The "Critics" comments in this article and the response by
     George Gilder, provides a second round of third party opinions
     and analysis that follows those published in the February of 1995.

The Gilder articles provide some interesting technological and cultural backround that helps prepare readers to better understand and place in proper perspective the events relative to the National Data Super Highway, which are unfolding almost daily in the national press. I contacted Forbes and George Gilder and obtained permission to post on the Internet.

The letters from Mr. Gilder's critics have posted without the express permission of each of their respective authors. The postings have been made under the doctorine of "Fair Use." If any author wishes to have his letter deleted, an Email message to such effect, addressed to, containing a reply address and/or telephone number will initiate a dialog.

Please note that the preface must be included when cross posting or uploading this article.

The following article, GEORGE GILDER AND HIS CRITICS, was first published in Forbes ASAP, October 9, 1995. The article contains letters from various correspondents commenting upon a wide variety of issues raised in George Gilder's "Telecosm" series which will be published in 1996 by Simon & Schuster, as a sequel to Microcosm, published in 1989 and Life After Television published by Norton in 1992. Subsequent chapters of Telecosm will be serialized in Forbes ASAP.

                       GEORGE GILDER AND HIS CRITICS

          Every now and then, timing is all.  Take George Gilder's
          Aug. 28 Forbes ASAP piece, "The Coming Software Shift."
          Lucky Forbes readers got their copies in mid-August, smack
          between the year's two biggest technology events--Netscape's
          record-hot IPO and the release of Windows 95.

          Gilder used the timing to explain why creative energy and
          profits in desktop software would soon migrate from Microsoft
          to the Internet and toward companies like Netscape.  Needless
          to say, Gilder's point is controversial, drawing fire and
          praise from a host of computerdom's biggest names.

Mac R. Holmes, Troy, Ala.

Elite Fallacies

   I recently finished reading "Mores:  Not So Hot Potatoes" by Owen Edwards and
"Telecosm:  The Coming Software Shift" by George Gilder, in your Aug. 28 issue.
As far as I am concerned, Mr. Edwards is getting real, but even he still misses
some major points.

   1. Just because it is on the Internet does not make it true or real.

   A noted scientist who has used the Internet for several years recently
commented that the "Net" is becoming so congested with traffic and garbage from
nutcases that he has begun to use the telephone and the U.S. Postal Service
again.  At least he knows that his messages are received and that they are really
private.  He does not have to waste [time] waiting.  And he also knows that the
information he receives has not been tampered with by some weirdo.

   2. It is unclear to me why I should have to maneuver through the Internet
when I can easily dial the bulletin boards directly or one of the commercial
databases directly.  Why waste my time and money?  I have zero interest in letting
anyone have direct access to the models or data that I use.

   I wonder how many companies really are interested in letting hackers and/or
competitors invade their books or their customer databases through the Internet.
And such an invasion is quite possible unless steps are taken to isolate the
computers tied into the Internet.

   Unquestionably, there are some uses for the Internet, but I think that those
uses must be very well thought out and researched lest the problems overshadow
the value of the uses.

   3. Mr. Edwards alludes to the "interpassives" as though anyone who has little
interest in "getting on the Net" must be dumb, lazy or out of touch.

   Frankly, I do not give a damn about computing after I leave the office.  I'd
rather read or fish or hunt or walk or write or shop in real stores or sightsee
or eat out.  And I find real friends stimulating.

Nathan Myhrvold
Group Vice-President, Applications and Content Group, Microsoft

As usual, George Gilder's diagnosis is dead-on.  The software industry functions
by turning its fuel-cheap processing power, as described by Moore's Law--into
value for consumers.  In turn, software soon soaks up CPU power, fueling market
demand, which sustains Moore's Law.  The software-hardware twosome is about to
become a menage a trois, with the addition of communications.  The three-way
cycle will drive growth and value creation in each area at an even faster rate
than we've seen so far.

   I begin to differ when George frames the computer industry in lurid terms of
battle.  In reality, computer industry shifts are quite dull affairs.  The
multidecade decline of the mainframe has been a lot like watching a glacier
melt.  A leading company fails to invest in new technology and ultimately finds
its product obsolete.  New entrants might capitalize on this, but most --
including Netscape and HotJava--blaze new trails instead.  When an established
leader does fall, the proximate cause is almost always slow suicide or atrophy
through internal mismanagement;  not the wounds of combat.

   The Internet means opportunity for all software developers--new and old.
Mosaic and its commercial descendants are currently at the center of the
cyclone, but their day in the sun will pass just as surely as it did for
previous Internet standards like Gopher or FTP.  The instant rise of Mosaic and
Netscape foreshadows what George might dub another Law of the Telecosm:  In a
world with easy electronic distribution, it is very hard to hang onto your
position--easy come, easy go.

   A particular weakness of HTML is that it is a data format no different in
spirit from the protocols of 1970s-vintage mainframe terminals.  Next-generation
Internet software will use programmability at both ends of the pipe.  Java is
noteworthy and may play a role, but its hype falls into a very old trap.  Every
new programming language is heralded as a breakthrough.  Fortran was going to
computerize science and engineering, Cobol would revolutionize business, and
Lisp (and later, Prolog) would deliver us to the promised land of AI.  The
reality is quite different--new languages offer some modest benefits, but all
of the real value resides in the programs created with them.

   Finally, I must comment on the obsession with finding the next Gates.  Bill
has been my friend and colleague for nearly 10 years, so the topic is familiar.
Writing great software and founding a company is a first step, but that only
matches Bill Gates circa 1976.  The secret that made the Bill Gates of 1995 is to
have the technical and business skills to develop new product lines repeatedly.
Microsoft is unique in this accomplishment.  Our competition is invariably
single-product companies, or ex post facto assemblages of them.  I don't doubt
that the next Gates will eventually appear, and the Internet may fuel her rise,
but it's one hell of a tough act to follow.

