these things that it's
a reasonable expectation that such a community will
ultimately drive that change.
What's important from a Google perspective is that we enable
that, not that we particularly endorse one political view or
the other, and that's where we draw the line, because we
don't do content.
WALT: You don't do content.
But you enable content.
ERIC SCHMIDT: We absolutely enable content,
especially for end users.
We're very end user-focused.
WALT: I mean, actually, Bill Gates and Steve Jobs were
saying, we're just platforms enabling content, in their
session last night, that they're not real entertainment
or content companies.
So in and sense, you're aligning yourself with their
point of view on that, of what the right role for the company
is in this?
ERIC SCHMIDT: I can't speak for them.
From a Google perspective, Google is run around end
users, right?
It's all around end user testing, end user happiness,
essentially.
And all of the innovation is really about helping them
understand information better, providing more valuable
services on various kinds of devices.
And there is a huge, huge--
essentially, in traditional terms-- platform shift from a
model which is sort of a hard-wired platform to this
much more diverse, sort of more porous, platform model
that is the web.
It's very exciting.
And it's how these platforms and services will evolve.
It's becoming more confusing because it's becoming more
embedded and more powerful.
There's no single answer to that anymore.
WALT: So Steve Ballmer was talking here yesterday, and of
course he was talking about how they're way behind you in
search and in advertising.
We know they just spent $6 billion to try to buy some
expertise on the advertising side in particular, and tried
to buy DoubleClick before you did, and he got very close to
my face and in a very loud voice said, we're determined--
what did he say?
We're determined, and we're patient, and we're going to
make big strides here.
But one of the things he said is that the presentation of
search results-- the user interface for
presenting the results--
has been relatively stagnant.
Would you agree with that?
Is there a need for--
I mean, for all of your phenomenal success, is there a
need to make some strides in the way
you present the results?
ERIC SCHMIDT: Well, this past week we announced something
which we call Universal Search, which is the blending
algorithmically of results from video, and audio, and
books, and so forth and so on--
news, which is a big component of it--
directly in the search results.
So far, the feedback is that people love it.
We, of course, had tested it before we rolled it out, and
it looks like we'll roll it out pretty soon worldwide.
So that's an example where a relatively simple presentation
has been augmented.
I don't think we're going to go very far from the single
search box.
One of the things to know is that we are seeing tremendous
success with personalization of Google.
It's called the iGoogle Page.
WALT: Yeah.
ERIC SCHMIDT: And many, many, many millions of people are
now personalizing their search experience.
From my perspective, this was a surprise, because I figured
people would like that very simple box.
But in fact, if you think about it, people know what
they care the most about, and so they
personalize their page.
And these new web gadgets, as they're called, allow you to
have a highly personalized experience.
So what's happening is you have sort of the corpus of
Google available to these gadgets, the gadgets
themselves can interact with each other, and then you can
have a personalized platform experience.
It's a very, very powerful model, and it's likely to be
one of the major ways in which people consume information.
The other thing to note about iGoogle is that that personal
experience also extends to non-PC, non-Mac platforms,
anything that can handle a reasonably
sophisticated browser.
So all the new generation of mobile phones, and new devices
should be able to run them, and that's pretty exciting.
WALT: It is.
It's similar to some other things that you can do on the
web, Netvibes, even My Yahoo, to some extent, and on the
desktop with the dashboard on the Mac, and
the sidebar on Vista.
But still, when you type in a search--
even from an iGoogle page in that box--
it looks a lot like it looked three years ago.
Maybe that's good.
Maybe that's fine.
But in the world you live in, three years is a really long
time for something to still look pretty much the same.
ERIC SCHMIDT: I disagree a little bit, because although
the look may be the same, the amount of content and the
types of content are vastly larger.
With all the new content that's available on Google,
the index is infinitely larger, there's so many
different services.
WALT: Right.
I'm talking about UI and the presentation.
ERIC SCHMIDT: Understand.
The other thing that we did is at the very top left hand
corner of most Google pages, you now get a menu of
services, essentially, online applications
of one kind or another.
So that you can not only have a personalized page, but you
can also then go into the depth of the data services
that are offered by Google, again, worldwide.
WALT: Why don't you have a better sense of the preview of
the pages that you bring up?
I know I'm being a little granular here, but some
competitors--
Ask.com is one-- have a little icon and if I mouse over it, I
can see what that page would look like before I commit to
clicking, going over there.
Is that--
ERIC SCHMIDT: We have looked at that.
