[2127] in Commercialization & Privatization of the Internet
Trickle Down? How About Bubble Up?
daemon@ATHENA.MIT.EDU (Barry Shein)
Mon Jan 27 20:41:08 1992
Date: Mon, 27 Jan 92 20:39:02 -0500
From: bzs@world.std.com (Barry Shein)
To: mfidelma@diamond.bbn.com
Cc: com-priv@psi.com
In-Reply-To: "Miles R. Fidelman"'s message of Mon, 27 Jan 92 19:48:46 EDT <9201280048.AA02852@diamond.bbn.com>
I published an article, "Information versus Processing" in a recent
Sun User Group quarterly exploring these exact issues.
Basically, you have two competing views: Lightweight desktop systems
who's primary role in life is displaying and interacting with
information ("informationist"), versus full-powered peer systems who's
primary role is processing and storing information ("processist").
Each model has its plusses and minuses. For example, the
informationist's model is low-cost to the end-user and very dependent
on competent, centralized systems. The processist's system costs more
(per seat, tho not necessarily overall, funding sources are important
to consider when comparing costs), is harder to amortize, but provides
a great deal of independence.
Centralized systems (used by informationists) can better leverage new
technology on behalf of the end-user, the processist has to wait for
it to become inexpensive and commodity (e.g. juke box storage systems
have been available for at least a decade to those who had $1M to
spend on one, but only lately in the $10K-ish range.)
Informationist's systems tend to be easy to maintain (at the lowest
end we have the dumb or X-terminal where merely cycling power fixes
most problems.) Processist's systems tend to need real system
administration per unit. This is particularly true if the processist's
systems are all network peers (relate that back to this discussion.)
Hybrid systems are possible, but get expensive fast. An informationist
might budget $1500/seat and $1M for central resources. For 100 seats
that's $1.15M. A processist might budget $10K/seat, $1M plus about the
same ($150K) for various semi-private resources (e.g. a few
high-capacity printers.) Hybridizing and we end up with about double
the costs ($1M in processist seats, $1M in centralized resources.)
Processists make a lot of their resources unavailable (e.g. a TB of
disk spread across 1,000 workstations versus concentrated on a
centralized system, it's hard to build a large, say 1/2TB, database
with 1,000 1GB disks on 1,000 workstations, relatively easy on a 1TB
central system.)
Informationist systems breed centralized control and least common
denominators. Processist systems breed distributed control and
customization (everyone can make their own decisions.)
It's important to keep the two models clear in one's mind as people
give answers to questions like Miles raises from each camp and they
sound very different and contradictory.
My money is on the informationists (tho I suspect something like
today's processist's system will be the "cheap desktop access
device".) This model is prevalent in telephony (what's a phone cost?
what's a phone system cost?), television, cable-tv, radio, current
information services (e.g. Dialog), etc.
-Barry Shein
Software Tool & Die | bzs@world.std.com | uunet!world!bzs
Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD