[okfn-discuss] GNU GPL non-conformant with Open Definition?

Sun Oct 12 10:01:17 UTC 2014

On Sun, 12 Oct 2014 02:00:05 -0400, Mike Chelen <mike at opensci.info> said:

    > For example: 2.2.5 Source
    >> 
    >> The license may require modified works to be made available in
    >> a form preferred for further modification.

This doesn't make sense. The intent seems to be to allow copyleft
clauses but it doesn't work because data is not software.

A common thing to do with data is to transform it from one form to
another. Maybe somebody has published a microsoft excel file and I
think it is better to have a comma separated value file. If they say
that excel is their preferred form then I have to somehow publish my
modified csv file in excel. That would be irrational but as far as I
can tell would be compatible with the Open Definition.

This kind of restriction makes sense for software. It is unusual to
translate a body of source code as a whole to a different language,
and doing that would normally be a completely different program. With
the GPL this is intended for situations like yacc grammars which might
normally be translated to C. Distributing only the generated C file is
counterproductive because to meaningfully be able to modify it you
need the original yacc source. This is very different from when the
main activity is translating some data from a less useful form to a
more useful form.

With data, the whole point of doing that kind of transformation is so
that the more useful form can be more meaningfully used
modified. Requiring the inverse transformation for derived works is
counterproductive and diminishes freedom because of the extra burden
on the person doing the work for no benefit.

Furthermore consider someone who takes a comma separated value file
and transforms it to linked data. It could reasonably be argued that
doing so imposes technical restrictions because the tooling required
for working with linked data is more complicated. So here we could run
afoul of,

    2.2.6 Technical Restriction Prohibition

    The license may prohibit distribution of the work in a manner
    where technical measures impose restrictions on the exercise of
    otherwise allowed rights.

Perhaps this argument is clearer if we substitute "ASN.1" for "linked
data". Does a technical barrier amount to a restriction?

I fully understand that these are pathological examples. This is just
to illustrate that the Open Definition can be used to perversely argue
that a license that says derived works must be distributed as excel
files and making linked data is not allowed is in fact an open
license.

I think that the best that can be said here is something along the
lines of,

    The license may require modified works to be made available in an
    open format and under an open license.

    The license may prohibit distribution with technical measures
    whose main purpose is to restrict otherwise allowed rights.

Cheers,
-w