It hasn't been a big--
it has not tested particularly strongly one way or the other,
and there are some latency issues with such previewing.
WALT: So that's not something you're interested in.
ERIC SCHMIDT: Well, we're always interested.
We're always experimenting.
So if you look, for example, in Korea, we just announced a
very different browsing interface, which actually has
pop-ups as you browse along the fundamental home page.
So we're experimenting.
The user interface stuff is culturally specific, country
specific, language specific, and you'll see more of that.
WALT: Also here at this year's D, we had a demonstration of
Mahalo, Jason Calacanis's new product, which is a kind of a
Back to the Future thing.
You probably know what it is, where he has human editors.
And his contention is that for a certain percentage of
particular kinds of searches, your algorithms aren't very
satisfying to people.
He showed a little video of people saying, I can't figure
this out, or I'm not getting what I want.
And they were areas like, I want a particular hotel in a
particular city.
There is this enormous search engine optimization industry,
which is essentially, gets up in the morning every day and
tries to game your system, and it isn't always
good for end users.
And in some categories, I can tell you just as a user, I
think he's right.
I'm not saying his solution is the right solution, but I
think he's right that there's an issue.
It does frustrate me, at least, when I go
to look for a hotel--
maybe this is just me, but the things that I want most are
the website of that hotel and some sort of review that I can
trust. But instead, it's just pages, it seems to me, too
often, of brokers, middlemen companies, that claim to be
able to get me a room for a certain price, but I don't
even know which hotel I want to stay at.
Is there a solution to this?
Is Jason's solution the right solution, to go back to the
Yahoo Directory approach and put-- at least for some
searches-- put some human editors in, or can you do it
with changing your algorithms?
ERIC SCHMIDT: We believe we can do it with changing the
algorithms. The problem with these human approaches is you
can always pick a category, and a very talented human
editor can always do a better job.
The problem, of course, is that doesn't scale.
It doesn't scale to 100 languages worldwide, and the
billions of pieces of information
that people care about.
There are new artificial intelligence techniques that
can use both the personal information that you have, as
well as all the other information and signals that
we have, that can produce a pretty good approximation of
what a human editor can do.
The thing that you're referring to, we call index
spam, people who are paying money to try to generate false
results and/or overwhelm the user.
WALT: Not paying money to you, but--
ERIC SCHMIDT: No, no, they're paying money to other people.
The classic example is called a link farm, where people take
5,000 computers and they cross link each other.
But we, of course, can detect link farms now, and our
software is very good at that.
But there are other attacks that people try, because it's
in their financial interest. I think one of the things that's
interesting about the web is the web is now so important
that there's a certain amount of misinformation that's
commercial or manipulative in nature, or you could call it
biz-information, or some term where it's sort of boosted, if
you will, by a financial interest. And people will have
to learn that sometimes the marketing has been infused
with the result, and we try to detect that as much as we can.
The particular issue with reviews is one that I, for
example, notice all the time, and we have better approaches
that are in development for that.
WALT: You do?
Because I think it is an issue.
Whatever we think about whether Jason has a business
or someone else has a business.
Let's talk about advertising for a minute.
You have this phenomenally successful contextual
advertising business.
I will tell you, just as a consumer, that I appreciate
that you haven't done garish, flashing ads, the ads or text
are clearly demarcated from the algorithmic results, which
some other search engines didn't do.
I mean, I think you've done a lot things there correctly.
At D1, I don't know if you were here,
but I remember that--
ERIC SCHMIDT: I was.
WALT: I think it was Sergei made the point that those ads
were great because he was able to find a green laser there.
And Larry said that he was able to find some kind of
metal shed for his yard or something, I don't know.
Those with their examples, but you know,
they're good business.
ERIC SCHMIDT: I'm sure that's true, given the two.
WALT: It's been a very good business.
You obviously have just bought a company that
does display ads.
So are we going to see your--
is this more of a strategic thing for you to be in the
business of serving and placing display ads all around
the web, or is this a portent of your changing your pages to
have a lot of display ads on them?
ERIC SCHMIDT: Well, we're certainly not
going to do the latter.
A little bit of background for everybody is we have a
different view about advertising.
We think that the ads themselves have value.
There are lots of examples where the value of ads shows
up in our systems. We show fewer ads now because we can
target them better.
Those ads themselves have become more valuable,
therefore, our revenue has grown.
So we're showing fewer ads per page now than we ever have.
So more targeted, more personalized, ads are the core
part of how our business model works.
Of course, we did that very successfully with text ads.
We believe that the same technology and approach can be
applied to what are traditionally known as display
ads and other kinds of ads-- radio ads, television ads--
WALT: To make them actually--
contextually relevant.
ERIC SCHMIDT: To make them more relevant in the scene.
And every time we have done that, we have seen a
significant improvement in user satisfaction, advertiser
satisfaction, and frankly, the value of the ads goes up.
So it's a good business.
It's different from traditional advertising models
where if you show more ads, you make more revenue.
In our case, if we show fewer, more targeted ads, we
literally make more revenue, because we run an auction.
And the auction is, as you know, is a second price
dynamic auction.
The second price dynamic auction produces a higher
price, therefore, all is good.
And we know the ads work, because there is a process
where the value of the ads and the ultimate cost per
acquisition--
so-called CPA--
is solved through ad and display tracking tools that
are in development at Google.
So the sum of all of that means that as we broaden our
mission from an advertising perspective, we really want to
solve the advertising problem in general.
In other words, we want to bring engineering to
advertising as best we can.
WALT: You're not just buying DoubleClick to go on with what
they've been doing.
I'm not trying to get you to criticize what they've been
doing, but you want to actually use that as a vehicle
to apply some of what you've done in the
search ads to this category?
ERIC SCHMIDT: That is correct.
If you look at what advertisers want, they want
measurability, efficiency, reliability.
They want the system to work, and these are big systems,
they work at scale.
DoubleClick's customers, as best we can tell given the
ability we can see into them under the current--
because we haven't closed yet--
are mission-critical, large corporations that use them
every day all the time.
It's a core component of their business.
If we can make that business, again, more measurable,
stronger, more targetable, it's good for the advertiser,
it's obviously good for the publisher of the information.
We can also, using our tools, show fewer, better ads, which
is good for the end user, which is the ultimate goal.
WALT: Why do you keep saying end user?
This is not a corporate IT product, Google's products for
the most part.
You have direct users.
They're not end users.
ERIC SCHMIDT: Thank you.
Direct users.
It's important to distinguish because the company is run
based on the testing and sampling of what you call
direct users rather than some other strategic goal, so we
will, for example, trade off revenue or other things in
favor of an end user benefit.
There have been numerous cases where some very smart set of
product managers have come in with some very clever revenue
enhancing idea, which when tested does not improve end
user satisfaction.
We don't do it.
WALT: So that's the don't be evil?
That's the do the right thing?
ERIC SCHMIDT: Don't be evil is both a slogan, but it's also a
way of getting people to think before they act.
So one way to think about don't be evil-- and when I
started in the company, which was a few years ago, I thought
this was a joke.
And so I'm sitting around the table.
It's a small company and we're in the one conference room,
and there's some conversation about this revenue issue.
And one of the engineers, whose name was Ron, slams his
fist on the table says, "That's evil." I thought, wow.
So then there was this big debate as to whether it's
evil, and it turns out, he was right and we did not do that
particular thing.
And my point here is it's an internal way for people to
understand how they should be making choices.
Since there's no book on what's evil and what's not, it
forces the conversation.
WALT: The collected speeches of George W. Bush have a lot
about evil and good in them.

Let's talk about evil for a minute, though, because
there's an increasing point of view that you are accumulating
so much power--
at first, a couple of years ago, it was so much power on
the web, and now it's so much power in the advertising
market in particular--
that you could be evil, whether you
say you are or not.
Is it a good thing--
for the advertising market-- is it a good thing for
publishers and content owners for you'd have like 80% of the
power to serve and place ads?
ERIC SCHMIDT: Our model is based on real performance.

I understand the concern, because I've obviously heard
it and read it.
Ultimately, we're different from some of the previous
incarnations of large companies, large monopoly
power, that people have been alleging, because we are one
click away from losing that end user.
Advertisers, by the way, can choose to advertise on many,
many sites, not just us.
We have a whole bunch of commitments that we made to
our constituencies.
In particular, we've made a commitment that we
won't trap user data.
That if you, for example, become dissatisfied with
Google services, we will make it easy for you to switch to
another service.
There's two reasons for that.
One is, clearly for you as a user, but it's also important
because it serves as a meter, if you will, or a discipline
on the company and on the teams in the company, because
it forces them to compete on a product-by-product or an
integration basis.
They can't just say, hey, we've got them trapped.
It forces excellence.
WALT: But Eric, you're an incredibly smart guy.
I mean, you obviously know that I'm not talking about
whether somebody would switch from iGoogle
to My Yahoo or something.
I'm talking about, consumers don't know who's serving these
ads on these different sites.
They're aren't even on a Google site.
They're on other site and DoubleClick is the company
that serves the ads.
Now it's you.
It's Google.
A page might have a bunch of Google Ad Sense, Ad Words,
whatever you call it, ads.
ERIC SCHMIDT: Ad Sense for content.
WALT: OK.
And then banner ads.
And all of a sudden, the difference between whenever
you close this deal and before is you're controlling all of
it on that guy's page, on that website.
Is that good?
Is that a healthy thing?
ERIC SCHMIDT: Well, the easiest answer to understand
is that the publisher of the content gets to make that
decision and they get to make the decision whether it's a
good thing or not.
So far, they've been voting that it is, but every day--
WALT: But they haven't had the chance to vote for a Google
that controls both the display and the contextual.
ERIC SCHMIDT: But as you pointed out, there's a direct
competitor with DoubleClick, which is [? Aquantum. ?]
And you started our conversation by mentioning
that acquisition.
So indeed, they do have a choice.
And they can decide to mix and match.
There's no particular reason why you have to choose one
versus the other.
Each one of our products has to stand on its own.
WALT: Talk to me about how you evaluate Microsoft as a
competitor.
In other words, you know what business you're in and you
know more than you're probably willing to say here about what
businesses you want to get into.
You know your strengths and you know your weaknesses, and
you've been competing with them for a long time in
different companies and incarnations.
Steve Ballmer said that they've had a couple of
traditional business models and lines of business, desktop
computing, obviously, Enterprise products, and they
now have two new ones, one of which is advertising and one
of which is consumer electronics kind of products,
the Zune and so forth, the XBox.
They actually demoed another hardware product here.
But he put advertising as one of their core businesses.
You know, essentially saying, obviously it's not as big and
they're not as successful at it, but that it's on a plane
with Windows and Office.
Do you worry about that?
They have 78,000 people.
They have a lot of money.
ERIC SCHMIDT: Well, I think you always worry about a
company that has that percentage of the platform
business in terms of Windows, and that's been true for many
years, certainly in my professional career.
Advertising is different from the other businesses.
If I, for example--
one of the things that people miss about advertising is it's
not a zero sum game.
If a new entrant enters the market, it just creates more
choices, more opportunities.
Because our auction clears at the market value to the
advertiser, they're not going to somehow reduce the amount
of money that they're giving us.
They may augment their money.
They may augment their spending.
And the competition is, in fact, good.
It creates more choices for end users.
It's more choices for advertiser.
As you pointed out earlier, it creates more choices for
publishers.
I think the competition is very good.
The advertising business is harder than it looks like.
Again, when I started at Google, I thought, it's sort
of a toy system.
It's very straightforward.
You take these key words and so forth and so on.
The computer science and the complexity of running these
systems, especially when they're under attacks like the
ones that you mentioned about commercial spam and so forth
are very, very sophisticated.
Google has some of the top computer scientists in the
world, and some of the largest data centers and super
computers and so forth that work on this.
So again, the challenge is high but the
opportunity is worth it.
The current structure of the market is you have Google's
network and Yahoo's network, and you have Microsoft trying
to enter, as they have for a while.
WALT: So talk about Yahoo in a similar way, as a competitor.

My partner, Kara Swisher, wrote a piece recently
suggesting that maybe they'd be better off taking
themselves as a media company and not being in this
advertising technology business.
Would that be your advice?
How do you look at their whole situation?
ERIC SCHMIDT: I don't think it's appropriate for me to
give Yahoo advice.
I'm not there.
Every company faces its own challenges in a different way.
Yahoo is a combination of both very interesting and very
successful content and a technology base, and many
people have suggested that those two should be separated.

From our perspective, Google is a technology company, a
product company, an enabler of these services.
We're not in the content business and we're very proud
of the quality of our search and advertising and all the
new applications and the new web applications that are
getting developed, and that's our focus.
WALT: There's been a lot of rumors over the last few
months on the web that you are getting into the cell phone
business in some way, shape, or form.
Some kind of software--
and other people say there's going to be an actual Google
handset, maybe a Google phone carrier.

It is pretty clear you have some work going on somehow,
somewhere around the cell phone business.
Can you talk about what you're doing there?
ERIC SCHMIDT: We've internally saying,
mobile, mobile, mobile.
And the simple reason is that everybody here in the room has
their mobile phone with them.
Everybody in the world has their mobile phone with them.
When I travel, all I hear is the sound of mobile
phones going off.
And mobile phones have historically been pretty weak
from a standpoint of data connectivity and browser
capability.
Their browsers have been crippled, their data networks
have been slow.
The data networks, now with the 3G rolled out, are roughly
one megabit or greater.
That's sufficient for some pretty interesting
applications.
The browsers--
there's a next generation of browsers,
epitomized by the iPhone.
I'm on the board of Apple so I should disclose that.
But the iPhone is a good example of a--
WALT: Oh, Steve was here waving it around, and I held
it for 20 minutes, so I know what you're talking about.
ERIC SCHMIDT: More than I've held it.
WALT: What?
You're a board member.
ERIC SCHMIDT: That's right.
I'm waiting for mine.
So in any case, this new generation of browsers is
very, very powerful.
So the combination of the 3G network and the browser really
means that instead of saying mobile, mobile, mobile, I
should be saying, apps, apps, apps.
And the model on phones is not going to be the simple Google
search page with this long set of results, and all the ads on
the right, and so forth and so on that everybody is used to,
because the screens are so constrained.
And as fast as the processors are getting in as good as the
battery life and so forth, the screens are not
going to be this big.
They're going to be smaller, and so they're going to have
different kind of applications.
The applications are going to be web enabled.
They're going to use either Java or Javascript, and we,
Google, are building a lot of those applications in
conjunction with the operators.
The operators have various complicated business models
for distributing those apps and using them, and a few of
them are beginning to be successful.
Our most successful partnership right now is with
KDDI, which is a Google--
it's the number two in Japan--
Google search, Google ads, very, very profitable.
What's interesting about the ads in the mobile phone is
that they are twice as profitable or more than the
non-mobile phone ads, because they're more personal.
So if you think about the phones, there are more camera
phones now that are in phones--
sorry, more digital cameras that are in phones than
digital cameras.
Every phone either has a GPS or a pseudo-GPS
and is also a computer.
Therefore, for purposes of argument, every end user
computer is a phone, a GPS, and a camera with a
modest-sized screen.
The sum of all that is a very interesting apps platform, and
those apps are going to be doing all
sorts of personal things.
You have all the Google data repository behind it, all the
personal information, the sum of that is a whole new
platform plug.
So indeed, Google--
WALT: But are you going to just--
I mean, for you to write apps for mobile phones
is not a new thing.
I mean, I have on my Trio Google Maps.
People in this room have various--
not just through their browser on their phone, but you've
written a number of clients that do Search and other
things, and you have specially formatted Google pages that
look better on these crummy cell phone browsers.
So the idea that you're going to write a bunch of apps is
mildly interesting, but that's not what all these web rumors
are talking about you doing something more fundamental.
Like I said, some say you're going to make a phone.
Some say you're going to make an entire platform for phones.
That's a much bigger deal than just writing apps for the
existing platforms.
ERIC SCHMIDT: In order to write those apps there are
also mid-layers that have to be built that actually,
essentially, federate the services
and make them available.
There's a lot of other technology
that has to be built.
So when I say apps, I don't mean just the simple app.
I mean the infrastructure required to host and make that
app personal, extensible, shareable, and so forth.
The new model of these phones is going to be person to
person, right?
It's going to be phone to phone.
People are going to be sending information, videos, humor,
data, and so forth and so on.
It's sort of SMS gone wild in a fundamental way.
So from our perspective, we're building that software and
we're doing it with partnerships.
We just announced a partnership with LG and with
Samsung to do some similar apps.
The most successful of the Google mobile phone apps so
far is Google Maps, but there are many more coming.
WALT: So this middle layer, this platform thing.

I'm sure you've heard Steve talk about this because you're
on his board, Steve Jobs.
Not Steve Ballmer, I don't think you're on his board yet.
But talking about crippled software--
I can't remember his terms-- baby software, and bad
software on phones today.
And he has obviously done a port of the Mac operating
system to this iPhone.

Is Google, as a technology company, a software company,
able to step up and do what Microsoft, for instance, does,
which is to write a whole software environment, an
operating system or something that could go on a whole bunch
of kinds of phones?
And then your apps, and maybe other people's apps, would run
against that?
