Subject: Re: anti/Law
From: "Stephen J. Turnbull" <>
Date: Thu, 15 Apr 1999 10:12:09 +0900 (JST)

>>>>> "Brian" == Brian Behlendorf <> writes:

    Brian> Doesn't a piece of software overtime collapse from
    Brian> the heavyweight of all the legacy interfaces it has to
    Brian> support?  What escape valve does an open source approach to
    Brian> software development have here?

Distributed development (facilitated by open source and characteristic
of many open source projects, especially in the "bazaar model", but
obviously not dependent on open source) lives or dies on compliance
with protocols (APIs), formally or informally defined.

Open source encourages open protocols; more potential users
(developers) are more likely to see them and comment, leading to
better protocols.  If better means "better targeted" and thus
narrower, proliferation of "little protocols" leads to finer
modularization; the cost of fixing any given "little protocol" is
low (but more modules may be involved, a trade-off here).

Possibly the biggest cost savings in protocol evolution that open
source provides, however, is by-definition availability of the best
implementation as a reference implementation.

N.B.  That's in theory, and economic, not software/technical, analysis
(somewhat informed by reading a few texts on software developement).
Nor do I know how to quantify "modularization"; that is a layman's
view from outside of the modularization of the Linux kernel and a
somewhat informed view of the recent proliferation of interal APIs in
XEmacs.  Two points!  Wow, Euclid says I can draw a trend line!  :-)

Let me know if I came close.  I did read Kragen's list of long-unfixed 
protocols <>
so I know it's not a "silver bullet in the bull's-eye".  :-)

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."