FDL again, was: My concerns about GPLv3 process

Frank Heckenbach frank at g-n-u.de
Sun Feb 12 03:44:11 UTC 2006

Alessandro Rubini wrote:

> Frank:
> > You mention a case of turning a program into a different form. A
> > valid scenario, indeed.
> > [...]
> > So let's also here consider a case of turning a (FDL) manual into a
> > different form, to have a fair comparison. Examples such as
> > reference cards or posters [...]
> Good point. Actually, It seems to me the FDL has been designed for
> linear evolution of a manual, and not for using the material of the
> manual in other contexts.  While the pool of GPL works allow a "mix
> and pick" approach, this is actually denied by the FDL.  People
> can see this as a problem or not, I personally do.

Well stated. I thought something like this, but I haven't seen it
put so clearly before. Thanks.

FWIW, I generally also see it as a problem. There may be cases where
it might not be an actual problem, but since the license decision is
final (at least if you can't/don't want to relicense later), I'd
have to be sure in advance that the problem could never occur for
any given work before putting it under the FDL, and I see few
possibilities to be sure of this in advance.

"Mix and pick" is a common approach in many fields (where
permitted), including free software (GNU and other), documentation
(e.g., new FAQs/HOWTOs including parts of older, perhaps otherwise
obsolete, documents), even in arts such as music (sampling) and
probably many areas I'm not aware of.

As a personal summary of this discussion (as there seem to be no new
relevant points anymore), I must say I haven't actually learned much
new about the FDL, partly because I was already familiar with many
issues before, and partly because many "arguments" presented were
simply unfounded claims or opinions.

Initially, when the FDL appeared, I was kind of sympathetic to it
and even used it for awhile. I saw the complexities and had some
doubts about possible consequences, but I thought since it was
created by the FSF it would be ok somehow. (And BTW, back to the
original subject, now that I'm no more so sure of it, I'll also
watch the GPLv3 process more critically. For most of my code, I've
used GPLv2 only so far, not "any later version", and now I'm not
sure if/when I'll switch to v3. If there will be a discussion about
the contents of the GPLv3 draft here, I'll probably raise some
points, but after the confusing messages I've seen so far, I'm not
even sure where/how/if the contents should be discussed to make any

Later I learned about the problems with the FDL that had been
discussed at Debian and elsewhere, and I found no satisfying answers
to them -- either explaining why the problematic situations, as
described in many places both abstractly and in concrete examples,
were not actually problematic, or explaining what greater benefit
would be gained in exchange for allowing those problems -- even
though I tried to find such answers. At this point I switched back
to the GPL myself, and my general impression was, FDL without
invariant sections and cover texts is harmless though too complex,
with any of them it's quite problematic.

Now you (Alessandro) described how even FDL without invariant
sections and cover texts can be harmful (because you can't prevent
others from adding some). That was basically the only new argument
to me in this whole discussion (thanks again), and it's made me even
more skeptical of the FDL.

Of course, your scenario does not only apply to your publishing of a
book. It might be interesting to apply it to GNU documentation:

- Someone takes an important FDL manual of the GNU project (Emacs or

- He adds some substantial and useful content to it (tutorials,
  detailed usage hints for certain purposes, whatever -- there are
  probably many relevant topics one could write about).

- He adds invariant sections that strongly disagree with many
  central items of the GNU philosophy (such as advocating "open
  source", disparaging copyleft, misrepresenting the history of free
  software and the important of the GNU project, or to be really
  evil perhaps advocating software patents and DRM -- you know, all
  the stuff we don't like to see, yet see all too often).

- And finally distributes the whole thing (for free or for a
  charge, as he likes).

Now the FSF can, of course, by the FDL, include the new content into
their distribution, but not without also adding the new invariant
sections. What would they do?

- Actually include the invariant sections? Doubtful, as it might do
  them some serious PR harm (even more so if they tried to explain
  they were forced to do so by their own license).

- Politely ask the author for different terms? They can, of course,
  try that (though I've always thought the point of free licenses
  was to avoid having to beg the authors for permission for
  legitimate uses), but the author can just say no, what then?

- Rewrite the new content? This would require a complete clean-room
  implementation under especially difficult conditions (if the
  original source is readily available, the danger of accidental
  copying is bigger, and such accusations will be harder to refute),
  and waste a lot of duplicated effort. (Which shouldn't be the goal
  of free development IMHO.)

- Ignore the new content and do without it? (Which also shouldn't be
  the goal of free development.) This seems to me the most likely
  outcome. What's the result?

  * The new ("evil") author has a superior (by content) manual than
    the FSF distributes.

  * He can include new material added to the FSF distribution
    whenever he likes, to ensure his version remains superior
    without much further effort on his part. (Unless/until some FSF
    contributor adds invariant sections that appall him, which then
    would have to stay forever in the official FSF manual, though.)

  * He can rightfully claim his enhanced manual is released under an
    FSF-approved free documentation license.

  * He can (and indeed must) write "A GNU Manual" on the front of
    his version, if that was the front-cover text of the original

Note: Obviously, this is only meant as a thought experiment. There's
no point in actually doing this to make a point, and the large
effort involved can much better be spent on other purposes. But I
wonder if it really takes someone to do it (which will then probably
be someone with really evil intentions) until the problems are
addressed, rather than denied.


Frank Heckenbach, frank at g-n-u.de
GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)

More information about the Discussion mailing list