ERIC SCHMIDT: Without announcing new products and
new API layers, let me describe that the model is not
as simple as an operating system and a platform, which
is sort of the traditional way we talk about this.
It's really a set of services.
This week, in fact, yesterday, Google had a developer day,
which is all focused on building the piece part apps,
the so-called gadgets and the infrastructure, that are
embeddable in this new platform.
The most likely scenario, from a Google perspective, is to
build some, if you will, inspirational platform apps,
but primarily focus on getting third parties to do it,
because that's where the innovation will come from.
WALT: OK.

Why don't we take some questions.
Thank you very much.
[APPLAUSE]
ERIC SCHMIDT: Thank you.

WALT: Right here
MALE SPEAKER: I wanted to ask about Google Gears.
like up until now, Google has mainly been something we use
when we're connected.
With Google Gears, are you going more directly against
traditional client software companies, and will we be
using Google when we don't have internet access more?
ERIC SCHMIDT: As you know, Google Gears is an open source
software that we released, essentially, today which
allows you to have your internet experience when
you're not connected to the internet, and then when you
reconnect, it does the necessary re-syncing.
It does it with a relatively straightforward database.
It's one of the biggest requests that we get in our
new Google apps model, where people are--
they want to depend on the web service.
They want to be able to keep their personal information
somewhere, but we all understand that we're on
airplanes or places which have poor connectivity, and it's a
pretty uniform solution to that.
So we'll see.
The important thing about Google Gears is it's not
targeted just at Google.
We're trying to do it with a whole bunch of companies to
try to, essentially, create this new opportunity on
platforms. The term you used was "Go against the
traditional client software companies." To my knowledge,
all of the existing client software companies are
offering variants of web services to access client
software for precisely this reason, everyone's
moving to the web.
MALE SPEAKER: Thank you.
WALT: Esther?
ESTHER: Hi, thank you.
Last year, I asked about personalization, maybe not
here, but somewhere, and it happened.
The question now, it's an advertising thing.
Both what Google's own experiences, and what you're
seeing, and what you expect DoubleClick will make happen.
At one end, there's a lot of click fraud, and there are
lots of arguments over how bad it is, whatever.
People don't really care about click fraud.
They care about paying for clicks that aren't valuable.
And so that means a better way to do it may be to pay per
action, whether it's
purchasing history, or whatever.
It requires more tracking, but you're pretty good at that.
So where do you see that going?

ERIC SCHMIDT: Today, the majority of our revenue comes
on a click basis, so people pay for the click.
And the sophisticated advertisers have teams, that
is, groups that actually measure their conversion rates
and essentially construct their own CPA.
I was surprised to discover how proprietary they view that
whole process.
They don't want, for example, to turn it over to Google or
to anyone else.
They view it as sort of a key insight into their business.
They know if they do this, the following thing
happens, and so forth.
So far, the click fraud issue has not turned
out to be a big one.
We're always worried about it.
We're always trying to detect it and so forth.
But remember that the auction is self-policing, in that if
there were, for example, an infinite amount of click
fraud, then eventually, the conversion rate, the actual
selling rate, would decline and the advertiser would say,
hey, I'm not getting my money out of this advertising.
I'll move my money to some other solution, or some other
company, or some other meta-group around advertising.
The same technology argument should work pretty well for
display ads.
WALT: Yes sir?
MALE SPEAKER: Hi Eric.
Taek Kwon from TPG, the [? Biotech. ?]
I had a question around the domain industry, domain
parking, domain tasting, and all the associated practices.
What is Google's position on that entire industry?
ERIC SCHMIDT: We have a business that we acquired that
does in fact do some domain hosting and
ad serving for domains.
We've been careful to try to evaluate that based on end
user feedback, as opposed to just purely trying to load
these things up with fake domains, because we don't want
to be part of the problem.
We want to be part of the solution.
So our solution seems to have better conversion and tests
better around, essentially, domain management.
There are companies in the industry, they essentially
take and they fool end users.
They claim they're one thing and they're
really something else.
We have software that can detect that pretty well,
although there are some counter examples.
And every few months, we come up with yet another attack and
then we eliminate it.
So with respect to domain hosting, there are some
legitimate players and there are some that are essentially
manipulative.
WALT: Yeah.
HERBERT KEN: Hi.
I'm Herbert Ken with Codeworks in the UK, and a fellow Tiger,
class of 1989.
It would be an understatement to say that you guys have
obviously had a pretty good run over the last six or seven
years, and you've obviously had a lot of momentum, et
cetera, et cetera.
