A Telecosm of Limitless
Bandwidth
varbusiness.com
Technological visionary George
Gilder has his eye on a future of
dumb distributed networks and
intelligent personal devices. What does it mean for
VARs? Plenty. The opportunities, he says, will be
as limitless as the bandwidth.
by Carol Ellison
March 11, 1999
..............
Futurist and technological visionary, George Gilder is one of the
most popular personalities on the high-tech speaker circuit.
Senior online editor Carol Ellison caught up with him recently at
the Xplor Marketspace '99 e-commerce conference, sponsored
by Xplor International in Atlanta, where he took some time out to
talk about his new book, Telecosm, the shifting focus from
hardware to software and services in the VAR world and new
developments in bandwidth innovation that he predicts will have
a significant impact on technology and how business makes use
of it.
Gilder was not always so focused on high-tech. His career is
rooted in the politics and economics of the 1960s, when he
served as a speech writer for Nelson Rockefeller, George
Romney and former President Richard Nixon. In 1981, he
published the best-selling Wealth And Poverty. He was one of
the most often-quoted authors in the speeches of former
President Ronald Reagan and is often considered the father of
supply-side economic theory. In 1989 he published Microcosm,
an exploration of the impact of microchips and personal
computers. That was followed, in 1992, by the publication of Life
After Television in which he predicted that computers and
telecommunications would usurp television's place in the media
empire. That book was a prelude to his upcoming book,
Telecosm, to be published this fall. Telecosm explores
developments in bandwidth, the explosion of cellular
communications and hand-held computing devices, and new
computing environments in which network design is determined
by a search for optimal organizational efficiencies and not--as
they have been in the past--by efforts to overcome bandwidth
limitations. Gilder is a senior fellow of the Discovery Institute and
a regular contributor to Forbes ASAP, and other journals and
newspapers, including the Wall Street Journal
VARBusiness: Publicly, you're probably best known as the father
of supply-side economics. How did you make the move from
macro-economist to technology visionary?
Gilder: During the course of writing Wealth And Poverty, the
discovery that most intrigued me was what was happening in the
semi-conductor industry. I was responding to John Kenneth
Galbraith's New Industrial State, that said all innovation was
increasingly going to come from gigantic corporations and
governments with global dominance and there would be a new
set of monopolies in league with government that would
dominate the scene of innovation, so that small-scale,
garage-scale innovation was no longer possible.
But then there was Silicon Valley. My first introduction to it was
an article in Time magazine, of all places. I became very
intrigued with semiconductors. I've had a sampler by my bed
ever since I got married which was given to me as a wedding
present. It says 'to see the world in a grain of sand.' And it
seemed to me that was what the semi-conductor industry was
doing--describing worlds in gains of sand. So I wanted to learn
more about them.
I went by Ben Rosen's office. He was then head of the Rosen
Electronics Letter. I asked for a free subscription and he said,
"Oh, one rule of a newsletter is 'never give free subscriptions;
everybody has a reason why they should get a free subscription.
You'll go out of business if you give free subscriptions.' But
instead, how would you like to write it?' At the time I knew almost
nothing about semiconductors. That was 1980, 1981, something
like that. I said I didn't know anything about semiconductors to
speak of and he said I could handle it, given Wealth And Poverty
and whatever, so I started writing the Rosen Electronics Letter. I
wrote about semiconductors for a year, did a conference on
semiconductors and met Carver Mead (founder of VLSI
production techniques for microchips and a professor at
California Institute of Technology) who was the most important
figure. I spent months with Carver and ended up writing
Microcosm.
VB: Tell me about Telecosm. What are its messages for our
readers?
Gilder: Well, my initial concept of Telecosm was to write it in the
pages of Forbes' ASAP. I did 14 to 16 chapters and they make
a pretty good book. You can read Telecosm as it is on the Web
[www.discovery.org/Gilder/frbsindx.html] and it's pretty good
reading. It's repetitive to a certain extent but it has a certain
drama of discovery as events occur. Forbes actually printed it up
at one point.
The premise of Telecosm is that essential theme [of change in a
new era] with a lot more on bandwidth: How it's coming and why
it's coming, and the physics of light and lightwave systems, how
you build them and how you look to the technology and not try to
reproduce central office switches or even super packet switches
in optics. Optics doesn't like to do that kind of processing. What
you want to do is multiply wavelengths as much as possible.
Last week, there was just a wonderful development at the Optical
Fiber Conference. Lucent introduced, essentially, a fiber that
increases the available bandwidth for wavelength division
multiplexing optics by 60 percent. If you've seen the graph of the
window of fiber where you transmit light down a fiber-optic
system, there's a mountain in the middle. It starts at 1,300
nanometers, which is the low dispersion point in the fiber and
you have this mountain where attenuation becomes impossibly
high. Then on the other side of the mountain is the 1,500
nanometer wavelength area. What Lucent has done has laid low
the mountain. It's flattened that whole mountain so now you can
go from 1,300 nanometers to 1,620 with no, what they call
'water-spike' in the middle. It's called that because it's the water
in fiber that causes the attenuation. That's a great thing. It really is
a huge deal and it becomes very important when moving the
fiber market to the metropolitan area. Wavelength-based
networking becomes really feasible.
VB: Could you elaborate on your remarks at the Xplor
conference and the issue of our moving away from centralized
intelligent networks and towards a more dispersed model based
on the Internet, as you discuss in Telecosm? How will that impact
Internet computing as we know it today?
Gilder: It's a model where the network in the center has to be as
dumb as a stone. You've got a dumb network in the center and,
increasingly, the intelligence migrates outside to the desktop, to
the palmtop or whatever. This means a change in the model
that's currently emerging for Internet commerce and a lot of
Internet operations. For example, take AOL. AOL is trying to put
quite a lot of intelligence in the middle of the network and in order
to do transactions as an AOL member you have to go through
the AOL conduit, the AOL network. If you go to the same kind of
material by some alternative route, you lose the benefits of being
an AOL member and you lose the authentication that the AOL
servers give you as you perform transactions at the same sites
you might otherwise access through AOL. In other words, AOL is
trying to combine conduit and contact. This violates the
telecosmic model that says the network should be as dumb as a
stone and the intelligence will mostly be on the edge of the
network.
AOL's going to find that, as bandwidth grows, it's going to be
harder and harder to keep up with all the new fiber, WDM DOM
systems being created worldwide. It's going to be increasingly
artificial and a bottleneck to channel everything to AOL's big
servers to create a trusted environment with an ID and a
password and credit and debit access and all those functions. I
believe that the next step in the evolution toward the center of the
network dumb-as-a-stone/fringe-of-the-network-smart is that
security will migrate to the device itself, whether it's the desktop
computer, or a digital cellular form factor, such as the QualComm
PDQ phone that will be launched next year with 2 megabit per
second wireless links to the Net.
Anyway, this is partly a pitch because I'm on the board of a
company called Wave, which I think really has a very good
solution to this problem. It's called Empathy. It puts on a chip, a
hardware implementation of all your memberships. All the
algorithms that you need to perform transactions and establish
your identity in a variety of different environments can be
downloaded into a single encryption, metering, authentication
and credit chip that can be incorporated in a super I/O chip in a
PC, for example, or another device. So all this migrates to
hardware in your own device. This confers a lot of advantages.
For instance, security. It's worth it to hack at AOL if you can set
up a super computer to do millions and trillions of tries and break
into the AOL system to collect lots of credit card numbers. But it's
not worth it to break into a single metering chip that has one
person's information. So the security is artificially exacerbated by
locating [critical data] at a centralized point. If it's distributed
security, it's in everybody's device and it's not worth the trouble to
break into each one.
Once you have security at the end and the network's as dumb as
a stone you can upgrade to a private network connected across
the Internet with end-to-end encryption established in the device
itself, the Empathy chip. I talked about this model when I wrote
Life After Television [1992] and it's just now coming to fruition.
We're ready now with this Empathy device in league with
VeraSign, HP and other companies. We'll see what happens to
it. But the key is keeping conduit and contact separate, and
that's dictated by the model of the dumb network and the
intelligent edges.
VB: But in recent years we've seen companies such as SAP and
PeopleSoft invest a great deal in ERP systems. Now we're
seeing an even greater investment to make the data in those
systems accessible to customers and employees via the Web.
Given the size of those investments, how do those environments
move to the distributed environment you discussed?
Gilder: That's a good question. Those people are ossifying
themselves in huge enterprise systems that will be sclerotic as
time passes. They'll be outmaneuvered by small companies that
are focusing on Java, component software, net-based security
and all those other advances.
But that's facile. There will be a mixture of centralization and
distribution of these types of functions. If you've got 500,000
movies you're renting, you've got to have some gigantic server
and some artful process for managing it and retaining security.
There are going to be different models and what a bandwidth
environment enables is that you concentrate and centralize
where it's optimal. You don't centralize because of bandwidth
considerations. You centralize because of other constraints and
considerations. There are some applications which are
fundamentally large and will need some form of centralized
control. Centralization in the past was dictated by the needs of
memory and the lack of bandwidth and now it occurs only where
it's optimal. Now...it's increasingly feasible to distribute functions
that previously seemed to require centralization.
VB: We've seen an increasing acceptance of Linux among
traditional software vendors, which seems somewhat surprising.
Is there a new business model emerging around this acceptance
of an operating system that is free and no one makes money on?
Gilder: I haven't followed Linux very closely but my impression is
that a set of services is arising around Linux for which people will
pay well. Linux is an operating system. The applications for Linux
aren't free and you'll be able to compile for Linux and other
systems. You can do one package in Java and it will be
acceptable to Linux, Unix, Solaris, whatever operating system
might prevail. In general, you've got to pay for stuff. The people
who are working on Linux will eventually need to pay their bills
and feed their families. From time to time it will be optimal to
give away products in this economy that is so powerfully
governed by the issues of market share and customer reach.
Those constraints mean that many businesses will find it useful to
give away stuff as a phase of their expansion--but I think those
phases tend to add products and new models will emerge. You
have to decide what to give away for free and what to charge for,
and there are going to be a variety of models tried. The Wave
system allows you to rent software by the use. It is a good model
if your particular type of package is an application that's only
used now and again.
Q. What advantages does e-commerce bring to small
companies that may not need or want global reach because they
simply don't have the means to support such a large and diverse
market?
Gilder: Some companies are local, tied to a specific area or
neighborhood in a way that renders it optimal for them to function
locally and not compete globally. However, they can now use
resources from around the globe to optimize their delivery of
services to their own customers. Rather than being restricted to
the resources of their hard drives, they have access to the
resources of millions of hard drives distributed worldwide. It
vastly increases their ability to differentiate products and initiate
new products and services. It makes their company more
dynamic and resourceful even if it's focused on a relatively small
number of customers at a particular time. I think those
companies will be more capable of resisting a global chain that
tries to enter their market if they can command the information
resources that the global chain previously held as a proprietary
advantage.
VB: We're increasingly seeing electronic distribution of products
across the Internet, particularly in the area of software being
download directly to the systems on which they are to be
installed. What other industries lend themselves well to electronic
delivery? Do you see us all ending up there?
Gilder: I think software is a broad category that includes most of
the information products. Clearly publishing is a candidate. The
biggest gainer so far from Internet commerce has probably been
Federal Express. Its revenue has been growing faster than
Amazon.com's for the past year, surprisingly enough. Maybe it's
not so surprising. Federal Express may make more money per
book than Amazon.com does. I'm not sure of that, but it's pretty
close. I think this is an information age and any information
product and service can increasingly be delivered over the Net.
And, encryption and metering certainly increases the number of
products that can be readily transmitted over the Net by making
the collection and transaction process much more convenient.
Communications with end-to-end encryption makes bandwidth a
model. You really have to sell by the bandwidth offered rather
than by the bit, if you can encrypt from one end to the other.
VB: At the Xplor conference there's been a lot of conversation
about data mining as a means of building customer profiles to
improve customer relationships. What do you think about the
notion of customers and their transactions becoming more
valuable to business than the transactions themselves?
Gilder: That's been true for a while. It's part of an advertising
model and part of a model that focuses on the customer and the
relationship with the customer as the key issue. Martha Rogers
[partner in Peppers and Rogers Group (www.1to1.com), a
customer relations management firm] goes deep into this
question. Establishing the lifelong relationship with the customer
is more crucial than selling a particular product. The ways by
which the lifelong relationship is cultivated become more
valuable to overall business success than any particular flow of
transactions income. The transaction and the product become
the means of establishing a relationship which, in turn, will yield a
flow of service income. By that means, the service becomes
more important than the product.
This really is a model that's going to be increasingly significant.
Even cell phones get given away. Set top boxes increasingly
may get given away. Computers may get given away. If
computers have a Wave meter in them you may give away the
computer in order to partake of the flow of transactions that pass
through this computer. So far the computer industry has been a
sort of last ditch upholder of the old model where the product is
key and once you've sold someone a product you forget about
them. This, really, is not in accord with the Internet model. So it
may be that computer companies will start giving away or selling,
very cheaply, the actual machine based on a single chip system.
The flow of income will come from the relationship that's
established by the reception of that product.
VB: In the VAR world, which stands for value-added resellers,
we've seen a pretty dramatic and sudden shift in that direction.
Hardware prices are falling. It's getting tougher to make money
on margins and many of the big vendors are taking their VAR
programs in the direction of software and services.
Gilder: I'm very aware of VARs and I've read your magazine. But
if you look at the companies with large reseller programs, these
are not shrinking at all. The value added is more and more in the
area of integration. As I go around the country, I'm increasingly
struck by the fact that you find Sun and HP integrators who
function just like VARs in the past. But they're adding value at a
different level.
*********************************************************************
|