Scott McNealy
Chief Executive Officer, Sun Microsystems

I'd like to compliment George Gilder for so eloquently articulating Sun's nearly
decade-old credo:  "The network is the computer."

   I believe in Gilder's thesis that we are, indeed, at the onset of the age of
network-accessed, disposable software.  Desktop operating systems, utilities and
applications will soon be relegated to peripheral status.  The network will be
the operating system, and, as Gilder points out, this change is being brought
about by the almost incomprehensible momentum of the Internet.  The ultimate
change agent, Gilder suggests, may in fact be Java, yet another technology that
Sun has offered to the industry in an open and barrier-free way.

   Java is a programming language that creates the first true network operating
system for the globe, because it is safe, secure, robust, fault tolerant and,
among many other attributes, platform-independent.  Here's how we see it playing
out.  Three computer science students from Berkeley hacking code late at night
will create a Java word processing program.  Let's call it "NetWord."

   They put it on their Web server at  It will be free
because these kids want fame first, knowing that will lead to fortune as with
their hero, Marc Andreessen.  It will be 100,000 lines of code versus 1.5 million
for WordPerfect, because NetWord isn't bogged down with all the APIs
[applications programming interfaces] and other extraneous baggage of typical
applications that must interoperate with a specific operating system, or have
become overfeatured.

   So you're shopping for a new word processing program.  You've seen the ads
offering Word on sale for $ 249.  Of course, you know that means you get the
privilege of upgrading in a year for another $ 89.  Your buddy sends an e-mail
and says check out NetWord.  You take a test drive.  It works!  Cool!  You are not
alone.  A few million others have heard about NetWord.  Now the makers have a
following and in a few months they start a company and offer NetWord 2.0 for a
mere $ 2 per copy.

   You and 10 million other avid followers have no qualms about paying the
nominal fee.  It sure beats the competition.  In fact, at that price, you'll be
checking the NetWord home page monthly to download the latest copy.  As for the
three college kids, they are now millionaires.

   Suddenly the OS that controls the CPU on your desktop is a legacy of an old
paradigm, relegated to a secondary tier in the software food chain.  This means
that Microsoft is now a severely overpriced personal productivity tools
applications vendor with an OS business that is no longer able to "captivate"
the end user by being the only platform to run desired applications.  Keeping
APIs secret won't work anymore.  Controlling the interfaces won't work anymore.
The wall is down and there's no turning back.

   So where will that leave Microsoft? Scrambling to explain to investors why
the company's market cap is 10 times revenues, scrambling to explain how they
can survive in the world of disposable, run-anywhere Internet software.  Of
course, the marketing muscle up in the Northwest will try to figure out a way to
play in this space.  But playing now means playing on a level field.  The rules
are the same for everyone.  They're spelled TCP/IP, Mosaic, WWW, HTTP, HTML and
now Java--all the open protocols and languages that make up the Internet.

   It's barrier-free computing.  It's consumer choice.  It's a free-market economy
in the information age, finally.  From there, just about anything is possible.

Larry Tesler 
Vice President and Chief Scientist, Apple Computer

To report technology history accurately is a difficult task requiring
percipience, perspiration and perspective.  Gilder has done it.  He has combed
through the attics of computing and assembled for the family a gripping album of
the events and circumstances that created the phenomenon we call the Web.

   If the past is buried in clutter, however, the future is still more difficult
to perceive, especially before the dazzle of Netscape and Java.  In my view, the
Web is a catalyst, but it is not all that matters about the Internet.  The
browser is to the Internet what the spreadsheet was to the desktop computer.

   Mosaic is the new VisiCale;  Netscape Navigator the new Lotus 1-2-3.  Microsoft
would like to make Explorer the new Excel.  Wall Street is as enamored with
Netscape today as it was with Lotus when that company went public.  Netscape is a
winner, but is it the winner? With the aid of its ally, the Sun King, will it
inherit the throne?

   Where is the proprietary operating system that runs not merely the equivalent
of spreadsheets, but thousands of applications? Where is the API that some
megalomaniac weaves into, well, a web, to attract, ensnare and ultimately
consume hapless applications developers and most of the profits they worked so
hard to earn?

   I hope that the nameless idealists who create the Internet on a daily basis
succeed in their resistance to central control more successfully than did their
computer club counterparts a decade ago.  If they do, there may be no single
successor to Bill Gates.  If they don't, then it's anyone's guess who will
maneuver into that role.  The drama promises to remain gripping.

Howard Anderson
Founder, Yankee Group

George, George, George--what took you so long?

   What is so surprising is that you were so surprised that the Netscape
revolution may supersede the Microsoft revolution.

   Think back!  Fifteen years ago, the office automation hype was about word
processing, but the real revolution was electronic mail.  Today 36 million
Americans get their mail electronically and no one really gives a damn about new
word processing packages.

   Your beatification of Marc Andreessen is a little premature, don't you think?
There are many visionaries, but few company builders.  We in the venture
community intuitively know that there is little likelihood that the visionaries
will turn into the operating managers new industries need.  Other than Gates,
who? What you have proposed (correctly) is that the peripheral becomes the major
focus.  Five years ago, the thought that IBM and the clone industry would become
"value-added resellers" for Intel and Microsoft would have been heresy.  Now you
are suggesting that the Internet will become the major focus--and Intel and
Microsoft will be relegated to that "value-added" focus.  You are right.

   Nonetheless, quite frankly, the Internet needs three years to be ready for
prime time.  Meanwhile, we will go through a sine curve of hype, backlash, then
reality.  Just when naysayers sound and seem logical, the Internet will begin to
fulfill its destiny, and that destiny is to become the most potent force in

   And George, invite your mother to dinner for me.  Of you technobabble family,
she seems the most interesting!

John Perry Barlow
Rancher, Writer, Web Guru, Grateful Dead Lyricist

     I am enormously encouraged by George Gilder's discovery of the World Wide Web.
I'm tempted to stoop to catty remarks about how long it took him.  Lately, he's
seemed almost alone among intelligent commentators in continuing to credit the
myth of the information superhighway--a belief that you can sustain only if
you don't know about the Web or if your perceptions have been profoundly altered
by the reality-distortion fields generated within large content conglomerates,
telcos or cable companies, or the vast cluelessness that is Washington, D.C.

   In fact, anyone using the phrase information superhighway is almost certainly
in the thrall of some large and doomed institution, or is a complete nimrod who
knows only what he hears on television.  Which is, for obvious reasons, filled
with people who believe in it.  They're not going to be roadkill on it, by golly.
They're going to find that on-ramp . . . hey, they're going to be the on-ramp!

   No way, San Jose.  The future of the Industrial Era media megaliths, whether
Disney, Time Warner or, more to the point, Microsoft, will arrive over the Web,
not the mythical infobahn.  The Web will be the end of them.  We are up against a
discontinuous leap as we hit the next layer of interactive complexity, both in
software and in thought.  It has now reached a point where about the only way to
develop all the scripts, scraps and code objects from which the next World Wide
Operating System (WWOS) will assemble itself is to grow them in the distributed,
massively parallel amalgam of minds and processes that make up the Net.

   The Web is alive, and filled with life, nearly as mysterious, complex and,
well, natural as a primordial swamp.  Microsoft is a factory.  You can't
manufacture life in a factory.  You have to grow it in nature.

   This is not so new, really.  Bell Labs' proprietary claims notwithstanding, I
take the first WWOS, Unix, as proof that if enough graduate students type into
enough terminals long enough, they will eventually produce an operating system.
And given enough graduate students, you can be sure that somewhere among them
will be a Bill Joy or a Marc Andreessen.  In accordance with Joy's Law, they
won't be working at Microsoft.

   Furthermore, the Web and most browsers give them a way to crib from one to
another, lavishly and easily.  All that's necessary to see the code that
generated what's on your screen is to click:  "View source." All that's necessary
to get your page employing some of the same innovations is to copy and paste.
This creates an environment of dense autocollaboration.  To see how productive
it's already been in less than two years of existence, get on the Web now.  This
will make the microcomputer revolution look like a good start.

   Gilder's article is a fine thing.  I've been talking about the End of
Microsoft for a couple of years and have felt like a madman parading a ragged
sign.  When it starts coming from such impeccable seersuckers as George Gilder,
it may be time to short a little of that Microsoft stock.

Bert C. Roberts, Jr.
Chairman & CEO MCI Communications

Gilder is dead-on right that hardware and operating systems will become
subordinate to the network;  that the network will, in effect, become everything
and the only thing.  And he's right, too, about the evolution and impact of the
Internet and the resultant focus on content.

   Gilder's allusions to Gates and Microsoft, however, yield a very real
question:  Can they deal with the coming transformation? Doing so will require a
new vision of the future and a new understanding of merging industries, evolving
ways of computing, and some hard new facts.  Here's a key one:  Companies that
don't address and act on the need for collaborative work from a widening
universe will fall hard--and fast.

   Confronting any rival with the market dominance of Microsoft will be very
difficult.  But trying to run in place could be even harder.  The challenge is not
only to understand, as Gilder writes, "that your computer will never be the
same"--but also that it's your company that will never be the same.

Dan Lynch
Founder, Interop Chairman, Cybercash

    Gilder has got it figured out.  His article is dead-on about how the Internet is
the architecture of the future for information dissemination.  Want proof?

   Pouring huge resources into its Microsoft Network and Microsoft Navigator
offerings, Microsoft admits that it does not really know the business model for
how to make money in this new arena, but it is determined to get on the playing
field in a big way.  Meanwhile, Sun is too hooked on hardware profits to make a
new fortune on this round.  No doubt that Java is the right technology for making
the Net be the center of the universe.  Sun should simply spin that group out as
a new company unfettered by dragging along hardware.

   The only fault I find with Gilder's article is his picking Marc Andreessen as
the next Bill Gates.  Marc only has a few percent of Netscape stock;  Bill has
tons of Microsoft stock.  Marc will make a lot of others much richer and he
surely won't suffer, but he will not become a feared/loved mogul because of
Netscape.  Maybe he and Bill Joy should bolt and start their own company? Marc is
still younger than Gates was when Microsoft got its plum from IBM.

   I asked Marc one question when I met him during his brief stint at EIT in
Menlo Park:  What was the central design assumption for Mosaic? He said 45
megabits per second!  I asked why.  He said that it makes it easy to program
because you do not have to develop hairy mechanisms to mask out the slowness of
the network.  Then commercial reality set in when he and Jim Clark got together
and they recentered their Netscape Navigator for the 14.4 kilobit world, thus
giving us the MS-DOS of the Internet.  It is a good start, but Marc was right
about 45 megabits--only it is not just to make the programmer's life easy, it
is to make the consumer's life joyful.

Nicholas Negroponte
Director, MIT Media Lab Author, Being Digital

Once again, George Gilder's got it right.  His guide to the past and future of
the Net is on the digital dime and, as best I hear from my colleagues at the MIT
Media Lab, Java is really hot.  We're betting our house on it.

