Re: Technological Path Dependencies and Inefficiency

From: John Conover <>
Subject: Re: Technological Path Dependencies and Inefficiency
Date: 28 Nov 1999 05:36:11 -0000

Burkhard C. Schipper writes:
> Recently there has been some silly noise by noted economists about the
> origin of the idea. Without doubt much work on this subject has been done by
> Arthur (e.g. 1989, Economic Journal, Vol. 99, p. 116-131) but also others
> were working on it. Anyway, the kindergarten-fight among economists just
> shows how important they value the idea in light of the current anti-trust
> politics.

Additional references:

    "Competing Technologies, Increasing Returns, and Lock-In by
    Historical Events", June 1987, Revised May, 1988, Econ Jnl, 99,
    pp. 106-131, 1989. Also available from

    "Increasing Returns and the Two Worlds of Business", appeared in
    Harvard Business Review, July-Aug., 1996, under the title
    "Increasing Returns and the New World of Business".  Also
    available from

    "Positive Feedbacks in the Economy", Scientific American,
    February, 1990. Also available from

At any rate, Arthur acknowledges that the concept of increasing
returns is not new. On page 4 of the HBR citation:

    Glimpsing some of these properties in 1939, John Hicks warned that
    admitting increasing returns would lead to "the wreckage of the
    greater part of economic theory."

and goes on to argue that Hicks had mis-assessed the issues-increasing
returns does not destroy the general/traditional theory of economics;
it complements it.


BTW, I have had some success in modeling the dynamics of increasing
returns in business and industrial markets with the Gambler's Ruin,
(eg., like increasing returns in business, the player with limited
capital has little chance against the casino, even if it is a perfect
50/50 chance game.) In the traditional Gambler's Ruin, the player's
capital, over time, is a Brownian motion fractal. Adding
leptokurtosis, makes it even worse-and the player/business with the
smallest resources/market share has a very definite disadvantage.

As an example, Linux is supposed to have about 10 million installed
seats; Microsoft's Win9X about 90 million, (depending on who is
telling the story, of course.) So, using those numbers, in the
Gambler's Ruin, Linux's chance of *_EVER_* becoming dominant would be
10 / 90 = 11%, but the duration of the "game" in installed seats would
be 10 million * (90 million - 10 million), or virtually
indefinite-ie., the market is locked in a fairly static
situation. However, for NT, (AKA, Win2K,) NT has about 35% of the
market, Linux 17%, (IDG numbers,) and the chances of Linux dominating
NT is almost 50%, (or MS has a 1 in 2 chance of loosing that market to
Linux,) comparatively, a much more favorable state of affairs for
Linux. But, suppose there were 10 players in the desktop market, each
with 10 million seats, and Microsoft with 90 million. Then, it becomes
a probability that Microsoft will loose its dominant position, 0.89^10
= about a one in three of surviving with its market intact.

So, from this POV, an oligopoly is not sufficient to prevent
"monopolistic" control of the market-it has to be a panoply.


John Conover,,

Copyright © 1999 John Conover, All Rights Reserved.
Last modified: Fri Dec 8 15:06:45 PST 2000 $Id: 991127213624.4966.html,v 1.0 2001/11/17 23:05:50 conover Exp $
Valid HTML 4.0!