Jonathan Rosenberg: Inside the Black Box


Uploaded by Google on 13.03.2008

Transcript:

MALE SPEAKER: Good afternoon.
Welcome to Marian Miner Cook Athenaeum.
It is always a pleasure to welcome one of our
distinguished alumni back to Claremont.
And as a graduating senior, it's even more exciting for me
to see just where a CMC education can take me.
Our guest today, Jonathan Rosenberg, graduated Phi Beta
Kappa from Claremont McKenna in 1983
with a degree in Economics.
After a short stint at the University of Chicago, where
he got his MBA, Rosenberg dove right into the burgeoning
information technology sector.
He first worked as Director of Product Marketing for Knight
Ridder Information Services, where he helped deploy one of
the first online relevance ranking engines before moving
to Apple, where he managed their eWorld
internet product line.
Rosenberg then went on to help found the @Home product group,
eventually becoming Senior Vice President of Online
Products and Services for Excite @Home home.
But that was just the beginning.
In 2002, Rosenberg started working for what may well be
the coolest company in the high-tech world, Google.
At Google, Rosenberg is the Senior Vice President of
Product Management and Marketing, where he oversees
the design, creation, and improvement of the internet
giant's product line.
Google's influence cannot be overstated.
It's the largest American company that is not part of
the Dow Jones Industrial Average and it continues to
grow unchecked.
Indeed, we here at CMC are the beneficiaries of some of
Google's latest technology, the Google Apps platform,
which now provides our college email
service, courtesy of Gmail.
Mr. Rosenberg is a true friend of the college and embodies
the spirit of leadership and innovation that we hope to
instill in our students.
I am honored to welcome Mr. Rosenberg back to Claremont to
discuss what really goes on behind the scenes at the
Googleplex, Google's immense Silicon Valley headquarters.
Please welcome Jonathan Rosenberg to the Ath.
[APPLAUSE]
JONATHAN ROSENBERG: Thank you.
So, the people at my table have just demanded stories.
Right?
You just said I tell good stories, you want more story.
So I'm going to interrupt regularly scheduled programing
to provide some stories.
I did in fact graduate 25 years ago this spring along
with Matt [? Pikan ?] who has come to visit me.
And it was just four years earlier that my father drove
me down to Claremont.
I had him do what most of us have our parents do.
I had him help me sherpa my trunk of stuff into my dorm
room and get set up.
Then I sort of put my hand out and said, you know, Dad, I
need a check.
So he foolishly handed one over.
And, of course, he was then hoping to go meet my
professors and look around the campus and I said,
you can go now, Dad.
And so, my dad sort of headed out to the car, and I went to
wave goodbye, and I figured he was going to go
directly, what was it?
The 210 to 5 to the 152 to the 101 back to the Bay Area.
But it turned out that he stopped in Claremont because
he had something to say.
And he sent a postcard to my Story House address, and it
said something like the following:
Dear Johnny, Mark Twain once said that when I was a young
man, my father was so ignorant I could hardly stand to have
the old man around.
Then, when I came back at 21, I discovered how much smarter
he became in the intervening years.
I'm hoping the 60 grand I'm paying Claremont for you will
in fact make me much smarter.
See you again in four years.
Love, Dad.
Four years later, I was a lot smarter.
I was an arrogant graduating senior, to which Matt can
attest, and I was off to conquer the University of
Chicago Business Schools you alluded to.
I was, in fact, a lot smarter, and I actually kind of
delighted in showing how smart I was to other people
generally at their expense.
And this, I assure you, was not what my
father had in mind.
So I went off to the University of Chicago and two
years later, I graduated even more arrogant and even smarter
and went out to conquer Silicon
Valley much as you described.
And my first job was for a company called Operations
Control Systems that made data center management software.
I was given a lot of responsibility for a product,
and it failed miserably in the first year.
So, Pops came and sat me down.
He said, so what did you learn, son?
I said, didn't learn much.
Team screwed up.
My vision for the product was great.
I don't think I have much to learn from this at all.
My dad said, this time he quoted John Wooden, you know,
Jonathan, it's what you learn after you
know it all that counts.
Think about that.
It's what you learn after you know it all that counts.
So in deference to the fact that my father's $60,000
investment in CMC did actually pay off quite handsomely, I
titled this lecture Inside the Black Box: Technology and
Innovation at Google.
The title is not, in fact, original.
My father wrote a book the year I graduated called
Technology and Innovation Inside the Black Box:
Technology and Economics.
So it's in deference to my father that I now realize that
in my entire career I have done nothing quite as
scholarly as he did in his work inside the black box.
That said, I think the CMC senior that graduated 25 years
ago knowing it all has in fact learned a great deal since
then, and that's at the heart of what I'm hoping to convey
to you all today.
Was that a story?
I'll do some more stories, OK.
So, how many have been to my previous talks?
OK, well, that's a problem, because this is sort of
Jonathan 301.
Let me do sort of a quick a few minutes on
Jonathan 101 and 201.
We'll bring you up to speed.
You're really going to get your money's worth.
See, I'll do a whole course in five minutes and then we'll
move on to new material.
In my previous talks I talk a lot about the internet as a
change agent, how it's reshaping information exchange
and discovery.
Some of you may remember I had a pretty slide of a kid on a
long dirt road in Mumbai and pointed out that even these
kids have access to the same information as the scholarly
researchers at Harvard or Stanford or Claremont.
I also talked a lot how there's been this secular
shift in information and an agent of online advertising,
which has basically evolved to finance free products and free
software, and how there's this huge Cambrian explosion of
services that are now motivated and funded by this
free engine of advertising.
I've also tried to put in perspective my view of
business in the 20th century and why the big shifts in
technology cause the communications industry, the
entertainment industry, the technology industry to miss
the next big trend.
I've tried to explain why dominant players tend to blow
these transitions.
The last time I came, I talked about how oil fueled the
Industrial Revolution.
Much as oil fueled that revolution, bandwidth and
access to information and people spending online,
spending time online, and doing search, which drives
down information costs, basically cause all sorts of
commerce to occur online.
So for those of you who missed Jonathan 101 and 201, that's
basically it.
Then there's three books.
There's only three books you need to read.
I covered those too, a little, to some degree.
They're Chris Anderson's The Long Tail, James Surowiecki's
The Wisdom of Crowds, and more recently Don Tapscott's
Wikinomics.
So I always figured, I paid my professors to tell me the crap
that was in the books.
They never did that.
They always made me read the books, right?
But I'm going to just give you in one minute or less the
essence of those three books.
The first one, The Long Tail, basically talks about how if
you lower the costs of production and distribution,
you can offer a lot more variety.
And with more variety and the tools to find it, which the
internet offers, then people will gravitate towards their
own tastes.
And this whole concept of this tyranny of shelf space
limiting people to the hit-driven items that we see
in stores gives way and people can actually find specific
goods and services that map to what it is that they're really
interested.
That's basically the whole Long Tail.
Information costs go down, transaction costs go down,
economic activity spontaneously
combusts and goes up.
The second book, The Wisdom of Crowds by James Surowiecki.
This basically talks--
it starts from the premise of a British anthropologist,
Francis Galton, in the 19th century, who went to a fair
where there was a contest to weigh an ox.
And he looked at the guesses of the ox, and he discovered
that the experts were wrong, but that the average of the
crowd actually turned out to be better than any of the
singular experts.
From which, of course, he concludes on page 79 that we
learn that the average of a crowd is often much better
than expert, something that we see today in many prediction
markets and others, if you look at the IO electronic
markets, and if you look at the efficiency
of the stock market.
Page rank and much of what you see on Google today is a
manifestation of this crowd wisdom.
The third and last book is Wikinomics by Don Tapscott.
Basically it talks about this era of internet togetherness
leading to mass and global collaboration and people being
able to contribute with open standards like you see with
Wikipedia trouncing Encyclopedia Britannica with
real users doing work.
Linux is an open operating system having
individuals do work.
So those are the trends that I've talked about before that
successful technology companies and web companies
understand.
They're stitching together services through sharing.
They're not over-engineering a closed experience.
Flickr kicking the heck out of Webshots.
Wikipedia beating Britannica.
Bloggers beating CNN.
Facebook beating Friendster.
These are all examples of this kind of enlightened thinking.
So that's Jonathan 101, 201, and 301.
We had a pop quiz a little earlier.
Some of you did pretty well on it.
I sat down over the weekend and I tried to come up with
sixteen principles of innovation, things that I've
learned at Google that are different.
One of them stems from the concept of network effects.
How many of you remember a network effect?
What's a network effect?

Anyone?
AUDIENCE: The more you use something,
the greater the value.
JONATHAN ROSENBERG: The more you something,
the greater the value.
Of course, the seminal example being the fax machine.
For those--
do you even remember what a fax machine is?
The cell phone, the telephone as well.
VCRs, of course, once people came out with video.
