Subject: Re: Vendor canonization
From: "Stephen J. Turnbull" <>
Date: Wed, 11 Aug 1999 11:19:39 +0900 (JST)

>>>>> "Brian" == Brian Bartholomew <> writes:

    Brian> Let me make the question a bit more general.  Do you
    Brian> recognize any quantitative determination of the percentage
    Brian> of value shipped to users by software business practices,
    Brian> other than libre/non-libre?

I don't think so.  I think we can characterize the _commitment_ of a
firm to a business practice, by looking at the sunk costs (in
training, lost revenues, reputation, etc) needed to implement that
practice.  Aside from the usual issues of accounting practices, that's 

But "share of value shipped to users" is hard to allocate.  For example,
suppose a consumer gets value $1 from product A, value $1 from product
B, and value $3 from using them together.  If A is free and B is not,
is the value from free practices 33%, 50%, or 67%?  Suppose that C is
free software and a substitute for B.  But using A and C together is
only worth $2 to the consumer (no synergy).  Isn't it plausible that
in that case free software only contributes 33% to the A + B
combination, that B is responsible for the synergy?[1]

But implementing such comparisons is very difficult.  And stepping
back from defined products with observable market prices to the
influence of their "business practice components" is going to be an
order of magnitude harder.

It can be done; labor and consumer economists do such "hedonic
pricing" all the time.  It would, however, be rather hard to do for
customers which are firms rather than workers/consumers, because it
would be impossible to preserve a reasonable amount of anonymity.
(Eg, the only way to get access to the Commerce Department's raw data
on firms is to go to work there and sign a strong non-disclosure
agreement; those computers are not hooked to the Internet.)  Thus
firms would be very unlikely to give good detailed data.  (Of course
one could do it once as an academic study; what I think would probably
not be possible would be a semi-annual study of delivered value for
wordprocessors in Consumer Reports.)

You can of course do what Consumer Reports does, and define many
objectively measurable categories, measure them, provide all the
statistics, and define what you think is a reasonable index.  Ie,
benchmarking, like PC Magazine etc already do.  But this would not
really be beleivably accurate for "business practices", and would be
rather hard to connect to "delivered value."

[1]  It is possible to cook this example so that there's no internally 
consistent conclusion but that all the synergy is due to product (or
practice) B, and of course there's a symmetric recipe to guarantee
it's all from product (practice) A.

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."