A discussion of hybrid open source and proprietary licensing models.

[Marshall Van Alstyne <marshall AT MIT.EDU>]

A few thoughts on questions from Matthias:

> before writing such a [new] license, the first question should be:
>
>         Do we in fact need (another) one?
>
> There are existing licenses that work and fulfill the requirements
> of, e.g., the Open Source Definition[0] or the Debian Free Software
> Guidelines[1]. The QPL (qt toolkit) is one example.

Fair question for which I believe the answer is yes.  In a sense, the 
proliferation of licenses is itself evidence of needs going unmet.

Note that one goal is to create not just a hybrid, a proliferation, or dual 
license but a "meta" license -- a clean format with parameters like those 
of a software subroutine.  With one set of parameters you might get BSD, 
with another you might get GPL, etc.

> Another approach is to dual-license. MySQL is perhaps the most prominent
> success story of that approach.

One problem with a "dual" license is the occasional difficulty of combining 
contributions made by either side -- open versus proprietary.  Contributors 
to the open version don't necessarily want their efforts sold, while 
contributors to the proprietary version want to earn a living.  This 
versioning problem represents a kind of forking that we think can be 
avoided by a better license.

A different example is Sun Star Office and Open Office.  The products exist 
side-by-side and it's not apparent that the features of one are a superset 
of the other.

Another take on solving this problem might be a variant on the dual 
licensing approach.  This would be a switch from side-by-side to 
before-and-after.  A contribution could start life as proprietary code, 
becoming open after a short delay.  Contributions made under the open model 
would always be open.

> So the first question I have for anybody who wants to develop a bundle
> of new licenses -- no single license is appropriate for everybody -- is:
>
>         Which goal (and _whose_) is not served by either of these approaches?

One goal is finding an answer to the continuing question "How do you build 
a business around open source?"  I've spoken with several large 
corporations that remain wary of current licenses.  Dual licenses seem 
tolerable but complex and the number of successful companies using them is 
small.

A second goal is to put as much software under the open source model as 
possible, where users can gain free access to it.  Rephrasing, the goal is 
to raise the welfare of those who use software.  This might happen if 
either (i) existing proprietary code could is re-licensed or (ii) new code 
is developed at a faster pace under a better license.

A third goal is to enumerate licensing tradeoffs then to put them into a 
simulation model of a socially optimal license.  The code for this would of 
course be fully open.  A simulation would move a debate over licensing 
tradeoffs into a kind of recursive open science where the design of systems 
for developing code becomes an open social construct in code.

To provide a more detailed answer to this question, I'm also posting a 
three page summary of such tradeoffs in the next message.

>         Is that goal compatible with an Open Source approach
>         in the first place?

Well, this would be consistent with the second goal above :)

MVA