[657] in Commercialization & Privatization of the Internet
re: Size of the NREN Market
daemon@ATHENA.MIT.EDU (Craig Partridge)
Mon May 6 02:47:02 1991
To: com-priv@psi.com
From: Craig Partridge <craig@sics.se>
Date: Mon, 06 May 91 08:37:37 +0200
Gordon:
I have two problems with this classification and count...
First, I'd put businesses with serious technical needs as a separate
category -- I believe industrial research labs are a *big* part of the
Internet user population (take IBM, General Motors, DEC, Hewlett-Packard,
SUN, + folks like BBN and SRI all of which provide at least some Internet
access to many of their research employees and you've got a lot of folks).
Second, your total population is only about 600,000 users. I believe
that's too small. I did a little, and not statistically valid study
to try to estimate (roughly) the host-user multiplier for the Internet
(by host-user multiplier, I mean the number by which you multiply the
number of hosts to get number of users). The study was done in preparation
for designing a statistically valid study, which the team I was on never
had time to carry out. Anyway, the results suggested that a multiplier of
between 5 and 10 was appropriate (which was surprising, because I thought
the multiplier would be closer to 1 -- 1 person per workstation -- but
it appears that mainframe computing is alive and well at some organizations).
As of 1 January 1991, there were 376,000 hosts listed in the domain name
system (and this is an undercount as some sites don't allow zone transfers),
which suggests a user population of around 1.88 million.
I confess that I find 1.88 million a staggering number (though I think
the capture-recapture studies for USENET may show a population of about
that size too) and wish we had better measurements.
Craig