The internet is all about this. eBay, the buyers go
where the sellers go, the sellers go where the buyers
go, and eBay keeps getting stronger.
Google is really based on this.
Users go where the information is so people bring more
information to us.
Advertisers go where the users are, so we get more
advertisers.
We get more users because we have more advertisers because
we can buy distribution on sites that understand that our
search engine monetizes better.
So more users, more information.
More information, more users.
More advertisers, more users.
More users, more advertisers.
It's a beautiful thing, lather, rinse, repeat.
That's what I do for a living.
So that's-- someone alluded to the engine
that can't be stopped.
Hiring is just like that.
If you think about all
institutions, think about colleges.
Colleges are great because they have great professors and
administrators.
And great students come.
Great students come because there are great professors.
Great professors come because they want to be
around great students.
Great alumni networks build out of great students, and so
more students come.
So everything, as information becomes easily accessible,
becomes a network effect.
Hiring is the same thing, and it's the single most important
thing that anybody does.
And at Google, we do it very differently.
We've always believed that As hire Bs and Bs
hire Cs and so forth.
You get As, Bs, and Cs.
You're in school, right?
You can train As to keep hiring As, but you can't train
second-rate Bs to hire anything better than Cs,
because second-rate Bs are threatened by As.
So you've got to set a very high bar
and maintain it forever.
You wanted a story.
I told the story in the previous discussion about my
second interview at Google.
The more interesting story is the first interview.
I came in, and basically the way we interview at Google is,
we like run a CAT scan.
We see if it reveals signs of electrical activity.
You can go, bluuuh, you know, or it can go, blip blip blip
blip blip blip blip blip.
And you say, hire her or leave him wherever he is.
So I had my CAT scan with Larry and Sergey.
And here's what they do.
They bring you in and they ask you a whole bunch of
questions, and then they have, some VC told them, ask the
person to explain something complicated.
And then you just sit there, and you make them talk.
And eventually, they tried to make it more and more
complicated, and they sound like blibbering idiots.
So I walked in.
Larry and Sergey said, mm.
Explain to me something complicated.
That's the way Sergey said it.
He sort of has a bit of a Russian accent which I can't
imitate that well.
You could imitate it.
AUDIENCE: Explain to me, Jonathan.
JONATHAN ROSENBERG: Something complicated.
Thank you.
God, I knew Matt came for something.
So I'm thinking, well, I got to explain something that I
know something about that they don't know something about.
Jonathan Theorem 12, which we talked a little bit earlier.
So I thought, well, theory of the firm.
I mean, you know, I used to sort of coach people at
microeconomics.
Let's start with a total cost function, Q of X. You guys
need to understand that there's an economic law that
states that marginal cost bisects average cost at the
latter's minimum, and this can be proven through Cartesian
geometry or calculus.
So I'm going to start with the cost function Q of X, define
average cost Q of X over X. You take the first derivative.
You set it equal to zero.
If you don't remember the quotient rule, because you
have X in the denominator it's easier.
You haven't had calculus in a while, you make it X
to the minus 1.
And then you basically take the derivative.
You have Q prime of X, which is marginal cost. You set it
equal to zero.
Q prime of X equals to Q of X over X. Right, Jerry?
Right, it works.
So I do this.
I'm very proud of myself.
And Sergey says, you can help with this, it looks like
first-year calculus.
So I'm like, shoot.
Something more complicated.
So I think, well, what did I study next?
Linear algebra.
I'm going to explain the simplex algorithm and how you
maximize an objective function at the corner solutions.
And then I'm like, oh, I just read their paper, and page
rank sums the eigenvalues and eigenvectors of
the web in end space.
I don't want to talk to these guys about linear algebra.
So I sort of looked at them, and they were like, I actually
interviewed two years before I took the job, so it was eight,
nine years ago.
And they were sort of young, disheveled
computer science types.
I thought, I need to move on to my home turf.
I will explain to you courtship.
And I basically gave them a lecture on
courtship and dating.

And they thought it was brilliant.
So brilliant people are a combustible situation.
We take brilliant people and we put them in small work
spaces together.
If you come to our offices, you'll see all the offices are
patterned after the size in the original garage.
Sergey and Larry actually measured it.
And people work in these little
offices in teams of three.
Working from home in a development environment is a
malignant metastasizing cancer.
You should never let people do it.
Doesn't work.
Put people together.
Don't hire specialists.
Small groups of specialists have lower standards and
aren't flexible.
Number one: hire great people.
I have fifteen more of these, so I'm going to go faster.
Number two: ideas come from anywhere.
Most companies say this, but they don't really do it.
They have like some little stupid idea box in the corner.
And they don't reward people for their ideas.
We actually, when we were small, had a meeting and we
called it The Idea Meeting.
And we do peer reviews, and we hire really smart people, and
they're rewarded for actually coming up with good ideas.
And the way you talked about your ideas was, you went in
front of your peers, like I'm in front of you, and you say,
here's my idea.
And you got to talk about it until they look bored or booed
you off of the stage.
Seriously.
So basically, we're taking the wisdom of crowds and forcing
you in front of your peers to showcase the idea that you
have. And later, when we became too big to do things
that way, we started to do the first round of
ideas through email.
People would send an email, and then we'd all
vote on that idea.
And the ones that got the most votes as potentially good
ideas got to go up on the stage and either get booed off
or show their friends that they had a great idea.
So you've got to make sure that you're accepting of ideas
coming from anywhere, but then you've got to implement a
system that actually delivers against it.
The third big thing that we do that's different is sharing
and openness.
The openness is the Don Tapscott stuff that I talked
about in Wikinomics, but sharing.
If you hire great people, you need to trust them with
everything.
So everything is on our intranet.
Everything.
We write the board letter to the board each quarter.
We send it to all of the employees so they know what we
think is important.
When somebody comes to me and asks for information, I tell
them, have you looked on our intranet?
If they say, no.
I say look there.
Then when I seem them again, I say, did you find it?
If they say no, I say, well then, go find it.
They go find it.
I say, did you put it on the intranet?
Everything is on our intranet.
The same standards of openness are applied to all of our
product development efforts.
If you look at the recent effort we did
with OpenSocial, right?
We basically have this protocol to share information
between websites that people have already visited.
The same thing with Android, the software stack that's open
that we're working on for mobile phones.
So, that's what you have to get out of Wikinomics.
You have to understand this concept of openness.
You have to understand this concept.
They talk-- how the Human Genome Project basically was
led by firms that chose to share with everyone because
they understood the Ken Arrow model of learning by doing.
You win, not by locking people in, but by being better at the
things that you've actually done.
We're better at search.
So even if we share all of what we do with people, that's
going to help them make us better and we're going to stay
better because we've learned more and we continue to hire
people who are better.
So we really personified this model of open that's eclipsed
the more traditional models, the AOL closed proprietary
systems model.
You can't control the platform anymore.
Consumers are in control.
Customers are free to move from one open platform to
another, and they're the key to the continued success of
web-based enterprises.
You also have, as I said, the internet driving costs and
transaction costs down.
And since you have high barriers to entry, you really
have the ideal model for creating value if you're in
this open and sharing world.
The next big idea that we have is to morph ideas
and not kill them.
And this also goes to a lot of what I talked about in some of
my previous talks about the economics of technological
innovation.
I've talked about how different inventions come on
and we don't foresee their real economic
impact for some time.
The transistor, the laser, the VCR.
I also talked about the steam engine, which was originally
developed for pumping water out of flooded mines.
Once you connected it to the railroads, basically, you
tamed the West. So there are all these technologies that
are proposed as point solutions to very narrow
problems. What we have today is fast change with the
underlying technology, the CPU power, the storage
requirements.
So you're constantly revisiting these ideas that
didn't work before and reapplying them.
Blogs were originally about publishing information so that
people could reach niche communities.
But the whole blog systems that we developed are now the
engines behind publishing information in Google Docs and
Spreadsheets, which is the next evolution for those who
are using Gmail.
We also used to have ideas that we've morphed like
convergence.
Convergence was originally about, remember, the device in
your pocket that would do everything?
How many power chargers do you take with you on vacation?
Six?
Convergence is not at the device level.
Devices will proliferate.
Convergence is in the cloud, where your Gmail is, where
your docs are, where your spreadsheets are.
All of your data needs to converge.
The devices that vector into your data
are going to diverge.
The opposite of what we originally expected, but still
the same idea.
The next big difference: users come first, not money.
Sounds simple to say.
Very, very few companies follow it.
You read the Founders' Letter.
You read Larry and Sergey's focus on the user.
I first met them and they said they didn't know how they were
going to make money, but they were going
to build great search.
They did.
Other companies started licensing our
search and our ads.
And it was almost comical the degree to which they would
become addicted to the heroin of revenue.
The more ads they put on their page, and the more they push
the user's search results down, the
more money they made.
And they were our partners.
And each quarter they would say to me, Jonathan, how do I
make more money?
And I said, well, you can make more money by dialing up the
ads on the page.
You end up being like Vogue magazine.
It's nothing but ads.
At least the people who read Vogue magazine want the ads.
On the internet, the people didn't want that scope of ads.
So every quarter, they basically would monetize
themselves out of share because they were focused on
money and not the user.
And we were the beneficiaries of that share, even though we
advise them against doing this.
Ultimately the winners become, in this new world where
information travels quickly, disinformation gets spread
online, but the truth also emerges faster.
This is true in politics and it's true in business.
So users ruthlessly punish companies that do the wrong
thing and richly reward companies that
do the right thing.
Did we need Steve Jobs to tell us that we didn't need to buy
ten songs when we bought music?
Did we need him to figure out that we all wanted a player
with only one button that normal people could figure out
how to use?
Sony Samsung, the whole MP3 world, didn't understand that,
and they lost, because Jobs, like Google, was focused on
the user, and maniacally so.
Users, not money.
Data drives all decisions.
Don't come into meeting at Google and say, I think.
Come in and show us.
And show us with data.
Not with data that you massaged in some spreadsheet
from some external source.
Data that comes from our log systems that
we know to be true.
In every conference room we have two projectors.
One you can put up your silly PowerPoint presentation if you
really want, and the other you have to show the source data
from that that others can question that show your
conclusions.
Argue everything about your conclusion based on data that
everybody else in the room can argue against. Don't come in
and say, I think.
Don't come in with your design and say, this
is a beautiful aesthetic.
Users will like it.
Try it online.
Come back.
Measure what worked.
And tell us with data.

Note the approach to data that the company took to
advertising.
There are all these famous sayings about advertising: 50%
works, I don't know which 50%.
Larry and Sergey did the opposite.
They took a return on investment-based approach to
measuring everything that people do.
Our users get conversion tracking software for free
that tells them whether or not they should lower their bids
or raise them, because we want them to understand with real
data what, on a bang-for-the-buck basis
they're getting.
In years, advertising is going to be a dashboard that a CMO
has, and he's going to figure out how much he wants to spend
to achieve a certain objective on television, radio, and the
online medium, and we're going to tell him on a
bang-for-the-buck basis against any particular
objective what works and what doesn't.
I have one other quick example about information and data.
I read an article of McKinsey Quarterly about sardine
fishermen in the Indian Ocean.
Sardine fisherman 101 basically look like this.
These poor guys go out in these
boats to find the sardines.
The goal is to get fish, get them on your boat, bring them
back, and sell them on the market.
And they're in these little villages that dot the
landscape like 15, 20 miles apart.
So the fishermen all go out in the morning and race to get
enough fish to bring it back to the local market.
The first guy back gets a good price for its fish.
But some days lots more fishermen come back with lots
more fish than the market expected.
The last fisherman in has a catch that basically spoils,
so he has to race back out to sea and go to the next town,
which is another day's ride on their little fishing boat, and
hope that they're not overstocked with fish.
Well, some genius basically said, wait a minute, data can
solve this, and put cell phones in the fish markets and
gave them to the fishermen.
Now the fisherman knows which village to head for
because he has data.
That fisherman doesn't want to think.
That fisherman wants real data.
And that real data is basically the livelihood
around their economic well-being.
Iterating products.
Most companies talk about this.
They don't really do it.
I used to be the king of writing product plans.
All my failed job experiences.
And when I arrived at Google, Larry basically said, when did
the engineers ever do a better job of adding features and
functions that you wrote in your product plan?
When did they ever do it faster than the
schedule that you had.
I said, well, never.
He said, well, then, don't write them.
Just get people working on a demo, iterate it, see what
users do, and make it better from there.
Eight: vision.
Not all companies share their vision with their people, or
when they do, it's not much of a vision, and it's a vision
that doesn't stand the test of time.
How many people worked for a company that
had a mission statement?
How many-- well, raise your hands.
OK, leave your hands up.
OK.
Leave your hands up now only if you
remember the mission statement.
Cynthia.
AUDIENCE: CMC's mission.
JONATHAN ROSENBERG: Outstanding.
Good job, Pamela.
They can all recite your mission by heart.
OK, that's good.
I like that.
Well, like I say, Google is more like a
university than a company.
So we've got this mission to organize the world's
information and make it universally
accessible to people.
It's going to take the rest of my lifetime and the rest of
yours and the rest of your children's children to achieve
this mission.
Everybody understands it.
Everybody internalizes it.
And it means something to them.
20% time.
No company has ever officially defined 20% time.
Our engineers work on anything they want 20% of their time,
and out of this 20 time comes pretty fabulous things, like
Google News, which was basically a guy by the name of
Krishna who was following the 9/11 efforts and wanted to
pull information on what was happening
together in a dashboard.
And he basically invented Google News which today is one
of the most visited news sites.
Thinking big.
The antibodies in all companies try
to reject big thinking.
Larry and Sergey do the opposite.
You come in with some little idea and they leverage it up
to a giant idea.
They even institutionalized it in what we
call the OKR process.
Objectives and Key Results.
One of the VCs told them when they first got started,
everybody needs objectives and key results.
So the company did what normal companies do.
You learn from McKinsey: under-promise, over-deliver to
your boss, right?
So everybody wrote these pathetically wimpy little
objectives and key results.
And Larry and Sergey went in to mock them and
said, what's this?
This isn't very difficult.
You're solving this problem for the Bay Area.
Solve it for the known universe.
It's pathetic.
So they created a process where you're only expected to
achieve a mark of about 60% on your
objectives and key results.
Other companies do the opposite.
In the middle of the quarter they all get together and they
say, what are all the commitments
that people have made?
Beat harder if they're not making them, or give more
resources to the ones that are successful.
We've done the opposite.
We've defined very big goals, and in many cases people don't
come close to achieving them, and we're rewarding the people
who do reasonably well.
Sergey, I recall like it was yesterday, when we only had
five-- there were only 500 million web searches a day,
like, several years ago.
And Sergey sort of looked around and had this look in
his eyes at this offset and said, you were all
thinking too small.
There are 10 billion web page views a day, and only 500
million searches.
I wish an ad on every page.
And then we had ad sets, which is basically the reverse of
search, putting the ads on the web pages.
The X Prize.
Somebody sending a rover to the moon, driving around,
sending back data.
I was just in a meeting on energy.
And somebody said, we can make renewable energy cheaper.
And Larry said, well, then, put an OKR up to make
renewable energy cheaper all over the world,
cheaper than coal.
That's what really will make a difference.
Think big.
Bet on a trend or fall victim to one.
Everything about success and technology is internalizing
and understanding what Eric calls the
technology base case.
You guys read about this, but you don't internalize it.
Moore's law, right?
Processing doubling every 18 months.
Improving by an order of magnitude ten
times in five years.
Kryder's Law of Storage.
Storage becoming half the price in a year.
It's hard in normal industries to think this way.
To think that the cell phone that in your pocket is a
hundred times more powerful than the PC that you
used ten years ago.
Or, how many do you have a Wii?
It's more powerful than the computers that powered the
Apollo missions.
That's very hard for people to think about when they're doing
development.
Larry and Sergey internalized that before they got started.
They built a system that they knew couldn't work at the
scale they needed it to work with the processing
power of the day.
All of you are using Gmail.
When we launched Gmail, we couldn't afford to give people
two gigabytes of storage, but we knew it would take time for
them to use up all of their storage.
And by then it will be cheaper.
It'd be a year later.
It would be half the price.
Again, betting on a trend.
Chad Hurley, who was one of the founders of YouTube.
You look at what he did today and it
doesn't seem so brilliant.
But you talk to him about what he saw and he said, you know,
there was a moment four years ago when I was sitting down
and I was thinking and I saw that it was hard to create
video, because we didn't have ubiquitous
deployment of cameras.
And bandwidth was scarce and it wasn't fast enough to
download these things.
And you needed special client software because we didn't
have the open standards and robust browsers
that we have today.
But he said, in a year or so, all of that's going to change.
And in a year he created $1.6 billion worth of value.
There are many other trends like this that
we're tracking today.
Cell phones, obviously.
we're getting over the form factor.
We're getting over the display issues.
We're getting over the battery life.
We're cramming all sorts of processing
power into these things.
We're adding GPS location enabled ability into them.
We're not far away from being able to walk around in stores,
take a picture of a bar code with your cell phone, and have
your cell phone tell you how far you have to drive or walk
to get the same device in inventory
somewhere else cheaper.
That's going to fundamentally change commerce.
People with devices in their hands ready to consummate
transactions who have knowledge, data, information,
about where they can get a better deal.
Accept the smaller piece of a larger pie, rather than
hogging a bigger pie.
We have Matt here, who's a writer.
I'm sorry, let's talk about your industry.
AUDIENCE: Please.
JONATHAN ROSENBERG: The Hollywood writers' strike.
They wanted a raise.
They wanted a higher cut of DVD sales, right?
I don't remember what the number was.
It was like four cents to eight cents, right?
So we held up the world for the future of DVD royalties.
How many of you students watch all your network shows online?
I know my niece does.
Put your hands up, come on.
Do you want a DVD?
If someone gave one to you?
You don't have a DVD.
Many of you don't have DVD players.
So we're arguing about a larger piece of a smaller pie
that's turning into a crumb.
Google's the opposite.
We're focusing on making the partner successful.
When we first developed AdSense, which monetized
websites, the business people with MBAs all said to Larry
and Sergey, we're the only game in town.
Offer them 40% percent of the revenue.
And Larry and Sergey said, no, no, no.
It's their revenue.
It's their pages.
They're the ones creating the content.
Give them the vast majority of the revenue.
We'll just take a tiny piece.
What did we end up with?
A small piece of a much, much larger pie.
I'm glad to fixed things in Hollywood, Matt.
You're working again.
OK, so, given that Matt's eating, again.
The next concept: feed the winners, starve the losers.
OK, this is the opposite of the way most companies work.
Most companies start out with people with these objectives.
Think about other portals.
All these general managers think about their own area,
not the greater good.
So the home page of the portal is where you get distribution.
The guy running the career site, the guy running the
finance site, the guy running the mail system,
they all want traffic.
So the one who's doing the poorest relative to his narrow
objective and key result comes and begs the person who owns
the distribution, my career site is doing really poorly.
It's not getting the traffic I need.
Push it on the home page so that more people use it.
Well, this is the opposite of what you should do.
Your career site sucks.
That's why--
I'm always picking on my right, not on my left.
Your career site sucks.
No one's visiting it.
Your finance site rocks.
We're getting all the traffic there.
Why don't I steer the users to the finance site, which is
good, instead of the career site, which sucks?
That's the way most general manager
oriented companies run.
That's why we are functionally oriented and will always be
functionally oriented.
You feed the winners and starve the losers.
Avoid hippos.
I went to Africa recently, and I learned about hippos.
Latin: river horse.
Hippo kills more people than any other animal in the world.
Hippos kill more projects in organizations
than any other person.
A hippo is the highest paid person's opinion.
I am a hippo.
When people in a room start talking, eventually the hippo
speaks, and says, I think.
I don't like it when people say, I think.
Somebody said they wanted a Sun story.
Was it Cynthia?
Sun.
Eric used to work at Sun and he tells a story about coming
in at Christmas in the late, mid-'90s.
He had a project he wanted to do and to accomplish it, he
needed one of the workstations.
So he went to the shipping dock and stole, borrowed,
whatever you want to call it, his PR people,
I think, say borrowed.
Anyway, he took one of the boxes off the dock and took it
to his office and unpacked it.
There were eight Read Me First documents in the box.
He put them all up on the wall and he looked at them, and he
realized that was the company's board chart.
Eight hippos insisted on a Read Me First
document in the box.
You could write an entire senior thesis on this
particular sentence.
Eight Read Me First documents.
Rule of thumb: if you can see the org chart in a product,
don't buy it.
That's the difference between the iPhone, maniacally driven,
accurately and correctly by a man of vision, Steve Jobs, and
these other phones that have multiple different
applications on them that don't work together, that all
vector into six different address books.
Avoid hippos.
Who did I say I would pick on today?
What kind of people?

Our auditors.
Yeah, well, that's the whole security thing.
Let's pick on lawyers and bureaucrats.
OK, I do like some lawyers.
I like pragmatic forward-thinking lawyers that
think through the issues relative
to a business objective.
The kind of people-- who comes from a place where they have
toll roads?
Anybody from like--
what's the process?

You queue, you wait, you burn gas, you throw money in, then
what happens when you come back?
The same thing.
Who was the brilliant guy that actually said, maybe we should
just have the toll roads in one direction, like on the
Golden Gate Bridge?
I'm sure many weenie bureaucrats
said that won't work.
How many of you have only paid that toll, and then driven all
the way around the Bay Area to get back to
Mill Valley, right?
We need more people like the guy that said, we only need
tolls in one direction.
Of course, we need more people paying attention to
[UNINTELLIGIBLE]
use taxes and fixing all these stupid taxes that we pay that
actually waste gas and that sort of thing.
We should have a big gas tax that would fix it, but that's
an entirely another lecture.
We can make it less regressive on the poor by fixing the
income tax rate.
And then people wouldn't drive such big cars and the world
would be a better place.
