Subject: Re: Miguel De Icaza interview
From: Tom Lord <>
Date: Sat, 16 Mar 2002 08:40:46 -0800 (PST)

	From: (David N. Welton)
	There is actually no way for us to profit directly from
	Evolution. As soon as Red Hat ships Evolution, and SuSE
	ships Evolution -- Mandrake [...] there is no way to recoup
	the investment that we put into developing Evolution, which
	as I said was two years of development and 17 developers.

I don't know much about the quality or design of Evolution (and I'm
not even from Ohio).  From that perspective, I have no opinion on
whether developing it is a good idea.  But in general, is this an
example of the absence of systematic Free Software R&D figured into
FSB business models?

Why not make the development of such software a paid service, paid for
up-front by the distribution and service companies with a low
(possibly negative) margin plus a limited time profit sharing
arrangement on distributions and services that include Evolution?  The
idea is that those distribution companies (and others) can pool the
cost of development, obtain good technology transfer and participate
in the design.  The Ximian investors still have a high-return
incentive to further subsidize development.  Paying out on the profit
sharing is 0-risk: if Evolution helps generate sales, it's a little
off the top; if it doesn't, there's no pay-out.  Part of the revenue
Ximian sees from profit sharing can pay dividends; part can become
money-in-the-bank to provide a buffer against later projects that
fail; part can replace the investor-supplied part of the development
budget for the next project.

If an R&D company follows that model successfully enough, they wind up
employing an interesting staff and holding a collection of profit
sharing agreements that are or have a high-probability of paying out
-- thus giving the original investors an alternative exit strategy.

Looking downstream at the customers for distributions and services,
especially those customers for whom cost is not the dominant
constraint, doesn't this give them a business justification for paying
per-seat fees or value-received fees, even for free software?  Such
fees can be for a limited time.  They can be in exchange for
participation in the design process.  They don't have to be
implemented by "anti-piracy" contracts.

If that's too complicated, or too risky to actually attract
independent R&D labs, you could average out the profit sharing, do
away with VC funding of R&D, and instead pay for R&D with
moderate-margin subscriptions: that's the idea I was trying to play
with a year or so ago.

Isn't that a more efficient way of expanding the set of available Free
Software than having to buy up every little development company on the
map (for those little companies that aren't being run with costs set
too high)?  It's efficient in the sense that it let's the lab's
funders change over time: R&D focus can change without having to buy
and sell labs; lab's aren't always constrained to warp their focus to
fit the changing needs of a single host company.