But given the dynamism of the technology and the media
industries, inevitably, one would suspect dark clouds will
arise, and problems will come, and competition,
et cetera, et cetera.
So I guess I'm wondering what you and your team are doing to
build into, if you will, the DNA of Google, that will allow
it to be so-called built to last when, perhaps, some of
these harder times arise?
ERIC SCHMIDT: Companies are motivated by culture, and the
culture is determined by the people, the leadership, the
tone, the way people view themselves.
We've worked very hard to build something that would
outlast me, outlast the current management, outlast
everybody, because we think that the opportunity for
building a different kind of company is correct.
The company is run in a different way.
The empowerment model is different.
And we've talked about it, so I don't need to repeat it for
this audience.
And those seem to be pretty much likely to survive any of
the possible--
either competitive issues or maturation issues, and I don't
think those are going to change.
The other thing to know about companies is that they do, in
fact, have to change.
Last night, Steve Jobs, when you asked him what was the
company like 10 years ago, he said he cast away the past to
move to the future.
And I think that's a good message to all companies.
We're always thinking about what's next, what is our great
opportunity, and not rest on our laurels.
And we try to instill values in our company.
There's a tremendous number of new people who are joining the
company for whom this is day one in their Google
experience, and they will shape the
culture going forward.
And the great companies--
and many of them represented by people who have been on the
stage before me-- have all been through those
transitions, and I'm quite confident that we will get
through them too.
MALE SPEAKER: Thanks.
CRAIG FOREMAN: Craig Foreman from Earthlink, and I should
disclose, we have a long history and flourishing
nuanced relationship with Google.
Eric, I'd like to ask you a little bit, lest this audience
go home with more paranoia.
Tell us a little bit about the experiences you've felt
personally as you've worked on radio ads, the D Mark
acquisition, and some of the experimentation that Google's
doing on non-IP based forms of media advertising.
Should we be thinking that it may take longer and be more
costly an effort to see some of the techniques that Google
has been so successful with emerge in other media types?
ERIC SCHMIDT: Hi, Craig.

Each of the advertising models is different and
they work for a reason.
Radio advertising is the obvious example.
If we could get a unique identifier for the radio
that's in your car, we could certainly target it with
higher quality ads, we'd show fewer ads.
You'd be happier, the advertiser would be happier,
and the programmer would be happier because they'd be
using less bandwidth for the ad, which of course then
allows them to even more programming or whatever they
want to do.
Unfortunately, the technology doesn't allow that.
Each of the economic models has a different
profitability model.
So for example, radio is not as profitable, for example,
today as text ads because it's not as well-developed.
We looked at the ad models, and there's a number of people
who have studied this and indicated that, for example,
radio as an advertising business is under-monetized.
It's relatively poorly distributed, in terms of the
number of advertisers and the reach that it has.
So in the theory of trying to help people out by building
stronger and more durable advertising businesses, we
have a radio, we have a series of television trials people
know about.
We have the radio business that we acquired.
We're doing one for print around newspapers where we
essentially build a cross connect, where we
show the ads to them.
All of these are more targetable than the previous
versions but not as targetable as the core business.
So it will take longer, and they'll probably not be as
profitable initially as we develop these new markets.
And we keep experimenting.
We keep trying.
I think it's fair to say that the ad models that we have,
that we're all dealing with now, will all be changed.
They will all morph into something more dynamic, more
personal, more entertaining, because that's
how advertising sells.
WALT: Jan.
FEMALE SPEAKER: Eric, you were talking here about culture.
I think that--
I mean, Google, when it started off was a fun place,
lots of color, toys to play with, roller hockey, and it
was really sort of almost counterculture.
You've suddenly gone very dry and corporate.
What happened to Froogle?
Why do we now have the Google Internet Product Search, or
whatever it's called?
What happened to Froogle?
I mean, Froogle was a brilliant name.
And don't go dweebing on us.

ERIC SCHMIDT: You've not met the dinosaur we just installed
in our lobby and the space shuttle replica that we have,
and all the balls and so forth and so on.
I don't agree that the company internally has abandoned its,
shall we say, colorful and bizarre roots.
With respect to Froogle, it didn't test that well.
And we're replacing it with a much more integrated product
search offering, which end users tell us
is what they want.
I happen to like Froogle as a fun name, but in fact, the
integrated answer that we're developing is a better one.