   The Net, however, is no place for kings.  It is a collective intelligence that
will work in spite of government and any attempts to control it.  No Carnegies,
Rockefellers or Harrimans this time.  The white envelope, George, is empty.  When
Larry Roberts invented the Internet in 1969, his decentralist approach made the
idea interesting for the military, because it was fail-safe.  More than 25 years
later, after the Vietnam and Cold wars, the Net is also safe from dominance of
any kind:  Supercomputer emigres, Unix hackers and Media Lab alumni have moved it
forward (not Windows or Mac users), because they are made of the right stuff.
This is serious media and computer science, not the world of Windows

   Here is where George is wrong.  The next Bill Gates is not Marc Andreessen.
Yes, Marc will do very well.  However, if you were a sucker to pay $ 75 per share
for Netscape at the opening on Aug. 9, keep it to yourself (if I hadn't been
traveling between Rio and Buenos Aires that day, I would have sold short and
made 25 points in a few minutes).  There will be many browsers, hundreds of them.
Today, I'll bet on Niki Grauso's which will come out in 37 languages.  Browsers
are the surface of a much deeper phenomenon.

   Sun Microsystems has it right.  Java.  Why? Because Java is a language and many
smart people will express themselves and invent new applications in that
language.  Netscape is but one awning on the Virtual Boulevard of Digital Cafes.
Java is the coffee.

Andy Grove
President and CEO, Intel

George, George, George--you haven't met a new technology you didn't like, and
conversely you haven't met an older one that you don't think is ready to be
toppled by the new.

   Do you remember when you wrote that "specialized computers-to-a-chip will
become the prevailing product category"? Well, it didn't happen.  Not that
silicon compilers were bad, not that ASICs are insignificant, but the
general-purpose computer that you pooh-poohed has become the prevailing paradigm
of the subsequent six years since the publication of Microcosm.

   Now you predict that the PC will become a peripheral attached to an
all-knowing, all-powerful Internet.  I don't think so, George.  Not that the
Internet won't be important;  I look at it as one of the most important
applications that will be used on my PC in the future.  And that is the
difference between us.  What matters to me is my computer.  I use my computer.  I
am using a word processor to write this response;  I use e-mail to communicate
with hundreds of coworkers from my desk and from my home;  I use CompuServe to
get updated industry news;  I use ProShare conferencing to do business with
partners across the world in real time.  And I use Netscape to check out a new
home page that I have heard about.

   A couple of years from now, I may spend more time in the Net-based
applications than in some of the others, just as today I spend my time
differently among my applications than I did five years ago.  My computer will
evolve and adapt (as PCs have been so good at doing) to the ever-changing mix of
applications with which I work.

   You are wrong in another area.  Implicit in so much of your writing is a
fantasy:  the notion of high-bandwidth communications reaching every PC.  Today my
computer is connected with ISDN to other computers;  it is also connected to LANs
and to an ordinary phone line.  Five years from now, my computer will still be
connected to an ordinary phone line and to ISDN, but also to broadband networks
via a cable modem and to an ATM network to reach other lucky computer users;  and
probably to do some kind of wireless connection.  Ten years from now, it will be
another set of communications transport media.  But it will never be a single
superconnection, because goodness doesn't arrive in a single step.  It comes a
little at a time.

   To be sure, if God erased all computers and networks and forced us to start
from scratch, we would certainly come a lot closer to your utopian view of
broadband to the computer, with intelligence cleverly distributed between
network and computers.  But then, if God erased New York City, it would not be
rebuilt the way it is today.  In real life, both cities and computer networks
evolve, increment by valuable increment.

   The bottom line is this:  the magic of the PC business is that there is a
computer that I can look at, put my hands on and say, "This computer is mine,
and it opens the universe of tens of millions of computers to me through its
screen and keyboard--and whatever connection scheme is made available to me."
Don't bury this magic;  cherish it.

Andy Kessler
Partner, Unterberg Harris

Once again, my colleague and trend-tender, George Gilder, has eloquently set
forth the next wave of innovation.  But in the process, the PC or client device
will not be hollowed out.  Rather than living with fast servers and PCs acting as
relatively dumb terminals, which is the architecture of Mosaic and the Web and
even Java so far, the endgame is instead a network where every device is a
server, even your own lousy PC.  This means terribly intelligent and "fat"
clients, enabling rather than just hanging off the network.

   Moreover, it is much too early to be handing out crowns to the new kings of
the network.  George's trends are dead-on, but the kings are coronated by natural
selection, not by just showing up and claiming the throne.

Stephen Manes
Columnist, New York Times, Author of Gates

Slow down, George!  Sake's alive!  Ma missed signs four and five!

   What sprang to mind upon reading Gilder's lates panegyric was not the rosy
technological future, but a Burma Shave ditty from my youth.  I don't know
Colombian beans about Java, but I've been watching this industry too long not to
reach for my methane detector when somebody announces in the language of
corporate propaganda that "your computer will never be the same." This
particular locution almost always signifies too much time spent amid the
rarefied gases that hover around chairmen and CEOs.

   A particularly toxic dose is to be suspected when the proclaimed agent of
change is an interpreted language.  Interpreted languages have a long line of
champions, not the least of whom is that great Basic patron, William Henry Gates
III (who seems less "Archimedean" than Charles Foster Kanean).  But their magic
has its limits, as Gates himself found out when Microsoft developed its
Multiplan spreadsheet on the interpreted runs-on-everything (but kinda slow)
model and watched the compiled runs-on-DOS-only (but really fast) Lotus 1-2-3
force some massive recalculations at Microsoft headquarters.  Wasn't it only a
few months ago that the press went gaga over General Magic's interpreted
Telescript, which now reportedly has, uh, some problems?

   The runs-everywhere promise of interpreted languages generally manages to get
broken when it hits some Least Common Denominator--say, another LCD, the
liquid crystal display.  If you want to run Gilder's helicopter model on your
crappy little Personal Digital Assistant, you may not have to worry about having
AutoCAD on your hard drive, but you probably will have to worry about having a
hard drive, not to mention a screen with enough detail and contrast to make the
thing look vaguely like the original.  The IBM PC version of Multiplan was no
prize, but it looked and felt great compared with the one that ran on the toy
Commodore 64.  Software is never truly independent of the hardware it runs on,
interpreted programs tend to run with all the speed of a tricycle, and bandwidth
is still a long, long way from being free.

Jesse Berst
Editor, Windows Watcher

Mr.  Gilder's article on the coming software shift is 10% fact and 90% wishful
thinking.  Fact:  Window's dominance will come to an end when it is rendered
irrelevant by a new platform.  Wishful thinking:  hoping the new platform will be
owned by somebody other than Microsoft.

   Gilder's fantasy of a Microsoft-free computer industry might come to pass if
Bill Gates spent his time in his vault counting his money.  He doesn't.  He
obsessively watches the horizon for threats to his hegemony.  When he spots a
danger, he works feverishly to use his current monopolies to leverage his way
into the new arena.  He spotted the Internet danger about two years ago and has
already spent more than $ 100 million to make sure that Microsoft isn't left

   Yes, George, we will all eventually emigrate from the Windows desktop empire.
But when we get to the new promised land, we're likely to find that Bill Gates
has already acquired the prime real estate.

Scott Cook
Chairman, Intuit

Gilder chronicles invention like no other.  He finds the history in the most
recent events.  He finds the significance in invention before that significance
is real or realized.  But Paul Saffo of the Institute for the Future tells how
the innovations that reshape our lives follow their enabling inventions by years
or decades.

   Gutenberg invented movable type.  It took another 40 years for an entrepreneur
named Aldus to assemble what created book publishing as we know it.  Marconi
invented radio exactly 100 years ago.  It took another 25 years for entrepreneurs
in New York to create broadcasting.  It was through broadcasting that radio
reshaped our lives.

   None of us expect it will take 25 years for this chapter of the Internet
story.  But I hope Gilder will tackle the next question:  How will these
inventions change our lives?

Gordon Bell
Research Fellow, Microsoft, Inventor of Digital VAX computer

Telecosm, like microcosm, is another interesting Gilder story.  Like transistors
that follow Moore's Law, bandwidth increases at 60% per year.  The Gilder Fallacy
is that transistors and bandwidth will also be cheap.  This may be true in the
very long run, and I hope to live long enough to see it.  Meanwhile, the memory
cartel has kept prices constant at $ 50 per megabyte and the telephone cartel
has metered outbits in 64 kilobit POTS [plain old telephone service] = chunks
for decades.  For example, full ISDN lines (two voice lines) cost at least twice
POTS.  ISDN is also very inelegant since it makes poor use of the precious last
mile,and is not adequate or scalable.  Only modems, created by a new datacom
industry, have gotten faster.

   My version of the Internet story is at  It is called
"Internet 1.0 (Arpanet), 2.0 (today), and 3.0 (what we need)," subtitled:  "It's
Bandwidth and Symmetry, Stupid." In 1987, I chaired the cross-agency federal
task force that proposed NREN [National Research and Education Network], aka NII
[National Information Infrastructure], aka GII [Global Information
Infrastructure].  Our recommendation was "that an advanced network be designed
and developed to interconnect academic, industrial and government research
facililties in the U.S.," with the plan for a factor of a ten thousandfold
increase in bandwidth by 2002.

   Things have gone according to our "vision" to maintain constant doubling for
the last 25 years.  Thank goodness Berners-Lee created the http://www network,
aka the Web, so that Andreessen could build Mosaic and its viewers;  that's the
serendipity part.  Public funding of research (this time at Illinois's National
Center for Supercomputing Applications) "saved" us again.

   Intermixed in Gilder's piece is another story about a new Sun programming
language, Java, which when combined with the Web, will render Microsoft
ineffective.  If that's the case, then Sun needn't support the anti-Microsoft
legal consortium and should concentrate on toppling it with their products.
However, as a newly hired Microsoft research, anxious to work his way down the
corporate hierarchy to programmer, I hope that he's wrong about our undoing.

   Microsoft is its own greatest enemy if it fails to make every product embrace
the Internet.  In an organization that has to keep up with chaotic hardware
platform evolution, having an external standard and threat is the best thing
that could happen.  Of the 50-plus million Word and PowerPoint users, however,
perhaps 10,000 will see Java as an Internet development tool.  More likely it's a
tool for developing Internet tools.  Word already produces HTML pages and
PowerPoint should have the goal of being the most used Internet content editor.
I'll bet that Windows NT (created by Dave Cutler, who also happens to be the
world's greatest programmer) will become the "standard" server platform, just as
the Wintel platform, like the Gilders use, runs nearly all the viewers.  As for
staying in the center of the sphere--Gilder's right:  Andreeessen's there, but
I wouldn't count Gates out . . . yet.

Charles Mann
Contributing Editor Atlantic Monthly

The vision Mr. Gilder proposes is wonderfully attractive--who could not like
the idea of a future in which nobody has to struggle with the maddening
limitations and incompatibilities of the PC? But the idea of plugging everyone
into the Net for their work and playposes problems of its own, pride of place
perhaps going to security.  The number of Internet crackers and wannabe crackers
is rising;  worse, the very speed and ubiquityof the Net means that each new
break-in technique becomes instantly available to all of them, as the spread of
Root Kit and its analogues attests.  Putting the whole world online will increase
the opportunity for mischief enormously.

   At present, we have two basic means for fending off these guys--firewalls
and encryption.  Put crudely, firewalls are small computers that stand between
you and the Net, scanning for unwanted incoming bits.  (CheckPoint Software
Technologies, in Israel, makes one of the more interesting examples.) In one way
or another, firewalls intercept and read parts of every Internet packet that
comes your way.  If everyone's software is based somewhere "out there" on the
web, the computing power and bandwidth required to filter every screen dump and
rewrite through the firewall is interesting to contemplate.  A worse problem
comes when the information is encrypted--how is the firewall to read it
without decoding it? And if it does decode it, aren't you decoding the
information outside the intended recipient network? It's a problem.

   For this reason, most Net aficionados argue that some form of public-key
encryption is the answer.  Leaving aside the extraordinary reserves of computing
power that will be required to encrypt every bit of info going in and out of
every system in the world--and the even more vexing question of who will pay
for it, a subject on which I am not nearly as sanguine as Mr. Gilder--one has
to worry about the possibilities of breakthroughs such as those reported
recently in Science, where researchers took advantage of the vagaries of quantum
mechanics to reduce the time required to factor large numbers enormously.
Because encryption in its present form depends on the huge amount of time it now
takes to factor such numbers, these advances have dismaying implications.

   We could face a future in which advances in encryption and decoding race each
other, Red Queen-style--hardly a situation favoring the confident use of a
global network.  Until these problems are resolved, they will be a big damper on
the kind of bustling, productive, interconnected future that Mr. Gilder and I
both hope is just around the corner.

Bob Metcalfe
Ethernet Inventor and 3Com Founder InfoWorld Columnist

To refine Gilder's model for how Gates gets dethroned, I'll briefly explain how
Gates made his billions.  Gates came of age just as Grosch's Law about the scale
economies of bigger computers was giving way to Moore's Law about the volume
economies of smaller computers.  Gates saw early and clearly the profit potential
in controlling standards, like the horizontal protocol standards for
communicating back and forth among computers, and like the vertical interface
standards for communicating up and down between applications software and
underlying computer operating systems software.

   I went with horizontal protocol standards, namely Ethernet, which I advanced
by giving it away to all other comers.  Gates went with vertical interface
standards, namely IBM's PC-DOS, which he advanced more shrewdly by giving it
away to IBM but selling it to all comers.  To build the value of PC-DOS, Gates
gathered a huge flock of application software developers and got them committed
to his interface standards.  Then by controlling changes to DOS's interfaces,
upgrading them step-by-step all the way up to Windows 95, Gates was also able to
sell them over and over again in upgraded forms.  With that money, he could
exploit inside knowledge of his own interface changes to pick off one after
another of the larger application developers, like those in word processing,
spreadsheets, presentations and, now, electronic mail and databases.  His only
major failure to date has been in networking, where he's failed so far to kill
Novell.  That failure involved control of Novell's protocol standards rather than
interface standards where Microsoft is king.

   Gilder says that Moore's Law is giving way to Metcalfe's Law, which projects
the runaway value of numerous internetworked computers.  According to Gilder's
point of view, this exponential effect shifts importance from the vertical
programming interface standards upon which the Redmond Rockefeller's fortune is
built to horizontal communication protocol standards with which the likes of
Marc Andreessen might be expected to dethrone Gates, perhaps becoming the
Microsoft of the Internet.

   But Gates, of course, aims to be the Microsoft of the Internet.  And he need
only broaden his tried-and-true interface strategy to include the protocols from
the World Wide Web, now threatening his standards dominance.  Gates has the
advantage of his current Windows monopoly, which he is now trying to leverage
into control of the Web--let's hope our antitrust sheepdogs eventually arrive
on the scene.  Andreessen and company have the disadvantage of coming from the
virtuously open but fractious Unix world.  Unless they can keep their Web
standards act together, they will lose the Web to the Windows Microsoft Network
in the same way they lost Unix to DOS.

   MSN and Netscape will likely shape up to be the major opponents in the coming
Web Wars.  MSN is making its play to replace Berners-Lee's Web standards with its
own under Blackbird.  Microsoft is now giving away MSN browsers like Netscape and
can be expected at some point to sell MSN servers and tools directly opposite
Netscape's servers.  Netscape, on the other hand, with its very busy Web pages is
looking more and more like an on-line service in competition with MSN,
advertising and all.  I must say that this battle is one that my InfoWorld
readers will enjoy, as the fierce competition will drive the Web towards its
full potential.

   Gates doesn't need it, and certainly doesn't expect it from me, but I do have
to defend him.  Andreessen and company have not yet made their billions, and they
now look innocently charming, gregarious and eager, attractive compared with the
shrewd and intense Mr. Gates.  But, I've seen close up what even $ 1 million can
do to people, and frankly Gates handles his billions about as well as it is
reasonable to expect.

Stephen S. Roach
Chief Economist, Morgan Stanley & Co.

Gilder's gushing hype has finally come full circle.  He has come to the key
recognition that there's more to the information age than the alluring power of
Moore's Law.  This time-worn mantra of a vendor-driven rhetoric extols the
miracles of performance but begs the basic question of payback--the
productivity paradox whereby startling advances in computational speed fail to
deliver the value-added solutions that truly matter for society as a whole.  But
now Gilder heralds a new "killer app" in user-friendly browsers and new
"minimalist" programming languages such as Java.

