Subject: Re: BSD-like licenses and the OSI approval process
From: "Chris Travers" <>
Date: Tue, 16 Oct 2007 16:28:27 -0700

On 10/16/07, Lawrence Rosen <> wrote:
> Donovan Hawkins asked:
> > Why would he want to switch to an unpopular
> > license that the OSI lists as redundant?
> OSL is the 11th most commonly used open source license in open source
> projects in the world. AFL is the 12th. (The Non-Profit OSL is too new to
> show up in the list of the most commonly used 20 licenses.) I'm proud of
> those statistics.

I don't think this argument should be about the merits of choosing the
OSL or AFL for new projects.  That is a decision best left to project

> OSI lists AFL as redundant on their website because the self-appointed
> committee that established categories didn't fully understand the legal
> implications of the licenses they were evaluating, and they also didn't
> understand how to solve the license proliferation problem they were
> chartered to address.

I would clearly agree with Larry on the above points.  The current
categorizations on the basis of popularity don't suit anyone well.

> For example, they included the CDDL among "popular"
> licenses because one of its authors was on the OSI committee, but CDDL
> doesn't even show up on the list of the 20 most commonly used licenses. OSI
> listed the Eclipse license as "popular", but it is number 14 on the list of
> 20 most commonly used licenses, below both AFL and OSL.
> Michael Tiemann and Russ Nelson both promised me privately that the OSI
> board will address this categorization problem, but as with so many other
> issues, they haven't yet.

My own view is that the categorization issue is done so poorly I don't
see a way to correct the problem without starting over.  I suppose
this should be made as a proposal on another list  but here is the
idea I would propose:

1)  Functional categories:  Copyleft, Permissive, Other (maybe more,
if the committee sees that as necessary).

2)  Within the categories, Licenses are ranked on the basis of rough
popularity.  (I see three major permissive licenses with different
wording structures: MIT, ISC, and BSD).   Assuming that Mr Rosen does
not intend the AFL to require source distribution to people who deploy
the software for public use, it would be there too (if he does, then I
would suggest placing it under "other").

3) Minor variations of wording ( license vs. MIT License, for
example, or the Intel Open Source License vs the New BSD License)
would be mentioned as footnotes to the main license.

I think in a system like this, we would see that the permissive
licenses have the fewest base licenses with the most footnotes, and
that the copyleft licenses have the most licenses, with the fewest

It would also satisfy Donovan and Rick's ideas regarding a more
modular approach to these licenses.

Best Wishes
Chris Travers