Robert Cringely: Internet use in American homes is still growing, but the pace of that
growth is slowing according to a study released this week by Parks
Associates, a market research firm from Dallas, Texas. Based on a
sample of 1,000 U.S, households, the survey concluded that 42 percent
of U.S. homes have broadband, 22 percent use dial-up, but that 29
percent have no computers at all, thus placing a firm upper limit on
potential Internet penetration.
Not to put too much faith in a
single study, this Parks report generally supports a similar Pew
report from last year. This has to be a kick in the head to pundits
who have seen Internet access quickly becoming as ubiquitous as
electric power, telephone, televisions, or indoor plumbing. It also
does a lot to explain much of the current Internet hype that
emphasizes non-PC applications. By applying a couple Cringely rules
of thumb, we can take these study results far beyond the Parks
analysis and figure out where things are really headed and why.
There are two ways to build market share for a major new technology:
1) by attracting early adopters followed by normal consumers, and 2)
through a generational change where non-users literally die-out to be
replaced by a whole new generation of consumers who are comfortable
with the technology. An innovation can generally reach about
two-thirds of the market through the first method, but to reach near
100 percent market share you need the latter. And that’s the way it
always is, even though we choose to forget that fact when it is more
convenient to do so.
That’s why electricity and telephones, both of which have been with
us for over a century, have almost 100 percent market share. People
aren’t born today who can imagine being without a telephone or
electricity, so of course they have both. The only communication
product class to buck this inter-generational trend is broadcast
television, which grew to 97 percent market penetration in less than
30 years, but that probably confirms my feeling that television was
really perceived by the consumer as an extension of radio, which
would given it a much longer effective adoption cycle.
Today Internet use and cable television use in the U.S. are roughly
comparable at just under 70 percent market penetration. That means
the commercial Internet, which effectively dates from the late 1980s,
has grown at about three times the rate of cable TV, which began in
the late 1940s and took until 1976 to reach 15 percent penetration.
In fact, cable TV market penetration stood at 50 percent in 1987,
about the time the commercial Internet came into being.
So the Internet has grown a lot faster than these earlier
communication technologies, but then the Internet is technically
dependent, for the most part, on some other host network. At the
very least you need to first be a telephone customer to then become
an Internet customer. That makes cable TV, even for its slower
growth rate, actually the more impressive act, since its numbers are
truly ab initio.
Yet it seems obvious to me that while a generation shift will make
Internet access almost universal in another 20 years, the same
probably won’t be said for cable TV, which may well peak and decline
before then simply because there will be other ways to get
television. That’s the distruptive nature of the Internet, which
threatens telephone companies, cable companies, and TV broadcasters,
The result is that each of these industries is trying to poach the
others. As such, cable TV is the very heart of the U.S. broadband
industry even though broadband is what will probably end up eating
cable’s lunch. Telephone companies like the Internet, too, just as
they are also trying to find ways to enter the television business
while their voice business is being savaged by VoIP.
While waiting for the inter-generational boom or bust that is
inevitable, each industry is building-out to maximize the revenue
from its existing subscriber base. Cable TV companies do this by
hawking pay-per-view, digital cable, and video-on-demand. Telephone
companies are starting to do the same. But Internet companies have a
slightly different task, and that’s finding ways to connect more
devices and more device types to their networks. That’s because, as
a stupid network that offers nothing more than bandwidth and a
gleaming smile to its users, the value of the Internet is increased
solely by the number of connections to it. So if you have already
maxed-out your PC connections, it is logical to start connecting
And that’s the motivation — the real motivation — for all these TV
shows and movies suddenly being made available over the Internet. It
is to get more types of devices connected to the net to, in turn,
increase the business value of that network to its nominal owners,
which unfortunately would not be you or me.
This is just the simple motion of tectonic plates — slow but
inevitable, and also irresistible. Against this glacial scene of
course there is still a subtext of local politics and short term
business advantage, but the trends are clear, and here they are:
1. All networks will eventually be subsumed into the Internet, because
only as a single network can their value be maximized. This is kind
of a cock-eyed interpretation of Metcalfe’s Law, which says the value
of a network is the square of its number of nodes. I think using a
square function was Bob Metcalfe going a bit far, and the real value
here is to acknowledge that network unification is a primal technical
urge and — whatever the actual value achieved =- will drive all the
existing networks into a single technology with wired and wireless
2. If all networks are eventually to merge, then all the whining from
this special interest or that is just whining. When SBC (now AT&T)
or BellSouth complain about having to carry Yahoo or Google bits for
free, they are just trying to fool us into paying even more than we
already are for the same network services they would give us for free
if they had to. Sharing the increased value of the larger network is
worth more to these companies than the incremental revenue of
bleeding Yahoo. Either they don’t get this, which is very possible,
or they are lying, which is equally possible.
3. This is a stretch, but it makes sense to me: If the prime derective
here is simply to grow the Net as big and as fast as possible, then
the best way to do that is through the balancing of data loads as
much as possible across the Net. This is contrary to the
client-server model that has dominated the Internet for most of its
existence. Put differently, the major impediment to eventual
Internet hegemony is the problem of scaling client-server
applications. How big a data center do you need before you realize
that no data center is big enough for some applications? Only a
server-server or peer-to-peer architecture makes sense in the long
Here we have what might be three new laws to guide future growth of
the Internet. And they tell us a lot about where things are headed.
It is logical to assume that all devices will eventually be networked
but we will start with devices that are either already intelligent
(computers) or are very important to us (televisions). Any political
or business obstacles in the path of this transition are most likely
temporary. And the only way for the network to maximize its value is
through server-to-server or peer-to-peer networking.
So if your ISP doesn’t want you to share or re-sell your Internet
connection, it is just because they don’t yet realize that such
sharing is really to their advantage. In time they will come to see
it or they will die. If your ISP doesn’t like all that Bit Torrent
traffic, it is just because they haven’t yet figured out that it is
the long term key to their success as a business.
For the transition of television to the Net, then, heavily data
center-based ventures are unlikely to be successful in the long-term.
Google and its distributed shipping containers might be right for
this moment or next, but they can do the math and that math tells
them that pumping tens of millions of simultaneous HDTV unicasts from
a Google server is no way to make money. It is much better to use
Google servers to seed video content, which is then shared by many
peers. This is pretty much the only way the system can scale high
enough to functionally replace and then improve on today’s TV.
The poster child for television 2.0, it seems to me, is a company
like Grid Networks, which might be a new name to you. Like Bit
Torrent, Grid has built a peer-to-peer distribution system. Like
Mike Homer’s Kontiki, Grid has married Digital Rights Management and
a business model of sorts to its Torrent-like distribution system.
But unlike most of the others, Grid (no, I am not an investor but
thanks for asking) most approximates what my Mom thinks of as TV.
Kontiki has been around for a long time, and are definitely first
mover in the market. They adopted a subscription model
(RSS+P2P=Video), but for that the videos have to be downloaded in
entirety, and there really is no "browse, click, watch" functionality
in their system (at least yet). It is certainly better than waiting
2 days for NetFlix at the mail box, but it’s just not quite user
friendly enough in my opinion.
Grid’s system, on the other hand, accomplishes two things from the
end-user perspective: it is point, click and watch; and it is very
very high quality. Using p2p, they can afford to send 1.5Mbps –
2Mbps video over their network because it costs the same as sending
150Kbps-200Kbps video. I was shocked by the video quality,
watching a DVD-quality movie at Starbucks on my notebook computer
with virtually no waiting.
But a successful commercial system can’t count on anarchy, so Grid
has a full Content Management System built into the network (called
Media Vaults) that allows for media uploaded to a media vault to be
monetized and then syndicated with XML to as many vending websites
you as the content owner wants. Mix that with a ubiquitous platform
player and, well, you get the point.
Apple is selling one-hour TV shows on iTunes for $1.99 with 67
percent of that revenue going to the content owner ($1.33), and a
cost of 25 cents for the network delivery leaves them $0.41 for
marketing and operations per video — very thin margins. Reducing
the cost of the video delivery to as low as five cents while raising
the bit rate would seem to be a no-brainer.
Right now, Grid only has about 100,000 clients installed, but that’s
enough for 80 percent lower prices for instant-on video. A new
multi-platform client is on its way; the company will become 95
percent "grid efficient" with as few as 1,000 downloaded clients.
This isn’t rocket science, and there may be other, even better,
solutions to the same problem, but it works, it is available now, and
it could change television forever.