http://www.computerhistory.org/events/lectures/schmidt_12062001/

Unwinnable Wars: Personal Perspectives on Technology Leadership
Eric Schmidt
Thursday, December 6 at 6:30 pm
Xerox PARC, Pake Auditorium
3333 Coyote Hill Road, Palo Alto, CA

Video online at
http://video.google.com/videoplay?docid=-5992089712538137005
http://www.youtube.com/watch?v=KFgOGgHS0Gg

Transcript at
http://archive.computerhistory.org/resources/text/CHM_Lecture_transcripts/Schmidt%20Transcript.doc

Transcript converted to ASCII, with a few typos fixed:

Toole: While we’re getting settled, I’m John Toole.  I’m the
Executive Director and CEO of the Computer History Museum. And I think
they’re gonna cut off – cut over my set of slides back there.  Hello.
__________.  Can I have my slides switched please?  Pretty please.
While we’re waiting for the technical difficulty, let me introduce a
couple of people from my organization that you probably know, some of you
may not know.  Karen Mathews, Executive Vice President of the Computer
History Museum.  David Miller out in back, Vice President of Development.
And we also have, Mike are you here?  Mike Williams – Mike Williams
is our new head curator, who just came for the University of Calgary.
I’m very proud to announce to have him on board.  And we’re gonna give
you a little sneak preview of a press conference we’re having tomorrow.
I’m gonna talk a little bit just very briefly about a little run
through dealing with the background about the museum.  I’m gonna ask
Dave House, our Vice Chairman of the Board of Trustees to talk just
a moment about what the museum is all about from his perspective.
And then I’m gonna come back and introduce our wonderful speaker
for the evening, Eric Schmidt.  As most of you I hope know by now, we
had roots over 20 years ago in the Boston Computer Museum founded by
Gordon and Gwen Bell.  We’re an independent organization.  Our older
logo, we just put on a new logo and I’m wearing that on a shirt
by the way that my wife says is available for sale out in the back.
The reality is the logo represents a very important thing to us.
It’s a new image.  It’s a new entity, an identity that we’re
putting together.  And really what that means is that it represents the
artifax, the preservation.  It represents the people and communities
that are so important to make computer history alive and real.  And it
also means the things that we’re gonna do for the digital artifax,
zeros and the ones that you see.  And we’ve had a rich history and
a very strong board and a strong organization.  And there’s a lot of
real things that I’m gonna show you tonight.  Our collection extends
to artifax, films and videos, photographs, documentations, and software.
Unfortunately our space limitations today don’t prevent us really from
doing the things we want to do and we need to do to preserve everything.
But I’ll tell you a little bit about that in a moment.  We really have
a world-class collection.  We just recently took a whole trip around
the world and in Europe, visited London, Munich, and Potaba in Germany.
And it really concluded that our focus of what we are to preserve the
artifax, the stories in the information age on an international scale
is really a very unique opportunity.  And it really isn’t being done.
We’re losing things all the time, which makes it really important
for all of us.  We have a visible storage exhibit area on __________
field today.  It really consists of a lot of very interesting pieces
computing artifax that you probably will never see anywhere else in the
world side by side.  I can’t share with you the smell and feel of those
artifax but I certainly want to invite each and everyone of you to come
join us.  This is a view of as we pane through the Cray of what – we
have piece of _________ that was on loan, or piece of sage, which was
semi-automatic ground environment system.  The kitchen computer, a lot
of very interesting artifax with a lot of interesting stories that are
Dosens and McWilliams and his team can tell you all the technical details,
the social details, and the wonderful stories that go along with it.
Again, we don’t have all of the things that we want over the space.
And that’s part of what we are in the process of doing both with the
people and the energy associated with it.  Took pictures of volunteers,
we hope each and everyone of you will either be a volunteer, a donor, or
become very much engaged with us now and in the future.  And there’s all
kinds of very interesting things.  But the exciting part is really where
the Nassau Research Park, which is a plan of record of where we’re
gonna be located.  Now the big hanger of course is gonna be converted
into an air and space center as part of this process.  And you probably
hear in the area, a number of public hearings.  I encourage each of you
to attend those if you’d like and voice you opinion pro or against,
or your concerns.  It’s all talking about what this is gonna be.
And I think it’s a wonderful opportunity.  We’ve go three acres
of land right in front that we’re gonna build 120,000 square foot
building in a couple of phases.  There’s a whole section back here
that’s gonna be a Carl Sagan Center for Astro Biology and a university
park complex, University of Santa Cruz, San Jose State, and Carnegie
Melon University.  So it’s a very exciting opportunity.  Our future
sight as look kind of toward the hanger is really very interesting way
in which we can build a very exciting kind of a building.  This happens
to be one of about nine different versions that we went through in an
ideas competition that we selected EHDD.  We haven’t selected this
particular building by the way.  We are only in the cement early parts
of this schematic design program in process phases.  But basically, we
pick – chosen these people as EHDD and then ______________ Molary, who
built by experienced music project in Seattle as our exhibit designers,
as a real architecture and exhibit design team that’s working closely
with our board, closely with our staff to really build a world class
institution for the long term.  Expecting to open in 2003, excuse me, oh
– great, ground in 2003 and open in 2005.  This is sort of a synopsis
of the people that I just stated.  We’re an international museum and,
you know, someone thought the building might look really nice if we put a
telephone booth there.  This is actually one of our artifax that was just
recently donated from Switzerland.  And you may look a little fund kind
of an artifax, but inside of it is really the world’s first tele-guide
that was used as a directory assistance for – in the public domain.
So there’s little interesting stories that you’re gonna find all
the way through this.  We really can’t succeed on our own even with
the space that we have waiting for 2005 spread over four buildings.
We are in the process of putting up an interim building.  We’re gonna
call this the beta building and we’re gonna announce this tomorrow.
It’s about 40,000 that consists of warehouse space, 10,000 square feet
of administrative office space to consolidate our people and to build
the culture we need to service the community that we think is important.
And the rest of the space would be some exhibit space and assembly space.
So this will give, and this will be completed late summer early fall of
next year.  So we have a program.  We have a process, and we’re really
running fast and hard to really preserve those artifax and stories.
We had a tremendous lecture series.  As many of you know, many familiar
faces, I welcome you this evening, you’ll – we gonna introduce Eric
in a moment.  And we’ve got some wonderful upcoming 2002 lectures that
we’re putting together that will be announced.  This is sort of a sneak
preview of just a few of those people that we’ve already got set up and
many more yet to come.  I really want to challenge you all and ask you all
to come – come join us.  Spread the word.  Take a look at – we’re
spreading really our logo around.  It’s our new logo, our new identity.
Volunteers are particularly important, both from the point of view of
helping us and the donations that they actually can bring to us in time,
energy, enthusiasm to building the community that’s so important to
make this a piece and part of everyone of you.  And, become a supporter.
And toward that end, we’ve – we’ve given you our new brochure of
– that each of you have.  This has a little snippet of some of the
things that talks about the timeline.  And I hope you will enjoy it.
Some of this is in the mail to some of you that some of our supporters.
I’ll now like to ask Dave House, our Vice President of the Board of
Trustees and CEO of Allegra Networks to come and tell us your perspective
of the museum.  Thanks Dave.

House: Thanks John.  You’re all part of a very important industry.
You’re here tonight because you are part of that industry.  You’ve
got an interest in that industry.  It’s an industry that’s changed
all mankind.  There’s two periods a time that history will record.
There’s the time before we had computers, and there’s the time after
we had computers.  And we – many of us had been alive in both of those
periods and part of creating that change that’s happened, in not just
industry, but in civilization today.  Today the information technology
industry, the information industry, computers and networking are the
largest industry in the world today.  It’s bigger than automotive.
It’s bigger than Petro Chemicals.  It’s bigger than any other
industry.  It generates more engineers and employs more engineers than
any other industry in the world.  Its economic importance is enormous.
Yet is an industry that is very, very brand new.  It’s happened in the
last 60 years.  Our charter is to preserve the artifax and the stories
about that industry for posterity.  And we’re a volunteer organization.
We’re a budding organization, an organization with an excellent staff,
John Leading and he’s introduced a few of the people on the staff,
a growing staff of individuals.  And a growing set of facilities, using
some buildings that we’re currently – four different buildings we’re
operating in on the Morfet campus.  Starting construction very soon on
our first purpose-built building, that is one that we’re building
ourselves, which will eventually become warehouse storage, will be
combination of office storage assembly for meetings and lectures like
this etcetera.  Now, this organization is not a government organization.
We – there’s no government going into this.  We don’t get money from
the federal or the state or any agencies.  You got a brochure that’s on
your – was on your chair.  Tells you about the Computer History Museum.
And in the back is some important – strictly important information.
One is, there’s a list with really a small mice-type.  These are
the people who have made this possible.  These are the donors that
have contributed to our campaign last year.  We had very little money
from industries almost entirely from individuals.  And also in here
is a pledge form, and an envelope in which you can mail that form.
What I’d like you to do is take a look at that and see if you can
be part of helping this organization.  It’s December, it’s the end
of the year.  It’s time to make your annual donations.  It’s time
to get them in on your taxes for 2001.  So what I’d like you to do
before the end of this year and maybe before the end of this evening,
take a look here and see if you’re able to contribute.  We appreciate
contributions at any level.  You notice we’re a little bit nerdy.
And so those of us in the hardware side realize that one K is 1024.
So we make our lowest quarter donation be 1024, you know, 4096, 16383,
which is actual 16k etcetera.  We’d like you to take a look at your
own situation and reach into your contributions.  And make your donation
towards preserving the artifax and the stories of the information age;
the industry that clearly is very important to you, or you wouldn’t
be here tonight.  So take this seriously.  We only get money from people
like – you – people in this room.  It’s completely an organization
financed from individual donations.  So please do what you can for us.
Thank you.  (Applause).

