Debian policy on copyright

On Wed, 2007-05-02 at 07:41 -0400, Marty wrote:
> The recent media debate over a released HD-DVD key, and resulting DMCA take-down
> notices, got me thinking about a broad range of questions regarding Debian
> policy about "IP rights" in general, and specifically about numbers as
> copyrighted "intellectual propery." Any piece of digital "content" is a single
> number, so the issues raised by the copyrighted key apply to any other type of
> binary data, including images, audio recordings, "ebooks," object code, etc.
>
> Like the banned key, an HD-DVD image is a single copyrighted number. Losslessly
> transcoded copies, as reversible mathematical transforms, would probably be
> covered by the same copyright. Some lossy transforms may also covered, but an
> interesting exception is the HDTV broadcast flag, which applies a lossy
> non-reversible tranform, presumably resulting in a public domain copy. It begs
> the question, at what level of transform lossiness does the copy lose its
> "protection" and become public domain?

One cannot copyright numbers. The issue with the key, is that it is
being used to circumvent a copy protection scheme, which is a violation
of the DMCA.

>
> I have long questioned whether copyright can be clearly enough defined to be
> generally enforceable. To be a useful, it must be very narrowly defined. It was
> initially limited by available technology, and the original purpose was to
> protect the investments of book publishers, indirectly protecting society's
> access to books. Now copyrights have been expanded and interpreted as broadly
> as possible, with clearly adverse effects on society's access to information.
> In one extreme example, copyright was used to keep popular folk songs out of
> Girl Scout song books. The HD_DVD key seems to yet another example of the abuse
> of copyright laws.
>
> I wonder how Debian policy is shaped in this area, and how it's applied to the
> free/non-free designation of various programs? Is there a democratic way to
> determine policy? If Debian lawyers are involved, which country's laws are
> followed? Is there any place where these policies are clearly publicly stated?
>

http://www.debian.org/doc/debian-policy/

-davidc

--
gpg-key: http://www.zettazebra.com/files/key.gpg

>
> The recent media debate over a released HD-DVD key, and resulting DMCA
> take-down notices, got me thinking about a broad range of questions
> regarding Debian policy about "IP rights" in general, and specifically
> about numbers as copyrighted "intellectual propery." Any piece of digital
> "content" is a single number,

s/is/can be represented as/ and I don't think it is the same. I agree
that this question easily becomes philosophical, but sometimes lawyers
have to be pragramatic. Sure, a DVD's content may be representable as a
single, incredibly huge number. But nobody really cares for that number,
it's the result of its interpretation that people want to copy. And if
this result's copyright holder doesn't allow free trading etc., copying
that "number" is illegal. I generally do not have a problem with that.

The issue about the HD-DVD key starting with 09-f9-11 is a different
story since a) the key *is* nothing but a number and b) AFAIK nobody
claims copyright on the key. It's just that this key may be used to
break a copyright protection scheme which is illegal under the DMCA.
(Note that I am not a US citizen but we have a similar law here in
Germany.)

Whether the distribution if this key is really illegal in the US or
other countries is beyond me.

> so the issues raised by the copyrighted key
> apply to any other type of binary data, including images, audio recordings,
> "ebooks," object code, etc.

Debian applies its policy on programs to all files in the distribution.
Images, sounds, fonts, whatever.

> Like the banned key, an HD-DVD image is a single copyrighted number.

NACK, as I already wrote. The key isn't copyrighted and a movie isn't
really a number (it just happens to be representable as such).

If the number was copyrighted, converting to a different format would
yield another, uncopyrighted number -- even when done losslessly. The
other way round, if you could copyright a number with a
song/movie/whatever, you could even go as far as to say that with just
one song you could copyright *every* large enough number you can
imagine, since you can always make up a function that transforms a given
number into your preferred format which then gets interpreted.

> Losslessly transcoded copies, as reversible mathematical transforms, would
> probably be covered by the same copyright. Some lossy transforms may also
> covered, but an interesting exception is the HDTV broadcast flag, which
> applies a lossy non-reversible tranform, presumably resulting in a public
> domain copy.

Is this really true for the US? I have never heard of the applicability
of copyright law depending on the quality of the reproduction. It's just
that copyright owners don't care very much if you broadcast their whole
CD catalogue as 32kBit/s MP3 files. Oh wait, they do. :)

> It begs the question, at what level of transform lossiness
> does the copy lose its "protection" and become public domain?

I don't think this is a very interesting questions since copyright
probably applies until your pile of bits has become unrecognizable.

> I have long questioned whether copyright can be clearly enough defined to
> be generally enforceable. To be a useful, it must be very narrowly defined.

Well, lawyers are accustomed to bend even narrowly defined rules anyway,
so I don't think that's gonna help much. :) I think copyright is not a
lot different from other laws in that it very often takes a close look
by a wise judge to tell what's right or wrong.

> It was initially limited by available technology, and the original purpose
> was to protect the investments of book publishers, indirectly protecting
> society's access to books. Now copyrights have been expanded and
> interpreted as broadly as possible, with clearly adverse effects on
> society's access to information. In one extreme example, copyright was used
> to keep popular folk songs out of Girl Scout song books. The HD_DVD key
> seems to yet another example of the abuse of copyright laws.

ACK.

> I wonder how Debian policy is shaped in this area, and how it's applied to
> the free/non-free designation of various programs? Is there a democratic
> way to determine policy?

Debian Developers may propose and vote in "General Resolutions" on
topics like these, so basically 'yes'. You might find that one
interesting: . It stirred
/very/ hot debates among DDs because it essentially expanded Debian's
license policy to every file in the distribution (but was still called
"editorial amendments").

> If Debian lawyers are involved, which country's laws are followed?

I might be wrong, but I think in the case of copyright, a specific
countries law doesn't need a lot of attention because they all boil down
to almost the same set of rules ("don't distribute without the copyright
holder's consent"). But Debian most probably infringes some patents
which may be valid in some countries. AFAICT Debian's stanza on patents
is that it ignores them unless they are actively enforced.

Issues like the HD-DVD key are avoided. You don't even find libdvdcss in
Debian although it has been distributed by a lot of people without them
getting sued.

> Is there any place where these policies are clearly
> publicly stated?

Dunno.

J.
--
In the west we kill people like chickens.
[Agree] [Disagree]

Marty wrote:
> Any piece of digital "content" is a single number

Which does not imply that any number can be a copyrightable work. At
least I don't recall sending royalty checks to the inventors of the zero.

In this specific case, IMHO the key in question is not copyrightable
because no “original work of authorship” was done to generate it. It was
produced randomly, the same as any other key. You can find a large
number of other such keys (although much, much longer than this weak
key) in the debian-keyring package. Debian does not seek any kind of
copyright permission to distribute those keys. The package's copyright
file says: "The keys in the keyrings don't fall under any copyright."
Another example is the ca-certificates package, which contains a lot of
public keys that are owned by large companies, which Debian does not
have to worry about copyright issues in distributing. The only copyright
claimed on those is a compilation copyright from the Mozilla project.

This particular number was a trade secret (which implies no particular
legal protection) and is currently something that a big corportation
with lots of lawyers is being a pest about people distributing, and that
is the only pragmatic reason that I can see that it won't be uploaded to
Debian tomorrow. Although using it as the version number for a peice of
software would be sorta amusing.

> I have long questioned whether copyright can be clearly enough defined to
> be generally enforceable.

The same can be said about anything from murder to jaywalking.
This is why we have judges who generate case law.

--
see shy jo