some interesting thoughts
The myth of Internet time
Andrew Odlyzko
AT&T Labs
Florham Park, NJ 07932, USA
amo@research.att.com
research.att.com
The bursting of the high-tech bubble is moving the prevailing social
mood from Internet worship to cynicism. The attitude that "the
Internet changes everything" is changing to denigration of the
Internet as the "Citizen's Band radio of the 1990s." Yet just as the
early attitude was overoptimistic, the new one could easily become
unjustifiably overpessimistic. To avoid overreactions, it might be
useful to analyze what propelled the dot-com craze to the ridiculous
heights it reached in 1999 and early 2000.
That this was a craze is becoming ever clearer. A recent article in
Fortune described how in the spring of 1999, "more than two dozen
Silicon Valley [venture capitalists] ... took part ... in one of the
most ludicrous of the dot-com feeding frenzies. Over a span of eight
weeks, [they] vied for the privilege of funding seven pet portal
companies." In retrospect, it is clear that not even one of those
companies could be successful. Yet somehow all those venture
capitalists, as well as the staff of the start-ups, went along for the
wild ride. The press and the general public also willingly and
enthusiastically joined in the celebration of what promised to be a
brave new world. Why were they all so wrong?
A few key ideas appear to have led all these experienced people
astray, ideas that are closely interrelated and reinforce each other.
The most important was "Internet time." This was the perception that
product development and consumer acceptance were now occurring in a
fraction of the traditional time. Closely related to the concept of
Internet time was the idea of "first-mover advantage." Further
support for the dot-com craze was provided by several other notions,
such as "network effects," "increasing returns," "control of open
standards," and "standards lock-in." Of all these ideas, though,
Internet time was crucial. If indeed seven years of traditional
product cycles were now compressed to one year, then anything might
change in the blink of an eye. A company that could quickly establish
itself as a pet portal might be able to exploit the first mover
advantage. The world, propelled by network effects, increasing
returns, and lock-in, would fall into utter (and lucrative for the
start-up) dependence on it for anything remotely related to pets. In
that environment, any notion of due diligence gave ground to the
overwhelming compulsion to grab a piece of "the new California gold
rush."
The fatal defect of this line of reasoning is that it was based on
misleading myths. Network effects, increasing returns, and some other
notions are real enough. As has often been pointed out, they are much
more important on the Internet than in the traditional economy,
although probably not as important as their main proponents argue, nor
as easy to practice. However, the two key ideas of the high-tech
stock market bubble, those of Internet time and of first-mover
advantage, are extremely questionable. As with all myths, they do
have some evidence supporting them. For example, Yahoo!, the first
portal company, has managed to maintain its preeminence ever since.
Amazon.com has also remained the dominant online retailer for many
years (although whether it can ever be profitable is increasingly
called into question). However, the first mover advantage is not
overwhelming. Just consider the early personal computer pioneers,
such as Atari. Where are they now? Even the recent history of the
Internet abounds in counterexamples to the thesis of first mover
advantage. Microsoft's victory over Netscape in the browser wars can
be written off as caused by illegal exploitation of overwhelming
monopoly power. However, consider the fully competitive market for
search engines. Alta Vista achieved a technical breakthrough that
propelled it to a dominant status on the world's desktops. Yet today
it is a distant also-ran.
The main reason the first-mover advantage is much less potent than is
commonly claimed is that Internet time, the dominant theme of the
dot-com bubble, is false. Yes, product development cycles have become
noticeably shorter. This is true not just in software, but also in
such old-economy products as cars. (The Internet is partially
responsible for this, as is Japanese competition of the 1980s.)
However, consumers do not operate on Internet time. Novel
technologies do not diffuse notably more rapidly than they used to.
We have basically just one example of rapid adoption by the masses of
a strikingly new product, namely the browser. The beta version of
Mosaic was released in early 1993, and by the end of 1994 Web traffic
dominated the Internet, as millions of people rushed online. However,
that was an exception. Just consider HTTP1.1. This new browser
protocol is finally slowly moving towards dominance, half a dozen
years since its inception. IPv6, the new Internet protocol designed
to solve several technical problems, including that of limited address
space, is still only being used on a small scale, although it is about
as old as HTTP1.1. Amazon.com is the dominant online book seller, but
after six years of operation, does not even account for 10 percent of
book sales in the United States. These three examples are not
aberrations. Internet telephony is another example that easily comes
to mind. Already in 1995 it was predicted to bring imminent doom for
the established phone companies, but it is barely noticeable today.
Similarly, eBay, for all its successes, has so far had little impact
on the classified ads that sustain newspapers. All these technologies
and companies are transforming the economy, but not in Internet time.
The general rule is that it takes around a decade for even a
compelling new product or service to be widely accepted. This was
already noted in 1965 by J.C.R. Licklider, the "grandfather of the
Internet." In the book "Libraries of the Future," based on the work
of his team at MIT in the early 1960s (and one of the finest
technological forecasting jobs ever done), he wrote:
A modern maxim says: "People tend to overestimate what can be done
in one year and underestimate what can be done in five or ten years".
Of course, that was decades ago, prehistoric by the standards of
Internet time. However, the Internet has not altered this pace much.
In a 1997 study of rates of technological change, I concluded that it
still took about a decade for anything novel to diffuse widely. This
applied even to such compelling new technologies as music CDs and cell
phones. They did become pervasive, but each took over a decade to do
so. This happened even though CDs provided quality superior to
competitive products, while cell phones provided the entirely novel
feature of mobility of communications, which we now regard as
indispensable. Today we are seeing similar rates of adoption of DVDs
as well as digital cable TV, PDAs, and other new technologies.
Some readers might at this stage ask about Napster. Isn't that an
example of a technology rivaling the browser in its rate of diffusion?
It probably is not. Yes, it did spread extremely rapidly among
college students. Within a year and a half of its release, it managed
to account for about a quarter of Internet traffic at several
universities. That is certainly rapid. However, on a wider scale,
its impact so far has been muted. The sales of recorded music in 2000
set records. Yes, Napster is a threat to the music industry, but it
has not driven it into the ground.
The slow pace of diffusion of novel technologies explains the ability
of traditional businesses to resist the onslaught of the start-ups.
Since their customers were not changing on Internet time, the
brick-and-mortar enterprises had time to adapt. This is similar to
what happened in the past. We did not have t-businesses (where t
stands for either the telegraph or the telephone) or e-businesses
(where e stands for electricity). What we had were two types of
businesses. There were those that reshaped themselves to use the
telegraph, the telephone, and electricity. And then there were the
dead ones, that refused to adapt. And that is something to bear in
mind about the Internet. It is a potent communication tool, offering
unprecedented volume, speed, and reach of information. It does change
everything, just like the telegraph, the telephone, and electricity
did. It is just that people do not operate on Internet time, and so
change is slower than the enthusiasts predicted. |
