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

Software copyright (was: Proposal -- Interpretation of DFSG on Artificial Intelligence (AI) Models)



At 2025-05-08T23:06:12-0500, Richard Laager wrote:
> On 2025-05-07 07:05, Simon Josefsson wrote:
> > Another example is the practice to drop copyright years from
> > copyright notices.  Some commercial players do this because they
> > save developer time, and believe that the likelyhood that a
> > copyright claim will have commercial effects depending on the
> > presence or lack of the copyright year is low.
> 
> Basically zero, right? Because (in at least 181 countries), copyright
> attaches automatically, regardless of whether a notice is present at
> all.
> https://en.wikipedia.org/wiki/Berne_Convention

As I understand it, the U.S. did not ratify the Berne Convention until
1989, and its provisions were not retroactive (revisions to U.S.
_statutory_ law, however, such as the Sonny Bono Copyright Term
Extension Act of 1998, have frequently applied retrospectively, perhaps
creating the illusion that treaty ratification did so).

For example, works that had already fallen into the public domain, or
failed to acquire copyrighted status in the first place due to a
defective or absent copyright notice, were not recaptured by the
copyright regime through either the Berne ratification or the Bono CTEA.

To my knowledge, defective or absent copyright notices no longer prevent
the attachment of copyright in the U.S., but they could prior to the
Copyright Act of 1976, which took effect on 1 January 1978.

At least some of the original Bell Labs Unix source code (kernel plus
userland programs), some of which dated back to 1971 (in PDP-11 assembly
language or B, since the C language did not yet exist--but particulars
are hard to know since much of this data has been irretrievably lost),
was affected by this provision because, as I understand it, AT&T
dithered over its legal strategy, and decided at some point that the
Unix source code was a trade secret, and therefore--the reasoning may
have been fallacious here, and/or there were conflicting or shifting
directives in AT&T leadership--copyright notices were not attached to
some files, or in some cases might have been deliberately removed.  (The
question "can something be both a trade secret and copyrightable?" did
not have as clear an answer in the 1970s as it does today; as far as I
know the issue wasn't settled in the U.S. until _Salinger v. Random
House_ in 1987.[0])

The later infamous _USL v. BSDI_ lawsuit turned on this fact, as well as
AT&T's practice of "unclean hands", placing false notice of its own
copyright on materials that originated wholly within UC Berkeley.

In summary, it is possible for some (very old) source code to be in the
public domain in the United States even when not deliberately given that
status by its author.

What _has_ never happened, as far as I know, is that no computer source
code in any form has ever _aged_ into the public domain.  What protected
"Steamboat Willie" until 1 January 2024 also protected everything from
Alonzo Church and Alan Turing forward.

Admittedly, even that summary is too simple.  For a time, whether
program source code could enjoy copyright protection at all was debated
by legal scholars and courts in the United States.  As I understand it,
The 1976 Copyright Act settled that question, and was one of the
premises upon which Bill Gates rested his "Open Letter to Hobbyists"[1]
and established the foundational principle of copyright rentierism in
software upon which he built his multi-billion dollar empire.

> > The copyright absolutist approach would be to look at laws and prior
> > cases and recommend what is the safest and most conservative
> > approach.  As far as I know, that is still to do increment copyright
> > years.
> 
> Can we all at least agree that annually incrementing the year in
> _every_ copyright statement in the project is wrong?  You don't get a
> new copyright every year.

I agree, but it's a popular practice nevertheless, including in several
GNU projects.  The GNU Maintainers' Guide articulates a sound and
legally correct premise but then waves its hands such that some people
end up robotically "bumping" all copyright notices in a code base to
include the current year on or about 1 January, every year, before and
irrespective of whether any "nontrivial" change is made in that year.

"To update the list of year numbers, add each year in which you have
made nontrivial changes to the package. (Here we assume you’re using a
publicly accessible revision control server, so that every revision
installed is also immediately and automatically published.) When you add
the new year, it is not required to keep track of which files have seen
significant changes in the new year and which have not. It is
recommended and simpler to add the new year to all files in the package,
and be done with it for the rest of the year."[2]

I find that practice unseemly and offer different, and in my view more
scrupulous, advice in the project I maintain.[3]

Regards,
Branden

[0] https://en.wikipedia.org/wiki/Salinger_v._Random_House,_Inc.
[1] https://en.wikipedia.org/wiki/An_Open_Letter_to_Hobbyists
[2] https://www.gnu.org/prep/maintain/html_node/Copyright-Notices.html
[3] https://git.savannah.gnu.org/cgit/groff.git/tree/HACKING?id=0cd44362696c9d65ab59f6014f15221ac53b57f3#n101

Attachment: signature.asc
Description: PGP signature


Reply to: