Subject: FS a subset of OS?
From: Brian C <brianwc@OCF.Berkeley.EDU>
Date: Tue, 21 Dec 2004 18:36:26 -0800


I'm a 2nd-year law student at UC Berkeley's School of Law (Boalt Hall).
I'm writing a paper for the Berkeley Technology Law Journal on
understanding and enforcing open source and free software licenses. I'm
a Debian user myself and so already familiar with the topic. Here's my

While the FSF lists several licenses that they consider free software
licenses that the OSI does not list as OSI-approved licenses (e.g.,
Cryptix General License, eCos license version 2.0, and the Expat
License, and probably others) one hypothesis I have is that such
licenses were simply never submitted for OSI approval, for if they could
meet the requirements of FSF's free software definition, then they would
necessarily meet the requirements of OSI's open source definition.

The converse is certainly not true. The Apple Public Source License 1.2
was OSI-approved but the FSF had several objections to that license and
it did not make FSF's list of free software licenses. This led to the
Apple Public Source License 2.0 which both groups approved.

But generally, does it sound right that if a license satisfied FSF's
free software definition it would almost certainly receive OSI approval
as an open source license?

I don't think this is a slight to OSI, because the 10 criteria that
OSI's definition provides are probably a better guide to a
license-writer hoping to create a good license than FSF's "four
freedoms". See It seems one
has to tease out the implications of the four freedoms. The conclusions
the FSF reaches through such a process could easily differ from the ones
a license-writer doing their best to guarantee the four freedoms might
reach. (I think the real conclusion is that the text below the four
freedoms is just as important as the four freedoms to the FSF and those
four points may not fully itemize all their requirements. They admit
something like this at the bottom of the page.)

For instance, the complaints about APSL 1.2 were that it had 1)
Disrespect for privacy 2) Central control and 3) Possibility of
revocation at any time. Now I can see how guaranteeing the four freedoms
could be inconsistent with such license terms, but those three things
aren't immediately obvious no-nos if you just read the four freedoms,
and so we've either learned through experience what the four freedoms
really imply or we've learned that the text below the four freedoms
should be read as important explications of the four freedoms that are
just as critical to being a free software license as anything else.

This is something of a side issue though, and I don't expect this group
to speak for the FSF. However, this seemed to me like a group of people
who might have thought about this before, and so you might approach the
question this way:

Can you think of a situation where the FSF would consider a license to
satisfy the free software definition but OSI would not consider that
license to satisfy the open source definition?

Granted it's hard to think about it in the abstract and we can't predict
the future, but you get the idea. Thanks for any comments.

Links relevant to the above: