Subject: Re: Open letter to those who believe in a right to free software
From: "Karsten M. Self" <>
Date: Thu, 28 Oct 1999 18:28:10 +0000

"Stephen J. Turnbull" wrote:
> >>>>> "kms" == Karsten M Self <> writes:

>     kms> A natural monopoly, in the classic definition, arises where
>     kms> marginal costs are falling over the entire range of
>     kms> production (output).
> That's an oops.  _Average costs_ need to fall to give natural

That's what happens when I shoot from the hip.  I need to revisit

> That is something I hadn't bothered to notice before.  Software
> (information goods) are not a natural monopoly!  Since it is costless
> (close enough) to transfer the product to a second firm, you now
> have two firms with no fixed cost and constant marginal cost.  Whoa,
> Nelly!  This requires some careful thought.

Well, I'm glad I helped you stumble into this insight....

> The development costs are sunk, and thus not properly accounted part
> of the fixed cost here.  OK, Ben, you're right; this is not analogous
> to any of the standards cases in applied economic theory.
> That doesn't mean we don't have to _cover_ the development costs
> somehow; we do.  The free rider problem still is central.  But the
> standard static definition of natural monopoly does not apply to a
> software product in the obvious way.

This introduces another element to free software:  there's a lowered bar
for introduction of product.  Proprietary product has to be of
sufficient quality to satisfy some *consumer* need.  Free software need
only be of sufficient quality, on first release, to gather further
*developer* interest.  In an uncrowded field this bar is relatively
low.  Where existing products (free or proprietary) exist, the initial
quality requirement may be higher.

I'm stumbling around the idea of gestation costs.  Once development
efforts are distributed, there's a broader allocation of costs, and the
share of costs to benefits for the individual developer becomes more
attractive.  Posit that the cost is divisible but the benefit is
(largely) non-rival.

This also says that your largest marginal impacts on development cost
come early in the recruitment stage.  Adding a second developer halves
the initial cost.  By the time you're at 100 developers, the marginal
cost benefit of adding an additional developer (assuming equal
contribution, natch) is only 1%, and it's possible that the costs of
coordinating development activities would at some point offset these

>     kms> If a rival good can be produced for lower cost or
>     kms> significantly higher quality, and the switching costs are
>     kms> sufficiently low, then the competitve advantages of the first
>     kms> proprietary program is lost (this should sound familiar: MS
>     kms> WinXX vs. Linux or xBSD).
> This is irrelevant to the issue of natural monopoly or not, though.
> The natural incumbent has changed, that's all.

Granted.  It is, however, a strategic concern, and points to two

 - It's in the interest of the incumbant to keep switching costs high.  
   At some point, this will probably impact switching costs between 
   versions of the incumbant's own products, as the incumbant has to
   itself a moving target.

 - Proprietary is ultimately at a disadvantage, all things equal.  It
   can attempt to buy its position (or pay to keep it), but the natural
   advantage is elsewhere.

Karsten M. Self (
    What part of "Gestalt" don't you understand?

SAS for Linux:
Mailing list:  "subscribe sas-linux" to    
 11:10am  up  1:19,  1 user,  load average: 0.00, 0.05, 0.09