Principle behind freedom zero

Valentino Giudice valentino.giudice96 at
Fri Jul 1 16:52:50 UTC 2022

(Comments by an irrelevant stranger).

I am rather surprised by this conversation.

This reminds me of the controversy related to Ethical Source software and
the concept of "open source". This is relevant here, given the discussion
revolves around the DFSG (Debian uses the two phrases synonymously, and the
OSD is almost a verbatim copy).

The reason the ESM got lashback from many, myself included, has nothing to
do with whether the kind of licensing they are promoting is good or bad,
but rather with the fact that they described them as licenses "for open
source", while this was not nearly the case.

"Free software" and "open source" are phrases relied upon by the community,
by several documents, and even by governments to have a stable and clear
meaning. The two movements are different and separate, but the class of
software is quite well defined and nearly the same, with disagreements only
on its very boundary.

It seems to me that something quite similar is happening here.
One question is whether a license which prevents running software to make
proprietary software is acceptable.
A very different question is whether it qualifies as a Free Software

Of course, a free software activist will answer in the same way to both
questions (since the free software movement believes only free software
licenses are acceptable), but they are not the same question.

While the first question is inherently subjective, I think the second can
be answered quite clearly with a resounding "no".
Freedom zero has always been understood to include participating in quite
morally tainted activities. In fact, agreeing on morality is not necessary
to know whether a piece of software meets this criterion.

DFSG guidelines 5 and 6 essentially codifies the same as freedom zero and,
from all that I've seen, they are also commonly understood to include any
kind of activity regardless of morality. This also makes the DFSG (or, more
commonly, the OSD) a reliable standard that can be agreed upon.

Even if we were to erase guidelines 5 and 6 altogether, I'd argue that the
meaning of the DFSG would essentially remain unchanged. The reason is that
the way in which a software license would discriminate against fields of
endeavour is by not providing those working in such fields the freedoms
required by the other guidelines.

I feel that if someone wants to create a movement, or simply to define a
set of licenses, which restrict the freedom to run the software when it is
run for the purpose of making proprietary software, or for any other
purpose that they disagree with, they should feel very welcome in doing so.

But the newly defined software licenses would not be free software
licenses. That is a standard that ought to remain reliable and which
people, communities and governments need to be able to continue reliably
using, or rejecting.

> Would it be a restriction on freedom to run, since downstream could no
> run the program for *any* purpose, they could only run it for the purpose
> compiling GPL'd software?  If not, why not?

The fact that free software *might* be copyleft is literally part of the
Free Software Definition. And the DFSG should be read with historical
context in mind: Debian clearly allows copyleft software and considers it
to be free software.

If you make a derivative of a copyleft program, you are affected by the
copyleft restriction. It really does not matter in the slightest which
program you are running to build such a derivative.

If you have a copyleft compiler, and if the output of such a compiler is a
derivative thereof, then copyleft does affect you when you distribute the
output of the compiler, just as much as when you build a derivative of the
compiler in any other way.

You are still, however, completely free and welcome to use the compiler for
the purpose of making proprietary software, in a company that exclusively
releases proprietary software as long as, of course, what you are
distributing isn't directly the output of the compiler.

In addition, if you are able to alter the output of the compiler such that
it no longer includes significant portions of it, then the copyleft clause
no longer affects you.

Copyleft is the strongest, and most controversial, restriction which can
exist in a free software license. I think trying to extend the meaning of
"free software" by watering it down with other, newer, restrictions (ones
that, by the way, can't even rely on the same legal hooks as the others)
would be a really dangerous move.

I find it also questionable whether such a clause (which, interestingly,
exists in some proprietary software, such as Visual Studio Community) even
works well in practice.
One can build some software components, which they need in their own
software, under a public-domain equivalent license, then use them in
whatever proprietary software they are building. Of course, this can be
mitigated with other newer stronger restrictions, thus getting even further
from software freedom.
I'd also like to hear what the plan is for allowing free distribution of a
piece of software, yet making sure any recipient signs contractual clauses.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the ai-assist mailing list