Subject: Re: CNET: The coming open monopoly in software
From: Stephen J. Turnbull <>
Date: 25 Oct 2001 22:32:15 +0900

>>>>> "Zimran" == Zimran Ahmed <> writes:

    Zimran> the economics in this paper are pretty off.

I don't think they're so far off as you claim.

    >> last December, IBM announced that it was investing $1 billion
    >> in Linux in 2001.  Aren't these vendors concerned that
    >> open-source software will undermine the traditional proprietary
    >> software model on which their businesses are based?

    Zimran> IBM is not based on selling prop. software. IBM uses linux
    Zimran> to commoditize away the complementary product (OS) to
    Zimran> reduce the overall cost of their server solutions (the
    Zimran> iron) without hurting their margins at all.

Au contraire.  AIX wouldn't exist if IBM, at least at some earlier
point in its history, didn't think that proprietary software bolstered
its hardware.  They also got their butts burned bad because they
didn't own MS-DOS and later Windows.  (Recommended but heavy reading:
Franklin Fisher et al, _Folded, Spindled, and Mutilated:
U.S. vs. IBM_.  Doesn't cover the PC era, but that's not so different
from the plug-compatible peripheral and mainframe wars.)

IBM, however, is like a sumo wrestler: surprisingly nimble and
tactical for its size.  If they know a strategy can't work they throw
it out and look for something else.

    Zimran> It's an infrastructure, and as such a natural monopoly in
    Zimran> the way most infrastructure is.

And now _you_'ve got it dead wrong.  In economics, "natural monopoly"
refers to globally decreasing average cost, or in a multi-product
firm, subadditive cost.  Eg, Shapiro and Varian never use the term,
AFAICT, let alone apply it to "infrastructure" or "tippy markets".

Thus, software that doesn't exist yet "is" a natural monopoly;
software that already exists "is not" naturally monopolistic at all.
Almost as delicious a paradox as Copyleft itself.

    >> of software economics: as a vendor's business grows, the
    >> average cost of reproducing its software decreases. With

    Zimran> you don't care about average cost when selling software
    Zimran> (or anything), you care about marginal cost.

This is true only for the polar cases of perfect competition and
monopoly.  How and how much average cost matters in cases of
oligopolistic rivalry is a matter of fierce debate.  One thing is
certain: entry and exit are determined by average cost, not marginal

    >> software. At the same time, each unit downloaded increases the
    >> barrier to competition.

    Zimran> so untrue and muddled, not even going to comment.

I don't really disagree about untrue.  However, Fisher et al thought
the notion of installed base as barrier to entry sufficiently
important to devote dozens of pages to debunking it.

And of course the heavy duty network externalities we see today in
office apps etc may change that equation dramatically.

    >> cannot buy, and one they can't attack in a price war because
    >> the competitor's products already sell for nothing. It is
    >> predictable that

    Zimran> Microsoft happily gives away products for nothing when it
    Zimran> helps them.

So?  This is still not a threat if the _normal_ price for OSS is
zero.  Price wars work by forcing your rival's price below average
cost for long enough to force them to give up (exit or return to
cartel discipline).  On the other hand, by that token OSS is a threat
to Microsoft because Microsoft cannot afford to match that price for
very long across its product line.

    >> open-source movement? Aren't they equally threatened? No,
    >> because they are not the monopoly holder.

    Zimran> no, it's because they don't use the OS platform to
    Zimran> maintain lock-in and control standards.

Same thing in different words.

As I wrote before, I don't see glaring errors in their analysis, as
far as it goes.  Some of the things they say are still controversial
among economists, but that doesn't mean they are incorrect to say

Institute of Policy and Planning Sciences
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
              Don't ask how you can "do" free software business;
              ask what your business can "do for" free software.