Toole: Thank you Dave.  And thank you all.  We wanna change gears of
course now because get into what we’re really here for the night
most of you came.  I think in general, you know, we have a CEO’s,
CEO this evening, Eric Schmidt, gonna talk to us.  And has really
done a lot of work.  It’s a very special evening as a result because
leadership, its economics innovation and how it all fits together, and
the experiences are something that’s very, very important to try to
understand, to learn about, and to apply for now and the future.  And so
Eric we’re very happy to have you this evening for a number of things.
Chairman and CEO of Google, certainly my home page that I personally
use for my browser I’ll have to admit, certainly the former CEO,
Chairman of Novell, and the 20 year history of executive experience,
we’re really looking forward to.  Eric Schmidt.  Thank you. (Applause)

Schmidt: Thanks.  It’s a great honor for to be here.  And I mean
that because I was in this audience when this building was opened.
And I remember sitting right there.  And I want to encourage you to
participate.  I think that what the museum is trying to do is very,
very important.  And I think what I want to address, for the purposes
of tonight is what can we learn, and what mistakes are we making now?
And I think ultimately the study of history is really about what can we
now do base on what we now understand.  So I sort of started thinking,
well, how many natural laws are there.  There’s __________ law, you
all heard these, right.  There’s the law of untended consequences,
right, that’s what governments do.  There’s Mores Law, you know that
Met cap’s Law, the value of networking pieces N squared in number of -
_________ devices on it.  There is Joy’s Law, ______ equal two to the
year minus 1984, and the original – remember that?  The definition of
______ is actually true even today.  Amazing, right.  There’s the law
of evolution.  There’s the law of protocols.  Every successful protocol
gives you a 100 – $1 billion in market value, right.  You’ve heard
that one.  And it was true until the crash.  There’s Gate’s law, you
can never be too rich.  There’s Socialite’s law, you can never be
too rich and too thin.  There is Allah Hate law; I’m in charge here.
I’m dating – it’s a history talk.  You know, you have to – the
fun one, right.  There’s Griseum law – Griseum’s law, you all
know that law.  That’s the law why chat room are not places where
normal people inhabit anymore, right.  You know what Griseum’s law.
There’s the law that CEO’s jobs are getting shorter and shorter,
right.  Some of my board members are here so they’ll listen to that one.
There’s the law that the Internet always wins, which I actually think
is true.  And then there’s a new law, which I’m gonna propose for
business tonight, which is that a lot of laws are really opportunities
that have since passed, right, and that the trick in our industry is to
do something really new and innovate all this and put the stuff together.
And I believe profoundly that there’s a set of real rules that govern
what we do.  And I think that in our – in our job of understand what’s
happened and planning the future, there are a number that really matter.
And in processor speed, network speed, disc capacity, all of those things,
_________ are all doubling roughly every 12-18 months.  And that’s
true whether we’re making money at it or not.  Network effects are
the next vendor walk-ins.  And we’re gonna talk about that.  There is
always some proprietary advantage in an open source strategy.  You just
have to figure it out.  But I think the real – the really true laws are
really two.  I think the first is that each and every generation makes the
same mistakes, right.  The older folks in the audience are saying, yes.
We’re gonna hammer this one, let me tell you.  And then there’s of
course what I think George ________ real law is, which is that the only
real scarcity is your own life, right.  And that’s what we really are
here – here to talk about.  Now what I did is I brought lots of artifax
but I scanned them, you know, and we’re sort of with it so.  This is
the license plate that we all – that we all had in roughly 1977, 1978.
And this is the way Unix felt.  For those of you who were in the Linux
community, it’s the same feeling, and we would – and by the way we
would like to be as young as are now.  You know, it’s the same –
it’s the same thing.  We’re only 20 years older, right.  I can see
my friends in the audience who were with me at the time.  It’s the same
thing, right?  And, you know, the early Unix history was – it was as you
know, and we’ll just go through this quickly is developed __________.
The first part was done ___________ on the inner data.  And the inner
data first board that __________ Bell labs has done to an IBM machine by
a set of students at Princeton University in ’75.  Bell labs had this
interesting things that they stumbling into without licensing cause they
were not profit.  If you are a university you could use Unix.  And if
you are – if you’re not a university you had to talk to this group,
which ultimately became Unix Systems Labs.  And of course, Unix at the
time was time sharing word processing assisted of NROF and TROF, right.
And I remember those commands.  And of course at the same time, and
I wanna bring you to sort of my first discussion point, the Internet
routing was invented through Paul Bearing and his work in ’64, Venton
Bob invented the concept of TCPIP in roughly 1974.  A number of sites
were done, which involved imps and specialized hardware.  There was a
computer here in this – here in the building next door, which is one of
the first – first thing connected to the Internet called Max.  BSD at
the same time, when we – when I was part of this had – and this is the
_________ distribution, had a licensing model, which was an evolution of
all of this.  Where all of a sudden if you’re a university, and you call
us up and you send us $50, we would send you a tape.  So our strategy,
our licensing strategy was we would send you a tape with a source.
And it was a Vax Mag tape.  And of course we would order it to the 780.
And the BSD – the BSD series, there was, you know, BSD1 and 2 and 3 and
4 and 4.1, and 4.1C, and 4.2 and 4.3 and 4.4.  And it sort of ended in
roughly 1993.  And Heyday was in this early period 1977 and 1970, 1979.
What happened was because we were shipping these tapes out, people were
getting all this source and they were using it to build the Internet.
And they would run these Vexes in a later and they would use it using
TCPIP.  And the first really production of limitation of TCPIP was done on
VST units.  So we – we were full of ourselves.  We thought this would
be just really neat.  We would change the world as a result, right.
Of course we had the wrong hardware and the wrong operating system.
And we didn’t exactly understand monopoly and all that.  We’ll
come to that.  Now AT&T at the same had a group that it formed called
System 5.  And System 5 was to productize all of this.  And the problem
was that the people who did Unix didn’t wanna go work in this group.
And so one of the first things to know is you gotta get your people,
you policies, and your customers in alignment.  And if you don’t, you
pay for.  We paid for it for maybe 15 years.  SUN at the same time was
founded in 1982.  And it’s interested that there was another variant.
And a number of people in the room participated in this.  We’re SUN,
which stands for Stanford University Network was a hardware board.
It was license by Stanford.  They’re very liberal to six companies.
Anyone can name the other five besides SUN?  No because Andy _________
who designed the boards and is the only guy who could debug them went
to SUN.  So again, what do we learn here?  We learn that the technology
had to work and you have to have the people doing it, another sort of
key early insight.  And at the same time Xerox again, literally in the
building attached to this one, is busy building, inventing the Alto, the
laser printer and all these incredible things that changed our world.
The PCs were not network.  The first PC that I – that I ever saw was
in the building next door again when I worked here.  And it cost $3,000.
It came with 64 kilobytes of RAM, a floppy, and a monochrome display.
We of course thought it was useless.  And we were right cause we were
busy running on Altos and Unix machines.  And in fact, what happened was,
I gave a talk, right.  And this is the first slide of my talk about Unix
at the time.  I did this in 1982.  This was – this diagram was done
with SIL, which is a predecessor, you remember it.  And this was the talk
that I gave about the evolution of Unix.  And you can see that even then
Unix was getting itself into trouble, right.  And this is before all
the commercialation etcetera.  So the reason I thank you for this is
I think __________, the first really interesting open source scenario.
Microsoft had delivered DOS.  System 5 had just begun to inform.  SUN Was
shipping Workstations based on BSD.  The ________ was beginning to use
__________ and BSD variants.  Apollo was shipping this Workstation and
the first really interesting choice happened.  And Bill Joy was the
first founder of SUN.  And he occupied at the time the – he’s a
close friend so I always have a buyer.  See I came out of a position
very similar to what Linus ________ has today in the industry, really
was very iconic.  And I remember in 1983 going to a conference with
him where all of these people wanted his autograph and wanted to talk
to him.  And I had first glimpse geek fame.  And this is very heavy.
And my job was to make sure the adoring crowds would not assault him so
he could continue to code.  And of course we’ve since – we’ve since
multiplied that many, many times.  But it works.  And it works because
the stuff really matters.  The first real attempt we did with licensing
was something called Network File System Manifest.  And it succeeded.
And let me tell you – let me tell you the early licensees in 1984.
Let me just read them to you.  Alliance, Pyramid, Gool, Mount Zino,
Lockman, CSCE, whoever they are, Eakins, GCA Tropel, Sequence, Solarity,
Data Generals, Cadnetics, The Instructions Net, Unisofts, Sparticus,
Deck, Seamans, Convex, SGI, ISI, Cronin, Willingong, MIPS, Bull, CCI,
Alexi, White Chapel, Encore, ATT, Toshiba, and Zilog, and ________.
How many of those companies like I don’t even remember what they did let
alone are they not around.  But what happened was that we put together
a program where we had the source code, we gave it – we followed the
BSD model.  And the particular technology called NFS was stateless.
And what happened was that you could implement it in such a way that
it had very easy fail over.  And all the other technologies that people
were using were very sophisticated distributing computing platforms that
probably worked under some assumptions but were impossible to implement.
So the first thing that we learned in this was that you can build
something really simple and it worked at its scale and it was portable.
And we can talk in the Q & A about some of the technical details for
that.  You could pull it off.  It was very simple.  It was easy to add.
It was compatible.  You didn’t even change your APIs.  It slid under
existing file system.  It wasn’t a distributed operating system.
And the strategy we adopted was novel at the time.  And as a result,
we’re building something so useful and making free to people, and giving
it to your competitors was hugely controversial.  Now those of you who
spent years and what happen subsequently don’t remember what a new idea
this was at the time.  It wasn’t mine.  I was simply implementing it.
But the fact of the matter is it really was a new strategy.  And it
worked brilliantly.  And all of a sudden we had something to talk about,
cause who wants to talk about the products you built.  You wanna talk
about the strategy.  The press have to get excited, all three of them,
you know, at the time.  And who cares about Unix at the time, right.
We’re busy doing our thing.  You know, there’s clients over computing.
They’re all having a good time and so fourth and so on.  The competitors
Apollo domain in particular did not have such a strategy.  And they had
more complicated technology.  And by the way their product was better.
On almost any comparison, it was better.  But we won.  And we won because
we follow simplicity.  And by the way this is 1984.  The exact same recipe
worked each – in each of every five-year increment sense.  So why do
we just keep doing that?  Why is it that the industry always does the
other maneuver?  And I still don’t know that, but I think it’s one
of the questions for all of us.  And what’s interesting as a result
is that the – it made a name for SUN in a very period as an innovator
in technology and business practices.  And SUN became embolden with the
_________ strategy.  This is also known as arrogance, so, although I
was trying to be polite in my notes.  So it seems to me that the first
lesson that you will learn from this is that there is an inevitability
principal.  But what you wanna do is the day you start; you wanna create
the inevitable success of your product to a series of strategic decision
licensing and adoption.  And at the time, SUN was a very small company.
The Unix industry was sort of, nobody even understood it let alone have
that kind of press coverage that we had subsequently until it worked.
And also another law I would suggest is, it pays to first with any motto.
And of course this is – this is another picture of Bill.  I’m gonna
have a bunch of magazine pictures, Fortune, Upside, and so fourth to try
to illustrate this.  And Bill has had an __________ credited with having
written a code that become initial definition of how the Internet would
actually work.  At the same time I’ll give you some asides.  This is
the first mention that I can find in my stack of history around Microsoft.
And look at the title.  Now would you be surprise if I told you that this
is dated 1984?  1984.  By the way they were running DOS, right, which
wasn’t any better then than it is now. (Audience Laughs) And Bill is
28 years old.  And if you read this it’s like – it’s a game plan
for the next 17 years.  So what do you learn?  They’re executing the
same plan they’ve had since then, and we should’ve read the article
at the time.  I filed it away. (Audience Laughs) Okay.  Seriously.
So anyway as a result we decided we were open and with open systems and
that was all – that was always the right answer.  And so what we did,
SUN for various reasons had this technology called SUN View was a – not
a network based model.  It’s a kernel based graphics model that’s
relatively straightforward.  It work extremely well.  People like it.
It only ran on ______.  And we had a problem of what to do.  Here we had
all these people building on top of it.  Apollo had something called DSEE,
which was a similar technology at the time.  So, we said, “Ah, perfect.
We are brilliant.  What we’re gonna do is take this innerfest model.
We’re gonna completely clone it and we’re gonna do it with this
technology called NEWS (Network Extensible Window System).”  Everybody
goes, yeah.  And the problem with NEWS was well there’s like I have a
whole page.  Well the first problem is it didn’t work.  I hate – I
wanted to say that for 15 years.  I couldn’t say it before. (Audience
Laughs) And by the way, it was a great product in a different year.
The performance – the performance wasn’t there.  The scalability
wasn’t there.  It was a brilliant idea.  It tried to combine three
different things; server based window systems, a postscript-imaging
model and programmability and extensibility.  Anyone of those is a major
challenge.  And you’ll see later that X who won only did one of those.
We had all sorts of problems.  We decided to clone the Adobe because we
didn’t like their business terms.  So what we now was a cloning problem.
I’ll come back to that.  We didn’t give our customers very much help.
The technical advantages, people would look at this and go, like, well,
this is like, I’m not sure what it does, but it’s quite interesting.
And then they would go back to whatever they were doing before.  And then,
and by the way that was not the thing that killed it because as we
know non working software can be fixed over time.  Well you can work on
it, right.  The real problem was the strategy didn’t work because now
we had a bunch of competitors who didn’t wanna give us a second win.
So you get on shot.  And we got our shot.  And we executed very well by
the way.  The second shot we’re not gonna get.  And in fact, what was
very interesting is the competitor, which is called X was a technology
developed at MIT.  And X was actually quite a good technology, which is
funded by ________.  Now this is – this is the NFS strategy being done
to us.  And of course we were too dull to figure this out at the time.
So we breast, you know, we beat our breast a lot and talked about it
and so fourth and so on.  And meanwhile everyone’s talking about the
McIntosh, which at the time, you know, was very strong and very exciting.
And we tried to get a deal done with Apple in this regard, but we
couldn’t pull it off for various complicated reasons.  So you had
venomous competition form HB Deck and IBM who were absolutely convinced
that they would do whatever it did to stop us.  Plus we had the problem
that the technology did not in fact solve the problem that everyone had.
Well this is not a recipe for success.  And furthermore, we ended up
in this weird problem of mechanism versus policy, because everyone
__________ produce mechanisms for things but won’t produce policy
for this.  This by the way was a very bad idea, and I was one of its
biggest proponents.  And the reason is that the person using the computer
wants a policy.  They could give a crap about your mechanism, right,
for those of you in engineering.  The mechanism is very interesting.
It’s nice to give talks.  People wants something that works, right.
So we gave general mechanisms both in X and NEWS, but in fact people
didn’t really wanna do it.  So, what does SUN do?  Well, again another
one of our brilliant ideas.  We decided to merge X and NEWS, right.
Those of you who used it have the correct reaction.  Now this took –
this took two years, because the architecture although they – the slide
ware was very good.  Right, they were slide ware compatible, right, right.
That’s how we do marketing.  When you actually merge them together, you
get – you get something that doesn’t work.  Now meanwhile, the team
that built this played an April fool’s joke on me and sold me a car
without my permission and then put it into my office.  And the joke we
had at the time was this is the only project that they had ever delivered
on time. (Audience Laughs)  So here is me in my care inside my office.
And it’s really hard to get rid of a Volkswagen from your office.
Here’s a – we had a lot of buttons in this period.  This is another
one, and I reproduced some of them for your enjoyment.  And make sure you
save the buttons so that when you get a chance to give a talk like this;
you can embarrass all the people who came up with these ideas.  The best
part about NEWS was we got to do our first deal with Microsoft.  As is
typical of Microsoft, they have untended consequences.  This one has the
interesting unattended consequence that it has a quote from Scott.  Quote,
“SUN like our systems and Microsoft are leaders in the development in
of an open systems approach to graphics in their respective markets.”
He denies the quote.  (Audience Laughs) And one of the things to know is
that Microsoft signed the contract, took the code, inspected it, learned
everything they could from it, and never used it again.  Interesting.
Now, so, what do you learn from this?  Somebody else can play the game.
Lets keep going, right.  So that – so we – we don’t get enough,
right.  We’ve gotta do it again.  So now we’ve got the problem
where you’ve got this System 5 release, System 5 folks and you got
the BSD folks, and everyone would be served better by one – by Unix
unification.  And we had done a deal about converging the two where
what we agreed to do is to take the interface definitions from System
5 and add them to BSD.  This is the deal that I did and we were gonna
– SUN would continue to work on network, ____________, __________,
____________, NDS, Yellow pages, Rex, Locking, and PCMFS.  And we plan
to add things like pipes and some of the other things people actually
wanted.  This strategy worked brilliantly because no one actually use
the library.  So the great thing about some of these deals is if no one
actually needs the deal and which you ____________ the marketing effect,
then you’re fine.  But if you actually try to do the merge, right,
it takes years out of your life.  So we then pushing on our open systems
strategy, which of course we called, you know, open systems for open mind.
And our joke was open system for open wallets, was to establish sort of
a new religion that there would be these interoperable interfaces that
everyone would base on and then these things would scale and so forth
and so on.  And we were consumed by this, and at the same time one of the
early inventors here at Xerox went to Microsoft and ___________ Windows
won that out, delivered in 1985.  It wasn’t very successful cause the
hardware platform was so weak.  But it set in motion the other path.
And I’m gonna keep bring you back to these two paths cause I think
that they’re illustrative as to what really happened.  So anyway as a
result through a long series of things in 1987, we announced the big deal
with AT&T.  And the big deal with AT&T was to create the Unix standard.
We announced this as a series of phases.  The first phase was this
interoperability phase.  The second phase was the merger, and the
third was a research project.  The research project never happened.
And the first interphase did and the second merger did indeed occur.
And that’s what’s now called System 5 Release 4.  I don’t know
what was going through our mind, but it never occurred to us that our
competitors having been warmed up by the success in NEWS would see this
as annihilation.  And in fact, the – if you go back to the speeches and
the positioning that we took at the time, maybe we were baiting them.
Maybe it was one of those unconscious things you do as humans that you
really shouldn’t be doing.  I don’t understand why we did that and I
played my role in it.  But the important thing is that when we announced
in October of 1987 that we were gonna unify Unix, and it was gonna be AT&T
and SUN along with Motorola, and we gonna establish a binary interface
standard for both hardware and software; the world blew up, right.  And it
blew up big time.  So the idea here was we would partner with AT&T, we
would work with these groups, and we would – we would literally have a
combined problem.  The problem was that this was change in our strategy,
because up until then, all of the Unix technology that I worked on, had
been shared through this variant of open source licensing.  Now the way
it used to work with the BSD is, you know, you’d have somebody like
Van Jacobson as a brilliant Berkeley person, we just add TCPIP code and
we would sort of add it in.  But because of the merger and because of
the commercialization, that wouldn’t happen anymore.  So eventually
AT&T and SUN formed a group called Unix International.  And I’ll come
back to this in a bit.  And of course they were initially called Archer
Group.  OSF then formed, and of course you know what OSF stands for,
Oppose SUN Forever, Open Soccer, Open Systems, Open Soccer Foundation.
Formed in 1988, January 1988 in 100 Hamilton, which coincidentally six
years later was the site of all the JAVA work.  So it shows you what
goes around comes around.  Now what’s interesting about this, and
I don’t wanna dwell on the product side is that it eventually won,
and it won because of the sheer weight, power, money, you know, hot
air, you name it.  It took a lot and it eventually won.  It was many
years late. And some of the implication of this was quite interesting.
A lot of the key technical people who were working in my part of Unix
in the world actually left because of it, cause they didn’t like what
they were doing.  It actually caused a transfer of technology people out
because working on merger is with technology and for reasons that are good
strategically but not very good technically.  It’s not a very good way
to maintain you top technical people.  And that’s an important lesson
that I learned the hard way.  And I think we all did.  The deal itself
was a great source of funding for SUN.  AT&T had invested $250 million to
make this all happen.  In ________ we should’ve just taken the money
and gone and taken a Mac user interface and put it on the top of BSD,
and change the world.  But that wasn’t part of the deal.  So, you know,
I started thinking, how do I characterize this, and I think the rule is,
you know when elephants make love, you should run out of the forest.
But I think that we’re beginning to see these ideas again.  People
started throwing these ideas around for various reasons and they just
don’t work.  So there was this continually refrain for unification of
Unix, 1989.  Unix popularity hinges on the emergence of a single version
of the operating system, which is designed to allow many users to tackle
jobs on one computer system. Currently because computers use their own
version of Unix, it’s difficult to make a single Unix applications
run on a variety of platforms, right.  And that was repeated many, many
thousands of time.  And if you go to Google and you look it up, you’ll
see them all.  It’s quite distressing.  At the time there was this group
called X-Open.  They decided that it would – it _________ in this cause
they were a standardization body.  They tried to define these interfaces
but of course they were always late to the fire.  Some was busy pushing
something called an ABI for Spark, Applications Binder Interface.  At the
same time, a derivative of OSF called the, I forgot their name, formed by
Compaq Deck and SGI, were forms to standardize on the MIPS architecture.
SUN then pushed something called Open Network Computing, which included
NFS.  I’ll do this quickly cause you’ll see it gets very boring.
OSF developed something called DCE, which is Distributed Computing
Environment.  I had to look these things up cause I had forgotten.
But they use their own RPC mechanism, which is incompatible than Unix
International and developed something called Open Systems Architecture,
which did not include the technology from their funder, but included the
other guys, which no one used, right.  Now if you’re a customer, what
is your reaction to this?  And meanwhile here we are, okay, and we got
the other guys.  The other guys are bringing out Presentation Manager in
1988 and Windows 3.0 comes out in 1990 and really kicks off what we know
as the modern Windows era.  Microsoft cleverly gave OS22.0 to IBM and took
OS3.0, which they then renamed NT, which has now become the basis what we
know as Windows XP.  So, we get to play the game again.  So Scott…so,
I don’t know how to introduce this except that, it got to the point
where we were the subject of all media coverage, right, as this went on.
And the media was very important because the media drove the behavior
a lot of the participants.  Here’s a particularly fun cover of Scott.
And in fact there was a series of articles, which said that these were
turning points for the company.  And so what happens is, everybody gets
all excited.  And are these strategies working?  So as you can see,
and of course they’re equivalent and I don’t have as many of the
non-SUN coverage.  But there’s equivalent coverage of the other guys.
See, everybody is being pitted now, right, and the wars are playing.
So, I couldn’t figure out what this – what this button was from.
This is called Open Network Computing.  And I couldn’t tell if it
was by the Open Network Computing people or against the Open Network
Computing people.  And I know __________ was in network computing except
that I could tell you it didn’t work, right.  So, this went on and
on and on and on.  So meanwhile X Windows had managed to get itself
established.  And Open Look was another brilliant idea we had.  And this
was a joint design between AT&T and SUN, again license to everyone,
but no one wanted it.  They were busy building something called Motif.
And Motif was – because X did not have a policy it just had mechanism,
you needed a Window manager, Motif was it.  Now the brilliant thing about
Motif is that everyone managed to convince themselves that Motif was
going win and so all of a sudden now there was yet another war, right.
And that war was around who would use each of these layers.  And you get
the idea, right.  Each layer creates a new war.  And the way we solve
these out is we have this sort of combat, meanwhile, the other guys are
busy making progress.  My important point here is that the Open Look in
Motif problem then became a problem once we’d resolved the previous one.
So we finally got to Unix unification.  In 1992 we had – SUN was
running SVR4 in Open Look.  HP was running System 5 in Motif.
And Deck was running BSD in Deck Windows.  What happened, right?
So COSY was formed in 1992.  Anyone remember what COSY stood for?
Common Object Software Environment, right.  Anyone remember what CDE
stood for?  Common Desktop Environment.  We’re running out of three
letter acronyms to define what we do.  So what happened was, we had the
grand – we had the grand agreement of all, which was that X Window
System and the Motif Window Manager and the Desktop Manager for HP,
so we’re done, right.  How long does it take to build this thing?
Well then another couple years go by and you get the idea.  The OSF
and OSF1 were largely a technical failure for the people who _________.
OSF was the name of the competitive operating system, and it survives
in various forms.  The Open Group was long story, but AT&T sold Unix to
Novel in 1993 before I was there.  They also sold it before I was there.
SCO bought Unix from Novel in 1985.  And X Open took over the trademark
in 1995.  X Open then began to add additional, you know, Unix versions.
It was at UNIX ’98 and so forth and so on.  There was version of X11R6
released in 1998 and a splinter group formed, a Linux version.  So you go,
why would that happen after all the things?  It turns out that the terms
and rules about Motif did not allow the Linux community to use it cause
the licensing rules about Motif were consistent with the model that the X
Open and the corporate had, but not the free software world.  What is the
law you conclude from this?  Sometimes no one cares.  And that of all is
the hardest of all of these stories to relate is that your energy level
and your tone and your interest level just goes down as you read this,
right, because people had moved on.  So then what happened, was we said,
“Okay, we got this huge problem, cause all of a sudden Windows take off,
so we need something called WABI.”  And WABI did not stand for Window
Application Binary Interface, according to our attorneys.  And the problem
we had was that as you know, the desktop is largely ruled by PC’s and
Windows is taking off.  And Microsoft done an excellent job of porting
on and perhaps done a legal job of porting all those things all over.
And all of a sudden everyone is using Office.  And Office required a set
of interfaces that Windows offered that were not in our view standard.
So we built a clone.  And the clone of Windows was called WABI.
And because of the licensing issues, and of course Microsoft wouldn’t
license us the necessary intellectual property to making a perfect clone.
It didn’t quite look right.  Anybody here using WABI today?  There were
a few at the time, right.  So anyway, we had this great intro.  This was
our WABI mania – WABI mania slide.  And we had a huge thing.  We had
all of these, you know, powerful speakers about how important it was that
we have choices.  And that by doing this and making it available, people
would actually have choices on the platform.  Why did WABI not work?
And it wasn’t the name by the way.  It was easy to keep breaking it
if you’re Microsoft.  It was easy for yet another thing to happen.
Now I’m not suggesting that they did it deliberately, it may just be
part of their normal, you know, they change things.  But the fact of the
matter is that against a fast competitor, a cloning strategy doesn’t
work.  And it doesn’t work at all.  So if you wanna do this, you’re
welcome to do it.  But I’ve done mine, right.  And I had the best team
in the world.  At the time there was a series of scandals involving
unpublished interfaces into the internal Widows interfaces, which we
had all reverse engineered.  Even with all that, it didn’t work.
So meanwhile, the Internet is taking off.  And we sort of hit another
sort of interesting point here.  The – this is a copy of Internet Tidal
Wave Memo, which you may have seen.  This is now in the public domain
and it was part of the – one of the many recent lawsuits.  And this
is the – it’s interesting if you read it.  This is again Microsoft
speaking, it was written by an assistant of Bill Gates.  The Internet is
at the forefront of all this.  And developments in the Internet over the
next several years will set the course of our industry for a long time
to come.  This memo was leaked out of Microsoft by people who wished
them ill will, I guess in spring of 1995.  And what has happened was the
industry had sort of gotten itself into a roughly stable point, right.
We had the Windows thing going on.  Finally we’d settle all these wars.
And all of a sudden we’re off to build – to build the next thing.
When we built that software, we began to understand that networking
software was a little bit different.  One talk that I used to give, and I
like – where I liked some of the ideas was I resurrected with, entitled,
The Seven Deadly Sins of Networking.  And these are the mistakes that
people made at the time, and still I think make today.  You always assume
the network connection is up.  You always assume there is no latency on
the network.  You assume the bandwit is infinite. You assume the network
is secure.  The interconnect of the single network doesn’t change.
That’s your assumption.  That there is a single administration point in
the network and that network use is free, cause isn’t it?  And we were
busy, all of us, in the industry building, busy building these.  And the
Internet changed the way we thought about all of this.  And at the same
time Netscape was formed, and everyone – this is sort of relatively
recent history.  So folks in the room remember this.  And then the JAVA
phenomena began.  The way the JAVA phenomena – there’s a lot been
said about JAVA and the way it was originally developed.  And I won’t
retread that here.  Most of you have heard, myself rather, talk about
it from a simple technology idea to just the right timing of its use.
All of a sudden we had a phenomena.  What I want to talk about is not
the technology, which I have a lot of pride in.  But how did it get so
successful given we had so many bad examples before us, which is why I
took you through each of these.  And by the way, there are many hours
of details, all of which you’ll find very boring about all of that.
It worked for a bunch of reasons.  But one of them was this article.
This was the article that was published in March of 1995. I was in
Aspen at one of these technical retreats that we were doing at the time.
And I remember getting a phone call about this particular reporter and
then she get in and was actually talking to the JAVA people who were of
course busy talking to them, cause no one told them not to tell her.

Schmidt: It’s part of the world.  And it concluded a quote from a rising
young star, Mark Andresen.  So his endorsement seen as the messenger
of this sort of next wave of smart young people was as important as
what the article actually said.  Now he gets a whole bunch of things
wrong cause we – our strategy had not evolved at the time.  But here
is a case where since we had gone from the NFS period where we had
essentially no understanding of the press, we understood the press,
and we understood how the press could be helpful until we used it.
We didn’t abuse it now, we used it, right, and we used it to achieve
some significant gains.  The product itself of course was compelling
technology.  It had a clear utility, people who could understand it.
There was both a browser and of course the partnership that we did
with Netscape the JAVA main systems programmers in that sense out of
ordinary applications programmers.  So the initial strategy we felt with
JAVA was that it would solve sort of three – it would solve three
new value points.  It would solve the problems people had with CNC++.
We developed and intermedia instruction set and a virtual machine, which
allowed for portability across platforms.  And we wanted to solve the
right ones run anywhere problem on public networks.  And the strategy
at the time was we had to get to ubiquity.  What’s the definition of
ubiquity?  The market share of Windows at the time was 100 million users.
So I said, the goal should be in five years to have a 100 million users
cause I’ve been through the previous wars and I wanted to know how to
decide we were successful.  The licensing model worked, because we had
created the inevitability principal.  We had created the fact that we
were gonna do it because since the Netscape browser was the hottest thing
in the world, that meant that all Netscape competitors would adopt it.
And then we did one deal after another.  The initial deals all had a fixed
licensing fee, not a pre-unit licensing fee for relatively modest fee.
And of course the deal was announced with Netscape in May.  And the first
Netscape browser containing JAVA was not actually shipped until January.
And we spent all that time actually getting the thing to work, right.
So we did this early in the product cycle.  The – what’s interesting
is that we also were much more inclusive on how we did our licensing.
We worked very hard to talk to people in different industries.  Having
learned this polarization, which in heights I was incredibly foolish, but
I think we were just too young.  We worked with each of the constituents
and so we wouldn’t put in for example the JDBC interface, which is a
database interface until we had talked to all the key database companies
in the community.  So again, this community model was beginning to work.
Now the interesting thing about JAVA was it had some value propositions
that changed.  The first one was the, what we call the Dancing Duke.
And many of you saw the demo where the JAVA ball moved up and down,
and you could do that.  Today those are done by animated gifts, right.
We don’t need JAVA for that.  We – the right ones run anywhere
became less important because the Microsoft monopoly got stronger,
and of course Netscape is no longer a factor in the browser side,
that browser side.  So now JAVA is primarily used in mid tier.  And it
turns out it’s a perfect architecture for a lot of the new services
on server side.  And of course the beauty of this is that is changed
the perception of the inventors.  It became a huge marketing success.
And JAVA is both a great language in this tremendous marketing success
precisely because the licensing strategy worked this time.  And of
course it ended in  lawsuit complicated story where SUN, I did the
contract so I was a party go all this.  SUN actually ended up suing
Microsoft for breech of contract and was settled.  And Microsoft no
longer supports JAVA and has now built something called C-sharp and CLR,
which are essentially proprietary technologies with a slightly different
twist with similar goals.  And their technology looks pretty good,
but of course it’s not at the same time.  In thinking about this, I
keep asking myself, did all of the shenanigans that all of us here in
the valley, did they fundamentally affect the outcome of the desktop?
Had we done things differently, would they have fundamentally done that?
And I think the hardest conclusion that I’ve come is that it probably
didn’t, which I find very distressing, because I would like to be
able to tell you that we screwed up because we spend all of these years
with all of these different wars, the duplication and murders and so
forth and so on.  But I think in practice, our model still wouldn’t
have been strong enough.  And I’d like to come back to that because I
think that’s very, very important.  So what happened the actually JAVA
marketing consisted of an article with picture of Mark.  And of course
this was at the height of this.  This was an article from George Gilder.
And George took the strategy that we articulated and wrote it in a way
as only George could do, where he decided that JAVA would take over the
entire world and solve every know problem, ___________, right.  And what
was funny about this is this ASAP is only distributive people who are
Forbes subscribers.  You can’t buy it on the newsstand.  So I was in
New York at a conference and I get this phone call from Scott, another
executive saying, what’s this thing, cause the stock had doubled, right.
And you should notice this.  And everyone was talking about this, and
can I get a copy?  So I took my copy and I faxed it around.  And it
shows you that a well-placed argument at just the right time can have
an enormous impact.  And I look back, and I look at the key points.
This was a clear one of them.  And of course we got to the point where
not only did the press believe this, but we believed this too.  And so
it became a marketing strategy for a much broader set of initiatives.
And you can see this one.  And of course we even had even more fun with
upside, mom really is picking Scotty again, right.  This is back in the
negotiations.  So, I’d like to take a few minutes and talk about where
we are now, and see if we can sort of integrate the historical look to
where we are now.  And maybe you all have some pretty strong opinions.
A lot of you live through this with me.  There are two massive fights
going on right now.  And by the way I have a theory.  And I have a theory
that part of the reason I work in this industry and why I love it so
much is that we get to run this race every few years.  You too are gonna
get tested again.  You know, yes Eric, you’re going to have to have
an opinion again on the question of how do you structure for success.
How do you get these people to collaborate, you know, etcetera?  We have
two that I think are very much pressing.  I think the first one is web
services.  And this is an historical speech.  I won’t start talking
about any detail.  In 1990’s we focused on something – something
where we looked on, how did business logic within an infrastructure,
within a corporation get restructured?  There was something called
CORBA ______ Common Object Request Broker Architecture, one of my more
brilliant ideas.  In terms of naming it had objects and architectures.
And it had something called an Interface Definition Language IBL,
which was a contract between interfaces.  The CORBA ______ has not been
particularly successful.  It was very heavy weight.  There’s a number
of people who disagree with me in the audience, so sorry about that.
And I think it was ultimately not very successful because it violated
some of the networking principles.  It was too heavy weight.  It worked
relatively well within Intranets.  It did not work across the Internet
for various important reasons.  The JAVA team didn’t like that for a
whole bunch of reasons.  So they build a JAVA specific solution called
RMI, which became remote method implication, which became one of the
major issues with Microsoft cause they didn’t like that one either.
And they changed and they weren’t allowed to and that was part of the
suit.  There was this thing called COM and DCOM, which is the Microsoft
equivalent.  And the problem with these technologies is that they were
not sensitive to the networking world.  They made a set of assumptions
about security and routability and so forth, that ultimately were seen
not to be correct.  So in 2000, we have now embarked on a new one, and
it’s the world according to XML.  XML stands for Extensible Markup
Language.  There is a thing called WSDL, Web Services Design Languages,
how you specify your objects.  There’s something called SOAP, which
is the access protocol that actually allows you to invoke these things.
There’s something called UDDI, which Universal Description Discovery
and Integration.  That’s _________ name.  I was like, you know,
something of a cow.  And of course the architecture sounds great.
You know you read the marketing article.  If you actually look into it,
it’s full of problems.  For example, there’s no standard way to do
authentication.  But over the network people care about security and
authentication and indeed a number of vendors have highly proprietary
authentication mechanisms, another example of a lock in.  So you’re
being tested again because you’ve got to find the lock in strategy.
You’ve got to find the thing that would break this.  And I’m still
thinking about it.  UDDR for example publishes something call the White
Pages, the Yellow Pages and the Green Pages.  The White Pages is about
people, Yellow Pages is what the companies does, and Green Pages is
access to the services that that company or organization publishes.
Makes sense on paper how you authenticate, how do you get access, how do
name it, how do you route to it and so forth and so on.  At the same time
Microsoft has announced something called Dot Net and something called Dot
Net My Services, which are their versions of this, right.  And all of a
sudden we see we have another opportunity.  We have a group called the
Liberty Group, which is busy building a competitor.  The Liberty Group is
organized roughly like these other folks.  Another example would be Linux.
This is an ad, which I found profoundly humorous because it’s from IBM.
And if you go back and you look at the role that IBM has played in terms
of licensing, and you look at the investment that they’re making in
Linux now, it is one of the greatest 180 degree reversals that I’ve seen
in strategy since I’ve been alive.  In the Linux community, give you
the number.  Linux has about 20 – these are IDC numbers from last year.
Twenty-seven percent of the server platforms are Linux, mostly running
Apache, 41% are Windows, or Windows related.  In the client, Microsoft
has 91% Windows, 3.6% for Mac, 1.4% for Linux, so competition on the
server, not so much on the client side.  In the Linux area there are many
people who talked about how to solve the problems on the client side.
And there are a lot of very smart people working on it.  And there are
too many smart people working on it, because now there’s two Window
systems.  One is called KDE, based on C++, and the other one’s called
Gnome based on C.  It was originally developed as a replacement of Motif.
There’s a group called Mono, which is not exactly a brilliant name for a
group, which is essentially a public open source organization funded by a
company by Zimmian and number of other companies that are trying to build
clones of all these protocols.  So perhaps where we’re going is we’re
going from open systems to open protocols, or processes, or something like
open services.  But the same dialectic, the same problems exist.  And ten
years from someone will stand up in front of this audience, perhaps in
the same room and talk about this phenomena.  And the best part of this
is they have Linux International just like we had Unix International.
So I started and I didn’t bother to bring them.  I read a whole bunch
of articles.  And the criticisms are exactly the same from literally
20 years ago.  Their difference licensing, there’s something called
the GPL, there’s something called the SUN Community Source License.
There’s a free BSD license, which allows some limited commercial use.
So now you go, well okay, since you’re participating on this, what
should we actually do?  Well let me pause at something, which makes
sense to me, but I can assure you will not happen.  The Linux Kernel had
tremendous support for it.  If the Solaris and other Kernel technologies
were open sourced in such a way that the Linux community could get the
benefit of the work that they’ve done, you would get a combined Kernel.
If some combination of the tool kit strategies ended up with a single
API that spand all of these systems, you can get tremendous market share,
especially in the server side.  And you have majority market share on the
server side.  JAVA could for example be open sourced.  It’s currently
controlled by a slightly different license.  And of course why would you
object to giving the JAVA source away, cause you know Microsoft is not
going to take it anyway.  That’s the one thing that they’re not gonna
to do.  Of all the things they’re gonna do, they’re not gonna do that.
And you can open source that on top of the _______ machine.  You could
convince Apple to offer the Mac OS20, the first good user interface on
top of Unix in my lifetime, right, cause remember it’s running on top of
Mac.  And you could put that in open source on top of the merged product.
So our office for example from SUN is already open sourced.  The beauty of
my proposal is that you would then have best of breed at each and every
protocol application and product layer in the industry.  And that’s
a starting point in my view.  It’s a required starting point for
dealing with the strengths of the competition in this particular case.
The reason this won’t happen is because we’ve proposed in various
forms these kinds of things over many years, and it doesn’t happen.
And it doesn’t happen because of the structure of the industry and cause
people don’t see it.  Now there’s an alternative.  There’s another
alternative.  And the other alternative is to say that, for example,
the US government, which is important __________, only purchase software
that deals with data formats that are published, right.  Make sense?
And now we have the property, we would guarantee competitive choice
at each level.  What are the odds of that?  It’s been discussed for
10 years.  Hasn’t happened yet.  An alternative is in my view to see
the desktop, right, and go focus on specialized areas.  And that’s what
most people are doing.  And I think that’s the weakest, but probably
the most likely to the alternatives.  So what are the new rules, really?
Let me put my collection of buttons up.  What are the new rules?
I’ll finish up get your comments and questions.  Merge software
doesn’t work, right.  We now have example after example. You can borrow
technology but you can’t merge it.  You have to _______ in some way.
Compatibility libraries do work as long as the underlying thing is
the stronger of the two.  ________ don’t work at all.  The people
who went off to join these consortatives all got absorbed into the
DOT COM bubble where they then of course _________ for other reasons.
But none of them fundamentally have the kind of impact that I’d hope
when the marketing and the pitches really________.  They don’t work.
There will always be an open source choice.  I was part of one.  There is
a wonderful on now involving Linux.  And a whole generation of people
Google, for example, perhaps the largest Linux cluster in the world with
more than 12,000 machines had tremendous success using Linux.  There will
always be a Microsoft or an in the role that they played 20 years ago.
So the things that I’m talking about, I now have the one benefit of age,
which is I have some perspective.  These problems are not going away.
I think there’s sort of two choices that you have when you think about
strategy.  The first is, cheap volume proprietary.  And I would suggest
that Microsoft, ebay, ________, great companies following a very focused
strategy, large market share, very low cost, excellent penetration,
good customer relations overall.  The second one is to standardize and
extend through collaboration.  But if you do that, you’ve got to start
from the premise that you have the best product in the world, which is
hard to assemble.  And then you have to execute the rest of the strategy
flawlessly.  You have to have full rights to implement the API, no fees,
no hidden tricks, and you need to innovate even faster than choice A,
because they’re the default choice, right.  And getting that right
and maintaining that turned out to be so much harder than any of us and
certainly I – that I did.  So everyday we run this contest, and I –
part of I think, what I wanted to communicate with you is that there is
a younger generation of people represented in this room who are smarter,
harder working, better trained than I every was with the people that
I worked with.  So I’m throwing this to you, right.  You get to run
the race again.  You got the same set of problems.  You got the issues
around consority, you got the issues around licensing, the stakes are
even higher because what we do is so incredibly important to the world.
So run the test again.  And this time it’s your success that’s on
the line.  And see if you can pull it off by combining just the right
set of laws around innovation, inevitability, and ubiquity.  And then I
think you could have a huge success.  So with that, thank you very much.
(Applause) So I think we wanted to do questions or comments.  And I’m
happy to have people debate or argue or say I’m wrong or whatever.
Go ahead.  Yes.  Yes Sir.

