Perhaps you are back for new evidence of the Internet's coming catastrophic
collapse, or to see whether I've finally recanted the unpopular prediction
made here in December. (See "Predicting the Internet's catastrophic collapse
and ghost sites galore in 1996," Dec. 4, 1995, page 61.)
Well, this week the collapse is foreshadowed in market research I've
extracted from International Data Corp., InfoWorld's sister company, of which
I'm a director.
At the end of 1995, IDC analyst Frank Gens polled many of his 300 colleagues
in 40 countries before predicting that during 1996 the Internet will move from
"intoxication" to "hangover."
He ventured that 20 percent of Fortune 500 companies' commercial World Wide
Web sites will either stabilize or be closed this year.
Gens anticipates disappointing financial results among Internet companies,
the Internet stretched to the breaking point, and high turnover among
underwhelmed Web users.
Frankly, Gens doesn't paint my picture of catastrophic collapse, but close.
Another Gens prediction is that Internet appliances will arrive in 1996, not
for $500, but for $100 to $300. He expects a Sony Playstation with Internet
access for $249.
Beyond 1996, he foresees free Internet appliances distributed by on-line services.
Gens, like me, is optimistic about the Internet in the long run and
subscribes to the IDC Internet Commerce Market Model. It predicts that in the
year 2000, more than 200 million Web users will conduct between $150 and $200
billion of Internet commerce. This is up from 8 million active Web users in
1995, which in turn was up from 1 million in 1994.
It's fair to ask whether these predictions are optimism or cluelessness, and,
oh, by the way, where is my collapse?
Out of its model, IDC produces an index of commercial opportunity on the Web,
for estimating returns on investments in Web pages.
The IDC Web Index is roughly total Web hours divided by total Web pages times
Each of the Index's major metrics is in turn dependent on a spreadsheet of
parameters, such as numbers of personal computers, modems, average Web hours,
and what IDC calls its Surf-to-Buy Ratio -- the fraction of users actually
buying on the Web (19 percent in 1995, climbing to 24 percent in 1997).
Well, does the IDC Web Index indicate Internet collapse in 1996? Not exactly.
IDC uses the terms "shake-out" and "backlash" to describe a modest decline of
its index at the end of 1996.
The decline is due to the growth of Web pages exceeding the growth of Web
users, "a classic scenario of a start-up market, where pent-up demand
generates investment in capacity that comes on-line just as early market
The IDC Model and Index are fascinating attempts to escape the cluelessness
that enshrouds commerce on the Internet. I'm sure IDC is in for considerable
For example, when IDC says $200 billion of Internet commerce will be
conducted during the year 2000, what do they mean? Is it just counting
transactions fulfilled over the Internet using some form of micromoney, such
as the purchase and delivery of a software component? Or transactions ordered
over the Internet but paid and/or fulfilled off-line? Or transactions begun
during on-line browsing but completed off-line? IDC answers all of the above
-- which is, I think, too broad.
When IDC divides Web hours by Web pages, are they coming up with any useful
measure, or does commercial activity not care if there are gazillions of Web
pages that hardly anybody visits? Should pages for the Index include just Home
pages -- published URLs -- or should it count all reachable pages?
Finally, IDC's bottoms-up estimates of Web hours begin with the huge database
that IDC has on the installed base of PCs.
What if $100-to-$500 Internet appliances start taking off during 1996? They
would make no difference during the collapse, but at those price points, IDC's
numbers leading to Web hours would be gross underestimates.
IDC is in Framingham, Mass., and can be reached at (508) 872-8200, and,
unless they've shuttered their Web site in preparation for the Internet's
impending collapse, at __http://www.idcresearch.com_.