But that's a different lecture.
OK, more on lawyers.
How many of you have cars where you have to hit I Agree
every time you get in?
Or you have parents who have these things?
It's a Japanese car.
Oh, what is it.
AUDIENCE: [UNINTELLIGIBLE]
JONATHAN ROSENBERG: An American, yeah.
Well, my cars are all Japanese, and I always have to
hit I Agree.
In German cars, you don't have to hit I Agree.
Are the German lawyers smarter than the Japanese lawyers?
Or do they make a better business decision about the
relative risks?
Every time you download a feature online you have to
click I Agree.
This is just stupid.
This is just lawyers saying that you have to do this who
don't understand the online medium anymore.
If your kid's in the--
if you guys go to the hospital and you haven't signed a
release, do you know the hospital can't tell your
parents what's wrong with you because
it's a HIPAA violation?
All of you should tell your parents to sign this, by the
way, which I'm going to do for my kids.
That's crazy.
It violates the spirit of HIPAA.
You may be over 18, but if you're unconscious because you
got in an auto accident, all of you would want your parents
to know and vice versa.
You see this--
who came from one of the banks?
Somebody said they were visiting
from Capital or something.
Every time I get a message from you it says, this may
contain confidential, you know, boring whatever, blah
blah blah blah blah blah blah at the bottom.
This is just wasting bits.
At Starbucks finally a genius decided that you didn't have
to wait for the transaction to be approved before they let
you go stand and wait for your latte.
It just goes through as though it's going to be approved and
an alarm bell can go off ten seconds later while they're
serving the next customer.
I mean what did the lawyer think?
Pamela was going to go get her chocolate mocha and run out of
the store with a VISA card that didn't work?
Somebody go get her.
This is just-- this is ridiculous.
These weenie in the world, in linear programming language,
create a constraint set that becomes the null set.
And and then you end up optimizing around that.
We talked about that in the IT discussion this morning with
respect to auditors and on-site backup.
The auditors just have the wrong rule.
You don't need on-site backup here with a key
to the vault anymore.
You can have it in eight data centers all over the world.
You have earthquakes here.
You don't want it in a vault that you can't get to.

At Google, we never would have done Images, Books, Google
Search, or any of the things that we did if we didn't have
reasonable lawyers who explained to us the risks and
then let us make an informed business decision.
Never surrender to the lawyers, the accountants, or
the bureaucrats, except when they're right, especially when
they're keeping you out of jail, like Reg FD violations.
That's very bad.
I won't be talking about anything that
could send me to jail.
There are no broadband connections in jail.
I wouldn't do well without a broadband connection.
Rewarding innovation.
Again, seems simple.
Most companies don't do it.
They have profit sharing.
At HP.
I have to pick on somebody, I have to pick on HP.
The profit sharing goes between 5.5% and 6.5% every
year and everybody gets the same.
Well, that doesn't work.
That's like some other country's approach to the
world and rewarding things.
Pay the people who deliver lots of money.
Baseball players, shortstops, they make a lot of money.
The best guys in an investment bank, they
make a lot of money.
Matt, if his voice had been as good as the best guys in
Hollywood, would have made more money, wouldn't be a
writer now.
But now he's a producer.
You pay the best people way disproportionately more.
Millions of dollars.
The guy that built Google News has made us millions of
dollars, billions of dollars.
Pay him millions of dollars.
And when some weenie says that isn't fair.
Say you don't care.
Life's not fair.

I wish I could hit a step back 23-foot jump
shot like Baron Davis.
I can't.
So I don't get to play in the NBA.
I wasn't as handsome as Matt was in college.
I didn't get as many dates.
It wasn't fair.
AUDIENCE: He's right.
JONATHAN ROSENBERG: It's still not fair.

There's a final ingredient in this.
And it's one that I haven't talked about.
And it's one that all of you, I think, have going for you.
And it's about learning.
It's why you go to a great liberal arts school.
Because underlying all the things that I talked about is
this one idea: it's learning how to learn.
I learned a lot of Claremont.
I can no longer articulate the cultural underpinnings of
Japan's economic miracle, as Leon Hollerman taught me.
I cannot articulate all the theories of personality that I
learned in Professor Snortum's class.
I might be able to say a few things about the theory of the
firm, matching wits with Jerry, if I was lucky.
But what I learned here was, I learned how to learn.
On graduation day, 25 years ago, I'm quite confident that
I knew it all.
It's amazing how much I've learned since then.
Thank you for letting me share some of it with you.