   But don't count on it--at least not yet.  In the fast-track 1990s,
technology must still come to grips with a profound upheaval in American
lifestyles.  For the first time ever, white-collar workers--fully two-thirds of
the U.S. work force--are being asked to up the ante of their own productivity
contributions through longer and harder work, squeezing family and leisure time
as never before in the last 50 years.

   Surfing the Net in a quest for the "best" application--and learning how to
use each of the new tools--takes considerable time.  And that's the rub:
Time-intensive surfing is in direct conflict with the harsh realities of a
time-constrained era.

   In the end, it all boils down to return.  The average U.S. worker who has
toiled under the burden of stagnant real wages for more than a decade will
insist on measurable improvements in personal productivity before squandering
ever-greater amounts of leisure and family time on an open-ended journey on the
telecosm.  However alluring the latest breakthroughs in electronic shopping,
banking, and video selection may seem to be, my guess is that they don't meet
the hurdle rate for the average American.

   That's the crux of the dilemma that Gilder still doesn't get.  The promises of
the telecosm will ring hollow until new means of income generation and wealth
creation for society as a whole are created.  At best, a flexible software
paradigm is merely another means to automate existing tasks.  The big steps come
with applications that truly change the functions of work and leisure.  And they
have yet to be taken.

Larry Ellison
Founder, CEO and Chairman, Oracle Systems Corp.

I believe George has it half right.

   The center of gravity is indeed shifting away from the personal computer and
toward the network--specifically toward the Internet's World Wide Web.

   Despite the Clinton administration's most determined regulatory efforts,
networks are getting faster and cheaper.  Various telecommunication bills pending
in Congress will decrease regulation, increase competition and accelerate this
trend.  Cost and ease-of-use improvements, plus video capability, will make the
World Wide Web as popular as the familiar TV and telephone networks for
communication and commerce, as well as for information and entertainment.
Bandwidth will expand exponentially andso will the quality and quantity of
information on the Net.  The information age will move from dawn to full
daylight.  And our world will be changed.

   But probably not by Netscape:  Wall Street and Mr. Gilder are greatly charmed
by the most exciting startup since 3DO, but it is important to remember that
Netscape's improbable $ 2.4 billion valuation comes without earnings or any
quick hope thereof.  In the absence of any commercial competition, Netscape was
able to achieve a 70% "market share" by distributing its browser for free via
the Internet.  When a clearly superior browser becomes available, people will
replace their Netscape browser just the way they got it--with the touch of a
button.  The most important new feature announced for a future version of the
Netscape browser is Sun's hot new Java programming language, which Sun is trying
to make an open standard.  We're all for it, but you don't get to be the next
Microsoft by distributing Sun software and helping create open standards.  You
get to be Microsoft by creating your own proprietary standard, while convincing
everybody it's actually an open standard.  Fortunately, this trick usually works
only once, so there will be no new Microsoft.  One is quite enough, thank you.

   Marc Andreessen is not Bill Gates.  It's probably possible to be great without
being Bill.  But it takes a lot of work.  Gates is the chairman of Microsoft, the
CEO, its founder and the man who makes every key hiring decision.  Andreesen is
neither chairman (that's Jim Clark), nor CEO (that's Jim Barksdale), but a hired
gun.  Hired, in fact, by Jim Clark.  History generally shows us only one
Rockefeller per industry--the first one tends to get rid of future rivals.  If
a cosmic shift causes King Bill to lose his throne, it will not be filled by
Andreessen or anybody else.  There is no next Bill Gates--in software--at

Steve Case
President and CEO America Online

I have a simple request:  Let's not forget about the customer.

   A "reality gap" exists between technologists (breathlessly enchanted with the
latest and greatest revolutionary developments) and the mass consumer market
(inevitably preferring evolutionary incrementalism).  Invariably, developers have
powerful workstations, superfast networks, and a certain affinity for
complexity.  Meanwhile, tens of millions of people out there struggle with
memory-derpived 386s, painfully slow modems (in the real world 14.4 kbps is
considered pretty fast, and 28.8 kbps is the promised land), and a tremendous
thirst for simplicity.

   Yes, telephone companies are getting more aggressive with ISDN pricing and
marketing, and developers are reaching for plug-and-play solutions, but it's
still too hard and still too expensive.  Yes, cable companies are testing cable
modems, but the tests are going slowly and broad deployment is likely to take
many years.

   Switching gears to software, Gilder embraces the commonly held notion that
Netscape's browser is the standard, commanding a dominant ("70%") market share.
But all of the commercial on-line services now provide Web access, and they have
overnight become the dominant force in the consumer Web.  neither of the two
largest existing players (AOL and CompuServe), nor the most significant entrant
(Microsoft), are Netscape licensees, so Netscape's market share is therefore
zero in this emerging segment.  The already fragmented Web market is likely to
get even more fragmented, creating havoc for publishers and confusion for
consumers.  Ultimately, the services with the largest audiences will drive

   So when I read about the "dynamically portable" Java, linked with the
Netscape browser "standard" and coupled with a "bandwidth tidal wave," leading a
"tele-cosmic advance" towards a world chock-full of custom applets filled with
tasty morsels of "executable content," Exciting, but I wonder how--and when --
it will all come together into a mass market for interactive services.

   People don't buy browsers.  They don't buy objects.  They don't buy bandwidth.
They buy services.  They want access to a broad range of content, packaged and
presented in a friendly, useful, engaging manner, priced simply and affordably,
with a strong underlying sense of community.  At AOL, our aim is to reach tens of
millions of people and help shape this new medium.  We will continue to leverage
technology to do--but we'll also remind ourselves that although technologists
believe the journey is the reward, for everybody else, the destination is what
really counts.


When the responses to an article exceed the original in pith and pertinence, the writer may be tempted merely to step back, point and applaud. However, at the risk of revealing Forbes ASAP trade secrets, I will correct the impression that I am some kind of sappy seersucker who gets his information from doting CEOs, perhaps in the telephone and TV industries. In fact, much to the irritation of the PR firms surrounding large companies, I invariably rely mostly on engineers, programmers and other technologists and consult the imperial suits chiefly enroute to the illuminati in the nether reaches of their companies.

   As Stephen Roach of Morgan Stanley has noticed with some impatience, I also
give short shrift to the kind of macrovisions, strategies and statistics that
float up to the top of economic models.  These data are too light on the
technical details in which reside the angels and devils of industrial destiny.
For example, Roach remains perplexed by the "productivity paradox," the lack of
evident payoff of information tools, the failures of the industry to surmount
the hurdle rates of humbler households, all in the face of some 60 million
worldwide prospective PC sales in 1995, exceeding TV unit sales in the U.S.,
with e-mail passing the U.S. Postal Service in message units.

   Meanwhile, U.S. companies are capturing some 50% of all the profits of the
industrial world and maintaining market share between 60% and 90% in most
leading-edge products, while deploying three times as much computer power per
capita as either Europe or Japan.

   Roach trusts government productivity data and wage levels more than the
demonstrable achievements of U.S. companies and calls the conflict a portentous
paradox.  I say that both the wage and productivity data are largely bunkum,
based on silly deflators that mostly fail to capture the plummeting prices in
the industry, while U.S. leadership in PCs, peripherals, networks and software
is the real payoff.  PC companies alone have generated some $ 250 billion in
market caps over the last decade;  Internet companies will generate trillions.
These bonanzas answer all the questions on hurdle rates and social yields
(though if you want to see some real high-speed hurdling in homes, wait until
cable modems proliferate next year).  The payoff of information technology is the
global leadership of the U.S. economy despite appalling policy mistakes from the
Bush and Clinton administrations.

   Like me, Andy Grove likes to wrangle and I hesitate to deny him his
pugilistic jollies.  Among the new technologies I did not like from the outset
are HDTV, interactive TV, TV shopping, game machines, Japanese fifth-generation
computers.  NMOS, videophones, flash memories, serial Crays, Thinking Machines,
3DO and pencentric PDAs.  But perhaps we can make a deal.  I will concede to him
and Andy Kessler that the use of the term "hollowing out of the computer,"
borrowed from the estimable Eric Schmidt of Sun, is hyperbolic, even misleading
in an absolute sense.  But they should acknowledge that in relative terms, the
balance between desktop and network is shifting sharply.  After all, in five out
of the six PC uses Grove cities, most of the value originates on the network.
Nonetheless, I do cherish the PC;  indeed, using the Internet, I believe it will
displace both the telephone and television over the next five years or so.

   Grove also might acknowledge that Microcosm in 1989 was right on target not
only in identifying Intel as the creative driving force of the most important
company in the world economy, with nothing to fear from the Japanese, but also
in prophesying a huge rise (actually some twentyfold) in the number of new chip
designs (now totaling more than half a million) stemming from the revolution in
chip design tools.  Microprocessors andother general-purpose chips for all their
marvelous profitability, still account for less than half the market.

   Meanwhile, John Perry Barlow and Howard Anderson--as brilliant and
knowledgeable as they are--should not disclose so blatantly in these pages
that they have little idea of what I have been writing about for the last
decade.  For example, they might have taken some note of my constant use of the
phrase "worldwide web" long before the World Wide Web became the industry
mantra, as well as of my long celebration of networks, internetworks, Arpanet
and Darpanet, for the past six years.

   However, I think I am beginning to understand some of their qualms about my
position.  Barlow's hostility to the notion of an information superhighway comes
not merely from the literary scruples that we share.  He seems to cherish a view
of the Net as immaculately conceived, spontaneously reared and capable of
broadband embodiment, bringing down telephony and television, without giant new
investments.  I, on the other hand, believe that a broadband Internet capable of
these feats of creative destruction will require heroic efforts and outlays from
scores of companies, including cable and telephone firms.  Merely to sneer at
them is not a satisfactory response to the challenges of the Internet.

   MCI, for example, is a crucial provider of backbone services for the Internet
and the spearhead of new broadband systems.  Thus I share Barlow's misgivings
about MCI's $ 2 billion investment in News Corp., and I regularly warned the
RBOC's (Regional Bell Operating Companies) against following the Japanese in
"content" binges in Hollywood, gulled by Michael Ovitz, among others.  But I do
not believe that we can have a broadband Web without the full participation of
these crucial infrastructure companies.

   Most of my critics demur at my hyperbolic projections for Marc Andreessen and
for Java.  All I can say is that they are probably right in many respects.  There
will, of course, be only one Bill Gates.  But they tend to overestimate the
business savvy of Andreessen, whom I use to personify a watershed in the
industry.  And they fail to grasp the full ingenuity of Java.  Java combines its
interpreter with security devices, dynamic compilation accelerators, and
familiarity for the C-programming establishment in ways that render it radically
different from other new programming languages that failed to sustain their
hype.  Moreover, even if Java programs are somewhat slower than compiled programs
line for line, these applets will be far more elegant and efficient than the
code-bloated applications now dominant on the desktop.

   As for the trenchant Steve Case, I believe that AOL, MSN and other
proprietary systems will prosper for the next couple of years, giving their
sponsors an Indian summer of success and perhaps diverting them from the
ultimately dominant domains of the Internet.  But then the proprietary regimes
will massively give way to the open systems of the Web.