Subject: Re: GPL and trademarks and brandnames...
From: Brian Bartholomew <>
Date: Mon, 16 Nov 1998 01:59:46 -0500

> John Gilmore wrote a while ago that freed software functions by
> reducing the transaction cost of contributing to a project.

> Brian is suggesting that increasing the transaction cost is going to
> translate the benefits of freed software into a proprietary
> advantage.

I think both statements are true, but not simultaneously for the same
developer and project.  There are many pathologies of contribution to
freed software; here are a few:

1.	I'm tried of tripping over something, and I'm going to stay up
	all weekend to add a feature or fix a bug.

2.	I'm a student who wants a hobby, and I'm going to code
	something in my spare time for a few months until I get tired
	of it.  Then I'll drop it.

3.	I make my living working with free software.  I have no other
	means of support beyond the revenue I generate from the
	software I write.  My job is to make a product work for real
	customers, and that includes doing the boring stuff, too.

The first statement may be true for #1 and #2, the second for #3.  The
terms needed to incent core developers and bug fixers may be entirely
different.  It may be possible to give a proprietary advantage to core
developers without turning off bug fixers.  I think this is exactly
what the Netscape license tried to do, but missed:

	The cost to join the core developers is too high and the
	benefits are too low: you must become a Netscape employee, but
	you do not share in setting the direction of the project.
	This is where an increase in transaction costs kills the

	The advantage given is too strong.  Bug fixers don't have a
	perception of good value from the core developers they are
	helping.  Netscape can take the code bug fixers write and sell
	it for arbitrarily large sums without giving them a cut, or
	even giving them the product to use.

Here is a rough outline of license terms to address these concerns:

	If I'm a stranger to the project, I must pay $10 per year per
	unit to use the software.  This money goes to compensate core
	developers.  The core developer company promises to refund my
	$10 pro-rata if it makes more than xx% profit a year, so I
	know my money is going to cover costs and not outrageous
	profits.  Having bought in, I get source, which I can modify
	and redistribute to others who have bought in.

	If I'm a bug fixer, my $10 is waived that year.

	If I contribute enough and the core developers like working
	with me, I may be invited to be a core developer.  Then I get
	paid to continue doing what I'm doing.

What this doesn't address is using part of one project in another
project.  Maybe the core developers could sublicense to another
project under similar terms, but that could lead to spending all the
cycles in the world on license payment computations.

League for Programming Freedom (LPF)
Brian Bartholomew - - - Working Version, Cambridge, MA