Subject: Re: Free Software in Embedded Systems (was Re: Studies)
From: Bob Weiner <weiner@wave.altrasoft.com>
Date: Tue, 25 Nov 1997 00:32:43 -0800

>>>>> "RN" == Russell Nelson <nelson@crynwr.com> writes:

   RN> Niels P. Mayer writes:
   >> I love Linux and GNU and use it on my own systems all the time. But
   >> people in the financial/banking/point-of-sale arena  just look at me
   >> funny whenever I bring it up and ask "what's that?"...

   RN> One of the things I got paid to do late last year was write a cs89x0
   RN> Linux Ethernet driver for a Schlumberger POS gas pump.  There's money
   RN> in them-thar free software embedded systems!

This tends to confirm our experience at Altrasoft in free software markets
that to obtain standard business revenue models from the software one must
obtain payment up-front to pay for all of your overhead.  In many contexts,
this is hard to do because large companies such as Schlumberger are not used
to paying up front for the development of external technologies unless they
are in dire need of them and can't find anything at all comparable.

From an economic perspective, this puts many potential free software
businesses at a severe disadvantage relative to proprietary ones which can
afford to get much of their money out the back-end by selling limited use
copies of their software and which have greater up-front investment
opportunities (because of the control of their intellectual property).

From a commercial perspective, I think a much better model than the GPL
would be to distribute source code and to allow free use in academia and
non-profit research institutions but to allow vendors to require payment
for any commercial/industrial use of the software.  This would produce
much more software where people could learn from and utilize the source
to adapt it as they need it but would still allow vendors to assure
themselves and their investors that they would be able to make dollars
from ongoing sales investments.

As a simple example of the problem, all of the clients that we have for
a particular product released under the GPL on UNIX have told us that
they want the product on NT (without any sales effort on our part targeted
at that platform).  Yet none of them have been willing to pay anything
up-front to pay for the porting work.  We tend to believe that they would
rather spend two or more times more money on the software once it is ready
for delivery than to support the work and alter their software purchasing
models.  So you have a chicken-and-egg sales situation that is good for no
one.

This is not to say that up-front funding fails to emerge in all circumstances
but just that it is too limiting a model to cover all of
commercially-applicable software that needs to be developed.  So far the
solution in the free software world seems to be largely that people work
on it for the pure joy and public exposure without any economic incentive.
That model of course will never scale and the lack of man-hours relative
to those spent on proprietary software will continue to allow proprietary
vendors to produce many more mass market (and thus mass mind-share) solutions
than the free software community, despite all of the other levels of benefit
that free software can provide to users.

Even Cygnus now sells proprietary software and I believe (though Michael
T. would know much better) makes much more of its money from up-front
funding of custom ports (where the vendor needs the toolset to support
developers on a new chipset) than from support service contracts on the
back-end.

This is not how I would like to see things evolve.  I believe in open
software solutions.  But it is the effect that I see emerging from
the application of GPL solutions over public domain or other
business-friendly licenses.

Regards,

Bob