Subject: Re: open source policy across the globe
From: "Stephen J. Turnbull" <stephen@xemacs.org>
Date: Thu, 11 Mar 2004 15:46:25 +0900

>>>>> "Tom" == Tom Lord <lord@emf.net> writes:

    >> From: "Stephen J. Turnbull" <stephen@xemacs.org>

    >> May I point out that [Yet Another Bogus Example snipped]?  As
    >> a matched pair (same market segment) you could take that as
    >> "proof" that there are cases where the exclusion model is
    >> necessary, even socially beneficial!  Statistics may lie, but
    >> examples are downright shameless!

    Tom> Why do you think that has anything at all to do with
    Tom> licensing?

I don't.  Why do you think I put "proof" in quotes and called examples
"shameless"?  The point was to debunk the idea that a few successes
prove anything at all, absent analysis of general principles.

    Tom> [economic discussions are] frequently frustrating when they
    Tom> just wildly oversimply the modeling of the technology.

*shrug*  If you try to model the technology accurately, you end up
with an uninterpretably complex economic model.  It turns out that
accurately modelling the technology is typically a red herring.
Stuart Kauffman's _The Origins of Order_ has an excellent discussion
of the use of idealized models where he discusses modeling sigmoidal
response functions with on-off components (pp. 183-189).

It is certainly true that technology matters.  It is never true that
getting the economics right will help you make money; that will only
keep you from losing "too much".  Technology, OTOH, can be a win.
However, it's unlikely that the best technology will make the most
money, or even enough to retire on.  But a talented entrepreneur can
take nothing and turn it into a fortune, even an empire.

    Tom> These days (actually for many years but growing and getting
    Tom> attention): uh.... how do you say this in "economics-speak"
    Tom> .... I think the term is "horizontal integration" is a rising

Actually, it's horizontal "disintegration".  What used to be performed
inside of one organization's command structure now is cooperatively
coordinated by several.

    Tom> concern.  That is to say: the source assets tend to be
    Tom> managed by processes that span organizational boundaries.  At
    Tom> the technical level, distributed operation is a critical
    Tom> feature.

    Tom> At the business level: since "distribution" means the
    Tom> software is installed all along the trans-organizational
    Tom> pipeline, licensing fees are an _impedement_.  That's called
    Tom> a "transaction cost", right?

The licensing fees are not a transaction cost in and of themselves.
If they are the same for all competitors, then they're just an
ordinary cost of business which will be passed on to the downstream
customers.  Of course the customer wants to reduce them and shift
profit away from the vendor toward herself, but this is zero-th order
from the social point of view.  Moving profits around is a wash.

Where the transactions costs come in, which actually do harm economic
performance, are if you have to renegotiate every seat (which could be
approximated by repeated additional purchases if you're growing fast),
or if you're on a per-seat or per-TCP-packet license and your rival(s)
have site licenses, etc.

This is where purchasing managers and lawyers earn their keep.  If it
really is harming the business, then the business that works out a way
to "solve the licensing problem" will win big.  Since "solving"
invariably means more trade, societ