[ohf-licenses] Affero license doesn't work for hardware

[Terry Hancock]

Greg London wrote:
> So, I finally got a chance to read through the GNU-Affero license
> 
> http://www.fsf.org/licensing/licenses/agpl-3.0.html
> 
> Unfortunately, it doesn't sound like it would be useful
> for a hardware license. Rather than dealing with the
> issue of non-distributed derivatives that the public
> accesses through non-copyright-derivative means
> (derivative on a server, users access output only)
> (derivative in an asic vendor company, users access
> physical hardware)
> the license instead is worded to address the specific
> medium of network access.

Yep. That was my assessment. Doesn't solve the OH problem at all.

> I think it still comes down to something like the
> Apple license.

If you read the specific objections to the APSL, you will find that they
generally come down to specifics about how the license would function in
certain test cases, and not the whole concept of imposing copyleft based
on derivation regardless of distribution.

Some extremists will insist that "manufacturing a product based on a
design" is a "mere use" of a design, and restricting it (by insisting on
distribution of design data (or access to design data) to persons
receiving the products would violate OSD point 6 (or the same point in
the Debian Free Software Guidelines, which is identically-worded):

"""
6. No Discrimination Against Fields of Endeavor

The license must not restrict anyone from making use of the program in a
specific field of endeavor. For example, it may not restrict the program
from being used in a business, or from being used for genetic research.
"""

The theory here is that "manufacturing products and keeping the design
secret" might be a "field of endeavor".

This is of course, an INCREDIBLE stretch of the language.

It is exactly the same objection as was raised to anti-DRM clauses such
as those found in the Creative Commons licenses (where the "field of
endeavor" was "selling DRM-locked versions of software that cannot be
modified and run on a given piece of hardware without access to secret
key data").

You can see how the "field of endeavor" language is subject to abuse.

I would make the counter-argument that "selling binary software without
access to the source code" is a "field of endeavor" too, and so the GPL
is not an "open source" license either. Given that "GPL" is _the_
standard for a "free" copyleft, I hope that constitutes _reductio ad
absurdum_.

> Either that, or try to get FSF to modify the Affero
> license so that it closes the copyright-derivative
> loophole without being limited only to network
> interfaces.

I don't think it's appropriate to use an FSF license, as I've said before.

The nearest thing you could do to the Affero approach is to insist that
hardware devices include a facility by which their design data could be
read from the hardware itself (e.g. the data is stored on a flash memory
chip on a PCB).

Not a good idea, IMHO. :-P

No, the APSL is conceptually the way to go. But it needs to be
re-written for hardware. The copyleft also needs to be drafted in such a
way that it will satisfy most people that the product is "free" -- I
think I've already managed that in my earlier drafts, since I
specifically created options that would satisfy the "Dissident" and
"Desert Island" tests that Debian folks insist upon (essentially they
avoid forced-publication -- allowing for the right to distribute the
source _only_ to recipients of the products, and not necessarily to the
rest of the world). I think that in almost every case, it will be _much_
more convenient to just publish the data broadly, so I'll leave that in
as an alternative.

I think I will go ahead and re-draft my conceptual license without the
whole "design domains" concept, and attempt to frame the "functional
replacement" exception instead (which is probably mostly a recognition
of existing limits of copyright, but -- if explicitly stated -- will
eliminate the future hazard created by exceptional decisions like the
one to make binary software copyrightable.

Although I still think that leaves in some ambiguity, it _is_ a huge
gain in simplicity.

Cheers,
Terry

-- 
Terry Hancock (hancock at AnansiSpaceworks.com)
Anansi Spaceworks http://www.AnansiSpaceworks.com