Re: The Project Gutenberg license, packages using its books as testdata

Soren Stoutner <[email protected]> Sat, 08 Mar 2025 20:24:57 -0700

Newsgroups	gmane.linux.debian.devel.general,gmane.linux.debian.devel.legal
Organization	Debian
Message-ID	<9953564.ag9G3TJQzC@soren-desktop>

On Saturday, March 8, 2025 7:46:58 PM MST Maytham Alsudany wrote:
> Hi debian-legal,
> 
> Today I was reviewing a package[1] that contains a file[2] from Project
> Gutenberg. d/copyright had listed it under Public-Domain, and it would
> seem that way from the website[3] where it says "Public domain in the
> USA", but the header in the file indicated that it was licensed "under
> the terms of the Project Gutenberg License".
> 
> This package has been accepted into Debian through NEW in the past with
> the d/copyright in this state, indicating that the FTP Masters are fine
> with it (I'm just guessing). The package was since removed and is now
> being reintroduced.

My guess is that the FTP Masters took the statement in debian/copyright that the file was 
in the public domain at face value and didn’t catch the complications with the Project 
Gutenberg License.

> This is not the only package to contain stuff from Project Gutenberg,
> codesearch.d.n[4] says 98 results when searching for "Project Gutenberg
> License".

That is concerning.

I am taking the liberties to copy my response to debian-devel to raise greater awareness 
of the problem as it affects a large number of packages.  The full text of the original email 
with the lengthy analysis is at:

https://lists.debian.org/debian-legal/2025/03/msg00004.html

> Is this really a public domain license? It doesn't seem to be that way
> after I initially read it, hence why I'm sending this mail. I've
> included the license (sourced from [5]) below with some comments.

> [snip excellent analysis]

I agree with your analysis, the Project Gutenberg license is not DFSG-free, most 
particularly because of the restrictions on commercial use.

However, as the license points out, some of the books they host are in the public domain 
(although it appears that they only verify they are in the public domain in the United 
States, so it would be up to the package maintainer to verify they are in the public domain 
worldwide).  Also, as you pointed out in your analysis, to exercise the public domain option 
the package maintainer would need to verify that “all references to Project Gutenberg are 
removed”, which you have stated is not currently the case with this package and doesn’t 
appear to be the case with other packages that codesearch identified.

-- 
Soren Stoutner
[email protected]

signature.asc (application/pgp-signature, 833 B)