[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GPL & Possible Derivative Work

On 6/18/05, Bernhard R. Link <brlink@debian.org> wrote:
> * Michael K. Edwards <m.k.edwards@gmail.com> [050618 01:30]:
> > Plausible isn't much help here.  Courts mostly only ask "is it
> > plausible that the resemblances are coincidental" when evaluating
> > claims of copying of non-literal elements from a rejected screenplay,
> > film scenario, creature sketch, etc. into a fictional work (such as a
> > movie) made later.  That sort of "constructive availability" test can
> > weaken a claim of infringement at the factual copying stage, but
> > usually only if it's demonstrably improbable that the later author
> > even knew what was in the work alleged to have been copied.
> I do not know about you, but I prefer adhereing to the law than to
> break it in a way you have the best chance of getting away.
> As was stated before, proper clean room implementation makes it impossible
> to have undeliverately copied something protected. With all the
> specifications made you also get a good documentation of what actually
> happened. Noone can force you to make it more 'different' to the orginal
> (and even less to make less verbatim copies, because there are no
>  verbatim copies, only things ended up to be verbatimly the same).
> The only possibility to attack this scheme it to persuade to court
> you did not do so and lied and produced specifications and
> documentations and the like to disguise the facts. It is not even
> your word against someone else's word, but your word against someone
> else's wild guessings.

The difficulty is that specification-followed-by-reimplementation
doesn't have any effect on whether you have copied expression from the
original.  Copying is judged on the facts of resemblance, not on the
process by which it happened -- except in the quite rare case
(generally a work of fiction) in which the author of the later work
demonstrably didn't know much about the original and the resemblances
are credibly coincidental.  That simply doesn't hold water when
reverse engineering is in play.

As I wrote, the procedure you describe may be of some use in
bolstering the affirmative defense that the resemblances were dictated
by ideas, methods of operation, and compatibility requirements.  But
the "clean-room" technique was largely conceived as a defense against
claims of trade secret misappropriation, not against copyright
infringement.  Protocol cloning _is_ a deliberate effort to copy the
ideas and methods of operation without running afoul of copyright in
the expressive content of the original, and should proceed accordingly
-- which means producing evidence, not that resemblances are
coincidental, but that they are _necessary_ for interoperation.  And
that usually means comparing the two at implementation level and
purging any resemblances that you can possibly reimplement

- Michael

Reply to: