[License-discuss] proposal to revise and slightly reorganize the OSI licensing pages

[Chris Travers]

On Fri, Jun 8, 2012 at 11:01 PM, Rick Moen <rick at linuxmafia.com> wrote:
> Quoting Chris Travers (chris at metatrontech.com):
>
>> Nowhere in these do I see any indication that mere inclusion of one
>> work in another creates derivation.
>
> You will not find a simple acid test there or anywhere else.  And yet,
> in my experience, if you read those cases, you will get the pattern of
> the way judges rule.  It's a matter of whether copyrighted expressive
> elements were incorporated into a new work without permission.

Not exclusively.  I cited cases (Lexmark, Sony, etc) where expressive
elements were included without permission but this was held to be de
minimis (Lexmark) or fair use (Sony, Galoob), or allowed on other
grounds.  Those cases are interesting because it is undisputed that
literal copying occurred.  Hence my initial point of copyright only
applying to the extent that the function and expressive elements are
separable (in these cases, I would argue, they were not.  You couldn't
achieve the functions without copying the expressions, so it was
allowed).  These courts went about things in different ways but the
pattern appears to be that copyright is not a legitimate tool for
restricting interoperability of software.
>
> You are not going to find sharp lines about what constitutes creation of
> a new work, versus what is a collection.  However, as I said, you will
> get the pattern and be able to predict fairly well how other cases are
> likely to turn out.

Evidently we read the tea leaves differently.   I suppose it is true
that two observers will always connect the dots differently.

I see the following patterns regarding proprietary software:

1)  Where one party is copying another party's copyrighted works to
their direct financial detriment the court is far more likely to side
against the one doing the copying.

2)  Where the copying party however, is doing so for interoperability
purposes, or functional purposes of interoperability that do not
create new audiovisual works, and do not directly implicate the other
party's sales, these are far more likely to be allowed either via fair
use (Sony v. Connectix) or de minimis exceptions (Lexmark).  This, as
in Lexmark, is a straightforward application of 17 USC 102(b) which
states that copyright cannot be used to own an idea, method, practical
process, etc. (The exact words are "(b) In no case does copyright
protection for an original work of authorship extend to any idea,
procedure, process, system, method of operation, concept, principle,
or discovery, regardless of the form in which it is described,
explained, illustrated, or embodied in such work. ")

>From that I would suggest that the chance of a court holding that the
necessity of linking to system libraries gives an OS vendor copyright
control over all software running on that platform is very low, and
the chance of that being upheld on appeal is effectively zero.  Indeed
I would argue that 17 USC 102(b) effectively prevents using copyright
alone as a barrier to functional software interoperability.  This
seems to me a straight-forward application of Sony and Lexmark as
well.  Moreover I think this is what concerned the court in Oracle v.
Google.

So if you see the Gemini Engine as a piece of software interoperating
with MySQL through a defined API, then static linking seems to my mind
to be creating a compiled work, not a derivative one.  If, however, we
argue that the only functional unit that makes sense is the server
binary as a while, then maybe it is derivative (however in that case,
surely dynamic linking would cure that).  I just don't think it is
settled or clear cut.


>
> No.
>
> NuSphere's product was obviously derivative of MySQL because of the
> incorporation of copyrighted expressive elements into a new work without
> permission.  The technological details are trivia.

I don't think that works. If it did, every compiled work would be
legally a derivative of all components, and I don't think you accept
that either.  If it was, then the work as a whole provision would
mandate that Fedora Linux is violating RMS's copyrights by including
OpenSSL on the same CD as the Readline library, which doesn't work.
If it did the mere aggregation clause of the GPL v2 and equivalents in
the GPL v3 would be meaningless.

Best Wishes,
Chris Travers