Free software business models (was: Evolution copyright assignment: Storm in a teacup)

[Miguel de Icaza <miguel ximian com>]

Richard said:

>     Have you ever think that without that document perhaps they would never
>     released evolution as GPL software? or even worse, Ximian would never
>     developed it.
> 
> There was no real possibility of this, because Ximian was founded by
> free software developers with the specific aim of developing free
> software.  And at first it developed only free software.
> 
> Ximian's practice of releasing proprietary software came later, after
> Ximian accepted outside investment from people who had no particular
> commitment to free software.  In other words, it was ambushed by the
> VC.  This has happened to many free software companies: it is one of
> the main dangers that they face.

No, we were not ambushed by the VCs. 

We wanted a company that developed and created software, not a company
that only did support and services.  I am not a services person I am a
software developer.  My desire remains firmly to create a completely
free operating environment, and one that is not year light away from
the commercial offerings and one that happens relatively quickly.

The policies that Ximian designed were a combination of balancing our
commitment with free software and having to find a way to create a self
sustaining company that would allow us to continue building software.

I still feel that doing a combination of free and proprietary software
is a good way of moving forward.  You see this on Red Hat's services
model, you see this with projects like Mono, Evolution, Qt and Berkeley
DB.

I believed as much as the next person on the free software business
mantras:

	`People will build the software on their own'

	`Free software just happens magically'

	`People do not need to be paid'

	`Software will get written because it will bother someone 
	 enough that it will happen'

	`People will choose us for support and services because we know the
	 code base better'

	`Freedom matter the most to all developers and users'

And although all of the above have an element of truth, they are only a
piece in the puzzle.  I would say that neither Nat nor myself wanted to
sit around for twenty years for the `magic wand' to sort things out.

Instead we came up with some compromises that we felt were appropriate
to create a business that would develop free software.  That would
accelerate the creation of key missing pieces on the desktop; would
productize what we felt was important for the adoption of free software
and at the same time keep the company running and create enough value
that it would bring a return to our stock holders.

The reality is more complex than the one-liners and the feedback loops
systems that we all engage in the free software world.  

We might not have designed the best strategy for the acceleration of
the adoption and development of free software, but we tried one path,
and we were reasonably happy with it.

Today, if I someone asks me about creating a company that will *build*
software, I make sure that they understand that services and support is
not going to sustain that business, at least not in my experience.

But the best way of arguing my points is to go and create new free
software companies and show the world how your model can work.  We did
not find the silver bullet for a pure 100% free software company, in
the meantime we continue to develop a large volume of free software.

The dynamics have changed by Novell's acquisitions of Ximian and SUSE. 
Novell has a big channel and a support and services organization that
can bring in revenue for the free software that Ximian and SUSE
provide.  Today the world is larger and large companies are working
together in joint projects: from Apache, to OpenOffice, to Mozilla, to
Evolution to the Linux kernel and they sell other kinds of services.

Most of the new free software being developed today is -in my
impression- the response from customer needs on large deployments: fix,
patch, improve features that are needed to make a Linux sale happen.
Linux sales that typically have services and support dollars attached
to it.  

Such opportunities did not exist for Ximian and probably do not exist
for most free software startups today.

In short, for making a living, if you are happy as a consultant doing
very limited software development, pure free software companies are
possible.

If on the other hand, you are mostly interested in creating new
software on a short period of time, I would advise a combination of
free and proprietary software.

I would love to be proved wrong, but so far there is no standard
answer, no magic recipe.  There is no simple business model for free
software and it will continue to be an extremely risky strategy for
investment.

Miguel.