M1: Hi Eric.

Schmidt: It’s high, go ahead.

M1: Enjoyed your talk, agree with it entirely.  The issue though, I think
is a social issue, and we tend to avoid the social issue, because we’re
all wrapped up in things.  But in order to deal with technology, you have
to have some ego involved in things.  There’s a massive amount of ego
in all of these structures, and it’s a bit like Afghanistan.  You’ve
got a bunch of competing gangs that tear things apart.  And trying to
get these people to collaborate in order to have the innovative muscle,
they’ll wish or think they can do it better on their own.  So that’s
where a lot of this dynamic comes from.  How do you latch that in a
direction where you can team a bunch of horses together?  That’s been
the problem running a company, but it’s kind of like you need to be the
CEO of CEO’s to be able to pull all of these other forces in together.
And that sounds like the challenge to me.  And I’m interested in
the way in which you might articulate, bringing together such people,
because your idea, which I’ve heard around SUN before back in the SUN
spark cluster days.  It’s been around there for a long time.  A lot of
reasons why that didn’t happen in the early stage is the same reasons
probably why it isn’t occurring now.  I’d like to see how you might
propose to latch these horses together and talk to that particular group.

Schmidt: I never worked at Microsoft, but I assume that within Microsoft
they have exactly the same battles that we do among our community, but
they’re within the company as oppose to in the press and across the
table and with huge negotiations.  I don’t know how to do it.  I know
however that you have an opportunity it would exist in industry.  I do
think with web service as in Linux, there’s an opportunity to assemble
just the right pieces with just the right motivation to run the race.
I really do, because if you look at the market share gains that Linux has
had are surprisingly good, both on the server and also on the – believe
it or not Linux _________ and the client has more or less doubled in a
year from like 1% to 2%.  But they’re at least making some progress.
And a lot of people want an alternative, right.  The people want choices.
And they want some differentiation and so forth.  And I think ultimately
it’s a leadership test, and the problem I – speaking as a CEO of
a public company now a private company, there’s enormous pressure
on quarterly earnings.  Part of the issues that, for example SUN and
IBM and HP have trouble is that they’re on a quarterly cycle and they
have shareholders.  And ultimately these are very long term strategies,
right, many, many year.  And it’s hard cause they’re hard to sell
and hard to maintain.  Yes Sir.

Man 2: I have a somewhat different point of view.  I’m an end user.
I’m a businessman.  So I’m not and IT developer.  And I think
with all due respect that you’ve missed the point completely, okay.
And I don’t – I don’t mean to be negative at all.  I wanna give
you something, okay.  Number one, if you are in a marketplace with a
dragon who intends to dominate completely, if your strategy does not
include killing the dragon, you will fail, okay.  That’s very important
to understand.  And you have talked about several dragons over time.
If you wanna survive in that marketplace, you must have a dragon killing
strategy within your business plan.  Number two; what you want in order to
succeed is my money.  You want me to spend my money on your innovation,
your product, and your effort.  And in order for me to do that, I want
to know that you’re solving my problem.  Not yours.  I want you to
be a designer who takes my problem, your genius and produces something.
I don’t want you to spend your time entertaining yourself and expect
me to pay for it.  I really want it to work for me.  And I’m not
being selfish about this.  And this is the secret.  This is the secret.
Make it work for me, and I will give you all of the money that I can that
I make in using your product.  I will give you a very large portion of
the profit that I make if your product works for me.

Schmidt: So, one of the things that I think we have learned about network
affects is that the market which – when I started in this industry,
the rule was that the first incumbent in a competitive market had, oh 50%
market share.  The second one had 35% market share.  The third one had
10% and everyone else was 5% or less.  I think what we will learn in
networked intellectual property markets, if that those percentages are
wrong, if for a variety of reasons, including vendor lock in, customer
stability, the real numbers are 90%, 9%, and .9%.  If I follow the line
– your line of reasoning, then there can only be only one inhabitant,
if I take that insight, which I believe is true, by the way.  I actually
believe that.  And I take your insight, and I combine the two, then
there can be only one choice each of the protocol our services ________.
And I think that’s where we are today.  The fact of the matter is
that because the competitors for Microsoft through this – all the
shenanigans that I took you through, did not ultimately build a model
that was significantly better from a customer perspective.  And that’s
hard to say, cause I – we tried.  And by the way, our business plans
all said we could pull it off, just to be clear.  We’re not, you
know, the business plan actually claimed all that you just said, but it
just didn’t happen.  And what that says is that the industry will be
forever sentenced – forever sentenced to have a unitary choice in each
of the important control points.  I don’t like that answer.  It may
be correct.  And I think there’s a lot of evidence to say that it is
where we are today.  But I think we can do better.  A classic example
is that today there’s a battle over instant messaging.  And most
customers would prefer to have a single instant messaging service.
In this particular case, AOL is providing the majority share and the
other guys are trying to block against them.  And nobody is committed to
an open approach, an open process, and open protocol approach.  And I
think we would be better served by and open interoperable solution.
So here’s an example where what I said I agree with in principle,
but in practices it doesn’t get produced that way, cause of the 90%
rule and the because of the competitive market.  Go ahead.  Yes Sir.

Man3: Question is somewhat less weighty perhaps.  You said earlier on
that the SUN University Network was licensed to six or seven companies
– I’m sorry, the Stanford University Network was licensed to six or
seven companies, and SUN was the only one that succeeded.  What – who
were the others, and what happened?

Schmidt: I actually tried to figure that out and I couldn’t remember.
They were graphics companies.  And the idea was that they would be
embedded in graphics, what were then thought of as graphics workstations.
This is back before the – remember this is roughly when the PC had
just happened.  PC didn’t have an integrative graphics frame buffer,
since graphics were terrible.  And so these people were off building
specialized graphics computers.  And none of them made it.  Thank you.
Vaughn, I think you did the deals.  This is the problem with an audience
like this; people actually know the answer, thank you.  Vaughn what was
the list again?  What was the list again?

Man4: (Inaudible) forward, fortune…there’s four more.

Schmidt: Okay.  We got it.  Good.  That’s a network effect if I ever
saw one.  Yes Sir.  Oh I’m sorry.

Woman1: Actually Eric, actually thank you for the talk.  I actually
work for SUN.  I’m actually a __________ technologist myself, and I
have a different perspective.  Today I see actually two battles going
on, and you are correct.  One of them surrounding common services.
The other one surrounding web services.  And there’s no real differences
between the two of them.  Your are correct that there is no registration,
authentication, standard.  There’s no way to actually port or do the
protocols.  But then I hear both sides of the story.  It’s really
hard to architect something that makes the business happy.  Yeah, we
love to take your money and make something out of it, but how would you
go about in changing what I call more like social issues, such as, you
may created the greatest application, say, Star Office for that matter.
But Star Office is like WordPerfect, clunky and slow.  So how you create
something so that you can accept, have the user community accept it
without having the backlash?

Schmidt: I think that the bar for new products and new vendors continues
to be higher, right.  And what happens at each and every level it gets
harder to overcome whatever level of satisfaction that they have.  If you
go back 20 years ago there wasn’t a personal computer, 20 years ago
there wasn’t a workstation industry.  People were on mini computers.
And so the bared entry was relatively low.  That people could enter it,
they could try things and so forth.  The bare to entry now for desktop
computing and servers is extremely high in what we think of as client
server computing.  I think in web services you can see that because
there’re actually very few web services in use.  If you go to the UDD
directory that exists, there is very little.  The only one that anyone
talks about is Passport and it’s down half the time cause they have to
turn it of because of security architecture problems.  They’re busy
working on it.  So I think that the problem here is once again we go
too good at talking to the press and talking about the marketing of
our products.  And the products themselves are either not shipping,
not integrated.  But the fact that I had to repeat to you what SOAP
was indicates a failure of marketing, right, because if you have to
understand what the access protocol to talk to the objects is – are,
then this is not an object you’re gonna be consuming anytime soon.
This is like discussing the details of spark caps in your car.  Yes Sir.

Man5: Eric thanks very much for your speech and I realize you’ve
ran this race before.  But if you have to wake up now, let us suppose
you’ll have to run the race now again, what would your focus be?

Schmidt: I’m actually a Google, so I woke up and went someplace else.
And Google is a pure example.  I don’t wanna turn this into a Google
pitch cause I don’t think that’s appropriate.  Google is a pure
example of a new model.  It’s a brand new product.  It’s architected
very differently.  It doesn’t carry over the strategic problems
that I highlighted.  Whatever strategic problems we have, will be the
subject of a subsequent talk by folks here in this audience 20 years
from now, because we’ll fight – we will also fight battles with
the standardization in access.  But they will be at the services layer,
not at the platform layer.  And I can tell you that the ____________.
There’s a book called, Computer Wars, which is a documentation of the
IBM battles around establishing the mainframe interfaces that brought
IBM to dominance in 1968 to 1972.  I wasn’t part of those, so I
didn’t talk about them.  The battles and strategies are identical
to what I described.  And the reason IBM was so successful during
that period of time is to a series of tactics, which were ultimately
found largely illegal through various _________ and you can go through
all the details there.  They got all the application vendors to favor
them and to get off of everyone else.  And they used software rental
and hardware rental and various clever ways to do it.  So controlling
an application and controlling a platform turns out to be the maximum
control point.  One way to think about the value of Microsoft, which is
a 100 plus billion dollar preparation, or the value of Apple, which is
much less than that.  You see what you’re really seeing is you’re
seeing a value of the architectural franchise that they have built,
cause it is that architectural franchise that allows them all forward.
And that’s what we do.  And now for example, the Linux communities,
they’re trying to establish an architectural franchise.  And it’s
their turn to do that, right.  Yes Sir.

Man6: You spoke briefly on the role of the government and why things do or
don’t happen within the government itself.  And again, being somewhat
older than you, hopefully I have a perspective.  I can think of several
things that happened over the last 25 years with IBM you mentioned was
one.  But through government action or inaction literally forced the issue
of adoption of what ultimately became the broader platform if you will.
I think the most obvious one that I saw in the industry in which I was
in, it’s different for most of the people here, it was all airspace
and defense, was when the IBM architecture Microsoft at all became the
platform of choice in the late ‘80’s as oppose to say, an excellent
system from deck.  And also some excellent systems from Apple at that
time, which many, many people were using.  The government literally
made the argument, as did a number of large businesses.  Our big iron
is big blue; therefore our desktop should be big blue.  We had a war in
my company because we were using lots of decks _________ and we wanted
deck desktops.  But the company ultimately listened to the insurance
and contract people and went with the IBM Microsoft architecture.
So did the government.

Schmidt: I think that it depends on the scale and place you in the
technology strategy.  But it’s actually useful to have smart customers.
And most of the initiatives that began successfully started with highly
knowledgeable customers who had specific need that was not served in
the market.  So any strategy going forward has to solve a new problem.
Let me give you an example.  If you take my little scenario, which is
Star Office, and the Mac, and X desktop and the merger of Linux on a
desktop, how many of you would switch to it because it was more useful
as I described it?  You’d switch to it because it had the Mac use
interface, something which you’re familiar with.  But I haven’t
described something that was a different experience completely.
I haven’t given a really compelling picture.  So it would have to be
something more than that.  It had to be better integration, some service,
which you can’t do.  So I think that at each stage in our industry there
is an incumbent and there is a challenger.  And your story is a good one,
because these wars were practiced since I think the founding of computing
because the economics work.  The difference now is because they’re
in software.  The market share games are much faster.  As an aside, we
use to have a – during the bubble, which I was proud participate in,
but didn’t sell at the top unfortunately.  We had this sort of clock.
And the idea was that you would call up a press person, and you would
announce that there’s going to be a new Web site.  And you would get
them to write an article about this new Web site.  And that a result
of this article, there would be a million visitors to that web site.
Then you’d call up the press person and you’d say, “Well there were
a million users to the Web site, so they’d write that article, right.
So then you’d get 10 million visitors to the Web site.  And then of
course you get the article about the next site, I mean, you’re caput.
So the thing that’s different now is the ability to pulse society.
There is a lot of theories about this.  But the access to information
broadly allows you to play these games more quickly.  And that’s why
I think there’ll be another bubble around technology.  There’ll be
another speck, not just financial, but also technology speculated
one. And I think all of those could be alive and able to participate
in it.  And say again, 15 years from now, yeah this is like the one we
saw last time.  David.

Man6: Eric this has been fascinating to me because I watched this whole
thing from a different position.  As you know at Intel is part of the
Wintel combination for running the microprocessor business from ’78
to ’91.  I watched all of this happened and you might think that we
would’ve been sort of anti this.  But on the other hand we were very pro
Unix simply because we were tied with Microsoft and fighting for power.
And should Unix be successful, then Microsoft’s power would be weaker.
And ours relative would’ve been stronger.  So we – we’re aligned
with this activity.  But I’ve got a little different perspective on what
happened during this, and therefore impacts what your recommendation is.
And what I observed is with Microsoft you did have the CEO of CEO’s.
There’s one person that could make decisions.  Unix really got started
– the thing that made it take off was BSD and these free – nearly
free copies.  And it was like share ware and it was very Linux like at that
point in time.  And that works very well for software people, and people
who are generating code.  But it’s not a business.  And the problems
all happen because these became businesses, and each force every group
of engineers reported to a separate set of shareholders.  And each set
of shareholders wanted primarily their net worth to increase, therefore
the quarterly reports and the financial requirements and as – that you
had mentioned before.  The possibility of getting these multiple forces
all to align our single direction, that is to have a single direction
in their own best economic interest is almost impossible because of the
number of combinations.  And it seems to me that any of these programs
or initiatives that require the longer term more than a few, you know,
one day for a press conference would be longer term.  But a longer term
of cooperation between multiple sets of shareholders is inherently flawed
and won’t work.  I’d like your comment on that.

Schmidt: Well that’s I think and accurate description of where we
are now.  I’d like everyone to remember – remember all the deals the
Apple did with IBM, right.  Remember all the special companies that they
built and so forth.  I think like maybe you were part of one of those
when your boss called up and said, “Yeah you’re working over there
for a while.”  We’ve tried to solve this problem.  We’ve tried joint
ownership. We’ve tried joint ventures.  We have not solved this problem.
I – and I agree, I think it the core problem.  It’s a governance
problem.  You said I think earlier that it was an ego problem.  I think
it is an ego problem.  And I think we had it.  But I also think it’s
really a governance problem.  If we organize it differently, I think the
right answer for this is to look at the models that have worked.  If you
look at the ITF, many of you have participated in that.  That’s a model
that worked very well.  It brought us the Internet, right.  People like
Vint and other heroes of mine, you know, who operated selflessly and
really did.  And I think there’s a category of things for which that
would work.  But it worked for 15 years.  And 15 years is a long time.
If you told me that for the next 15 years, we have a Vint type person
in an ITF structure that’s going to work on services that are gonna
be truly interoperable and extensible, I’d sign up in a heartbeat,
cause that model works.  And it’s one of the few that I can point at
that’s been highly successful for us.  And again it brought us the
Internet and brought us all the successes we’ve seen today.  Yes sir.

Man7: this is actually right along the same comment, but it’s a
slightly different point see that the governance problem, that if only
if you could get these folks to work together, in other words, if you
just got them to all see that their personal long term benefit is by
working as team, that actually there’s another problem that may make
it even harder.  And that’s really a technical and a business problem.
Okay, first of all, technically it’s always hard to get different groups
to work together, okay.  And you will always need some czar that makes the
decision that is suboptimal in some fashion.  But any decision is better
than no decision, okay, and that there’re real technical problems there.
But the other thing is that also, I mean, if you’re fighting a dragon,
what Microsoft does is it takes serious loses on one side, okay, to make
sure that it captures that, okay.  Whenever you have different business,
I mean unless you can come up with some strategy, and I can’t see
how you could come up with some strategy, or you can take a big loss
in one company, because it’s gonna help the other company, okay, and
you’re gonna help the staff.  And I don’t, you know, that’s sort
of an argument that says that won’t work, you know, no matter how good
the intentions people are.

Schmidt: Let me give you another example that I think has worked well.
The structure of the Internet, this is not an Internet talk.  And as you
probably know, I spent the last five years talking about the Internet.
This is kind of different from what I normally talk about.  The Internet
is organized in a way that I think it gives us hope, because none of the
existing players, the long haul carriers, the ISP’s, people like UUNET,
who at various times had large market shares either chose to or were able
to turn them into dominant monopolistic structures.  There’s a couple
reasons for that.  The Internet itself and the routing architecture made
it hard.  The business economics did not promote it.  The community was
more collegial, right, as a group.  Speaking as a member I can tell you
it’s more collegial, that group rather than the other.  And I think
we benefited from it.  So for example, when the leaders of Cisco and
of course they played and important role in this.  When they would all
get together, they were actually nice to each other.  They actually
collaborated on – I mean this is actually important.  I mean as a
note, you know, President Bush and President Putin met at Crawford Rams.
This is actually important in the thing.  It helps avoid war, which is a
good thing to avoid.  And so I think that there is a model in networking
and it’s unfortunate that in the platform economic world, that model
is followed.  Couple more questions, and I wanna be sensitive to time.
Go ahead.  Yes Sir.

Man8: In your talk you mentioned some of your big disappointments over the
last 20 or 30 years of computer history, and maybe these are side effects
of what you’re talking about.  But one of the things that I’ve seen
as a big disappointment is that the best technology always loses.  And in
fact it’s worse than that because it keeps getting reinvented and losing
again, you know.  How many times are we going to reinvent virtual memory?
How many times are we gonna reinvent ACL’s?  How many times are we
gonna reinvent, you know, just a few examples like that?  Can you touch
on that topic?  I mean it’s just – it’s very disappointing.

Schmidt: And I know this is a very broadly held view.  Can I again just
disagree with you on one thing?  The best technology as defined by the
customer wins.  And turns out that sometimes their perception of what
technology matters and what the aspects that matter turn out to be very
different from what we as technologist think that are.  Turns out in the
PC industry, the single most important thing was an interoperable hardware
platform that had declining cost economics, right.  And it turned out
that that was more important than the elegance and simplicity of the Mac
user interface, which is at the time was an extraordinary gap.  There are
lots and it’s a much longer conversation.  But one of the things I’ve
learned is that the customer – going back to this earlier gentleman
who disagree with my point of view.  In this particular case we agree.
But the customer ultimately does speak with – and they eventually win.
And any strategy that involves sort of continually disappointing them, is
a problem.  Part of the reason that why the Microsoft and Intel alliance
with all their partnerships ultimately evolved, which they solved the
customer problem that the customer had at the time.  If you go back to
NEWS, which is a technologically brilliant product, it didn’t solve
the problem that the customer had as ____________ technology, which
was clearly inferior.  X has become the dominant standard.  And we all
love X now.  Also the people who built – well maybe not everyone.
It turns out that in the NEWS case, the most interesting thing about
NEWS was that the people who did that were the JAVA team 10 years later.
So sometimes you have to reinvent it in a different context.  JAVA, which
again is a crowing achievement for all concerned, and I think a great –
a wonderful thing.  It solved a lot of problems, was the right integration
of a whole bunch of technologies, which had been around for a long time.
I can tell you by the way, we’re not gonna be reinventing ACL’s
for a long because there’s two dominant operating systems, Windows
and Linux.  And they both already have them in various forms. So again it
may be that we’re now gonna go reinvent everything one level higher.
Our classic example is instant messaging, another one is in Kodak’s,
the various video interchange formats.  It’s like why do I need another
one, right?  Wasn’t that one good enough.  We keep inventing them and
inventing them.  And that seems to be a structural problem.  Couple more.
Yes Sir.  We’ll try to do it quickly.  Yes maam, excuse me.

Woman2: All right, so, you kept talking about this and the customer that.
And you mentioned highly knowledgeable customers, but of the customers out
there aren’t highly knowledgeable.  And they’re not driven by facts
and technology, but by emotions __________________ Mark Anthony thing.
Brutus went first, he appealed to logic and abusing and then Mark
Anthony went and stirred everybody up _________ Brutus.  Well the JAVA
article that was written about why JAVA will solve everyone’s problems,
that’s what got SUN’s stock to double.  That’s what got everyone
really interested.  It’s also social issues and communication now.
I was as Packard’s conference a month ago, and that place was talking
about how like criticism is not allowed anymore, because of the MCA.
And there’s all these social problems, and if you can decide what
the technology – what we want technology to do, then we can do it.
It’s just defining the problem.  We can do it.  It’s getting out to
the people, because the 100 million users that use Windows, those are
the normal people, the secretaries, the (Audience Laughs) – I know,
seriously those are the majority of the country.  We like to think that
they don’t exist near the valley, but they do.  And they’re the
ones that buy everything.  And so I think we need to start figuring out
what the normal old problems are that people care about, because when
you solve big things to solve your problems then you have to educate
the people as to what the problem ___________.  That’s a whole big
problem. (Audience Laughs)  Pretty much, how do you…

Schmidt: In my – my experience with the average person, which consists
of my mother, is that they struggle with the complexity of what we build.
And I think that as proud as we are of all of us for what we’ve done
in terms of technology, and I took you through my variant of our history,
you know, my sort of 25 years of history.  Our products are hard to use,
hard to install, hard to administer for normal people.

Woman2: He is a MDU.  He has to come, he happened to ask me, how do I
do this in Windows?

Schmidt: Well thank goodness he had you.  Couple more questions, quickly.
Thank you very much.  Thank you. (Applause)

Man9: Eric we have one very small token of appreciation.

Schmidt: Oh wonderful.  Oh thank you very much.

Man9: Thank you Sir.

Schmidt: Thank you very much.  Okay, take care.

Man9: I think Eric will be around for a few more minutes.  There’s a
reception out in the back.  There’s some refreshments.  I hope enjoy
yourself, and please keep us in mind beginning of the calendar year.
Thank you very much.