FEMALE SPEAKER: You can't call it Froogle dash the internet
dot search?
ERIC SCHMIDT: Every idea has been tried.
WALT: Yes.
NINA LITTON: Hey Eric, it's Nina Litton.
I was wondering, when you were describing what you're working
on with your phone friends, and you were mentioning that
you would you inspirational applications, would be a fair
guess that what you're really building is a middleware
platform that's going to be in the server somewhere and what
you're showing people on the new phones will be ideas that
the open community can build off of and create more of?
ERIC SCHMIDT: Well, the term middleware is sort of a--
the term middleware, as you know, is an IT term.
It's better to think of it as a set of programmatic layers.
NINA LITTON: --services, OK, programmatic services that
live in the cloud.
ERIC SCHMIDT: And you actually need them on both devices.
You need both on the client as well as on the server.
The client needs to be able to authenticate, to offer the
services, to be able to talk, and so forth and so on.
The computational power of phones now enables that.
So as part of what we're doing, such layers are being
developed as part of building these interesting phone apps.
It doesn't have a name, and it's not some new platform
strategy, and it doesn't have a book that says, you must do
this or else, or so forth and so on.
The new developer model is different in that sense, that
it's much more grab bag--
take this, take this, take this-- from different vendors.
It's much more porous.
The other thing that's interesting about these new
developer programs-- because, of course, you've covered them
for many years as many people in the audience--
is that they're much more viral, and they're much more
community based.
They're viral in the sense that they spread out, rather
than being purchased in a store or whatever.
And they're community based because Walt shows it to me, I
show it to you, and then the rest of us talk about it.
NINA LITTON: Cool, OK thanks.
WALT: OK, we can only do two more, and
these two were standing.
I'm sorry.
MALE SPEAKER: A few months ago, a large company
introduced a product similar to ours, which is fine.
Shortly after that, if you Googled our company name, our
company name was the first one that came up.
But if you clicked on our company, you immediately went
to the ear phone site of the other manufacturer.
We spent a fair amount of time on the phone in correspondence
with Google, and was told that there was nothing illegal, and
that that was-- we could basically get into a bidding
war for our company name.
Did we just not talk to the right people, or is that
actually your policy?
ERIC SCHMIDT: You're talking about the ad side of the page,
or the algorithmic search?
MALE SPEAKER: The algorithmic search.
We came up the top of the search, and you clicked there
and you went.
ERIC SCHMIDT: So again, there must be some miscommunication,
because there are two parts of Google.
There's the algorithmic search and there's the ads.
We do allow competitors to advertise on other people's
trademarks.
This is controversial.
We've allowed that for a long time.
So it is possible that someone is advertising in a way that's
not to your liking on the ad side.
We do not allow that and we police it very, very carefully
around algorithmic search.
So if, for example, your web site comes up under a query
for your product, it should absolutely
take it to your website.
There are situations where people have done something
called interstitials, where they'll actually superimpose
the wrong website, and we detect those now.
So again, there must be some miscommunication on your
specific case.
MALE SPEAKER: I'm glad to hear that.
I'll send you the correspondence.
ERIC SCHMIDT: Thank you.
WALT: Susan, last question.
SUSAN LYNE: Yes.
Hi, Susan Lyne.
Could you talk a little bit about your experience in China
and the enormous growth of Baidu, and maybe what you've
learned about cultural barriers in
countries like China?
ERIC SCHMIDT: Well, you know, China is a challenge for many
American companies, and maybe all foreign companies.
As everybody in the room knows, we're subject to all of
the appropriate Chinese laws, and we're not necessarily wild
about these laws.
In fact, we don't like them at all.
As a result, we delayed our entry into the market for a
pretty long time.
During that time, Baidu became more successful, and of
course, they're also subject to the same laws.
When we decided to enter the market, we also decided to
enter it in a somewhat different way.
So for example, if information is omitted, we actually inform
the end user.
We also don't keep any user generated content.
The sum of that is consistent with the principles that
Google has set out, and after a long and complicated debate
that everybody participated in, and we are comfortable
that we made the right decision in doing that.
So we'll accept the consequences.
What's interesting is since we've entered and since in the
last six months, our market share is growing and our
traffic has grown very significantly, which tells you
that the Chinese end user also wants more information, global
information, and so forth, quickly
under the current model.
So I think the strategy seems to be
working with a late entry.
WALT: Thank you very much, Eric.
ERIC SCHMIDT: Thank you very much, Walt.
Thank you all.