[2531] in Commercialization & Privatization of the Internet
Growth and Media
daemon@ATHENA.MIT.EDU (Dick Karpinski)
Fri Mar 13 21:08:14 1992
Date: Fri, 13 Mar 92 18:05:37 PST
From: dick@ccnext.ucsf.edu (Dick Karpinski)
To: com-priv@psi.com
Whenever a new medium for publication arises which offers new opportunities
for low cost dissemination of information, we expect to see an explosion of
that information. We all look back to the invention of the printing press
as the archetypical example, but that explosion took decades. Now it is
happening in months and years.
To start with, I am going to assume that there exists a large body of
potential consumers of information who have (at roughly zero marginal cost)
access to the Internet and to the publically accessible data held there.
That is, many people can already access such network resources as WAIS
servers, campus information systems, FTP archives, and library catalogs.
To make things a bit more concrete, let's consider a typical technical
report on A4 or 8 1/2 by 11 inch paper in double spaced typewritten form.
A page of such a document holds about 2,000 characters. One megabyte of
such data represents about 500 pages. It would thus cost about $25.00 to
make a single photocopy at 5 cents per page.
If we instead bought (a part of) a hard disk connected to an Internet
connected computer with our money, we could publish that same information
to the entire Internet. It would cost the same (marginally, not counting
either the network cost nor the cpu cost to handle anonymous FTP) if the
hard disk cost $25/megabyte. It used to cost more for such hard disks,
but now the cost is below $2/MB and falling quickly. Thus, the cost for
Internet publication is down to 0.4 cents per page. This looks to me like
the kind of situation which would lead to an information explosion.
Now take a look at the growth factors in computing and communication. For
about forty years, the bang-for-the-buck in solid state electronics has
been improving at a rock-solid 23 percent per year. This means that you
get twice as much for an inflation adjusted dollar every three years.
That is very nice. It beats inflation and goes in the right direction for
those of us in the high-tech fields.
But wait! It gets better. In RAM (especially DRAM) which is a high profit
and high volume business, they've been doing so well that they skip a mere
factor of two. It wouldn't last long enough. Chip vendors have been
providing a factor of four every three years for at least three decades.
That was fast, but check out the workstation market. This has been quite
incredible. For about the last decade, we've seen the improvement in such
machines occur at a factor of two EVERY YEAR! This represents a factor of
eight every three years, and makes us wonder (some, with gaping jaws) how
long they can keep it up. They shot past the mini-computers in the first
few years and have just passed the main-frame machines in the last year or
two.
That is what's been happening in computing. What is the communications
world like? I'm not sure that these numbers are really accurate and also
meaningful, but the Merit folks have been reporting a growth in packets
carried on the NSFnet backbone of 15 to 20 percent per MONTH! That means
and annual growth of a factor of 5 to 7 fold. In turn, to compare it to
the computing growth, this represents a factor of 125 to 350 per three
years. That sounds like an explosion to me!
That is all very nice if you happen to be employed by one of those firms
or Universities which is privileged to have Internet connections, but who
can afford their own T1 line at $500/mo plus CSU & DSU startup costs?
I've only just noticed that there may well be a good answer for such real
and hearty connections. At least one vendor has a product of interest in
this situation: NCR has a PC board for around $1000 called WaveLAN.
WaveLAN is a wireless, spread-spectrum modem handling about 2 megabits
per second. It uses the 902-928 MHz *unlicensed* band to make connection
to its mate a few hundred feet away. On the other hand, if you add a
small aimed dish antenna to a line-of-sight partner, similarly equipped,
the distance can often be stretched to four or five miles. With gear
like this available now (and for a year or two already) we might be able
to sustain that growth in traffic for another decade or two. This could
make direct access to the Internet rather more widely available, I think.
Dick Karpinski (dick@ccnext.ucsf.edu)
