[Debian-med-packaging] Bug#915541: Removal of upstream "--will-cite" functionality has been reverted

Lucas Nussbaum lucas at debian.org
Tue Sep 7 10:05:45 BST 2021


On 06/09/21 at 20:56 +0200, Ole Tange wrote:
> On Fri, Sep 3, 2021 at 5:05 PM Felix Lechner <felix.lechner at lease-up.com> wrote:
> > On Fri, Sep 3, 2021 at 7:50 AM Tobias Frost <tobi at debian.org> wrote:
> > >
> > > But as said earlier: This is not a license issue; the license of GNU parallel
> > > would allow removal, but this would make upstream sad.
> > > The status quo is likely to mke our users sad, though.
> 
> Maybe it would help if the consequences were explained to them:
> 
> * Do you want the software with no citation notice and risk that the
> maintainer will step down because he cannot afford spending time on it
> - thus getting less free software in the long run?
> * Or do you want to spend the 10 seconds it takes to silence the
> notice if you don't want to see it?
> 
> The ultimate goal has never been to have a license notice. The goal is
> to make it possible for me to spend time developing free software. In
> practice this means either pay my salary or have GNU Parallel cited,
> so it is easier for me to get a job that pays my salary.
> 
> It is unlikely that the Debian project will provide my salary, so let
> us focus on the second part.
> 
> Before the license notice was implemented researchers forgot to cite
> GNU Parallel; not because they did not want to honor the tradition,
> but simply because they forgot. The citation notice changed this for
> the better.
> 
> If there is a different way that will ensure researchers will not
> forget, it would be acceptable to me.

It is common for scientific tools or research infrastructures to provide
a "how to cite" section in their documentation. This is not a problem:
when researchers freely determine that they should cite something (for
example to provide the necessary context for their work) it's a good
idea to help them do that. See for example
https://www.open-mpi.org/papers/

Alternatively, infrastructures (and funding organizations) often require
an acknowledgement in publications. See for example
https://prace-ri.eu/hpc-access/project-access/project-access-information-for-awarded-projects/

However GNU parallel goes much further than that, because:
(1) the wording almost requires citation
(2) it does so by providing a version-specific citation, not a generic
one. In the past, it was asking to cite:
    O. Tange (2011): GNU Parallel - The Command-Line Power Tool,
    ;login: The USENIX Magazine, February 2011:42-47.
(This publication attracted 939 citations according to Google Scholar)
But now it asks to cite (in the current development version):

  @software{tange_2021_5233953,
        author       = {Tange, Ole},
        title        = {GNU Parallel 20210822 ('Kabul')},
        month        = Aug,
        year         = 2021,
        note         = {{GNU Parallel is a general parallelizer to run
                         multiple serial command line programs in parallel
                         without changing them.}},
        publisher    = {Zenodo},
        doi          = {10.5281/zenodo.5233953},
        url          = {https://doi.org/10.5281/zenodo.5233953}
  }

In a way, this is a way to manipulate metrics such as h-index, by
artificially generating several entries that will collect some
citations.

It is also marginally an abuse of the Zenodo archive: Zenodo provides a
way to track versions for the same software or dataset (see for example the
Versions box in https://zenodo.org/record/5458943). GNU parallel uses
different records, as if each version was different software. (I'm not
sure if this is done by mistake or also serves the purpose of generating
different entries to accumulate citations)

Thus while asking users to acknowledge their use of GNU Parallel in
publications would be totally OK, My personal opinion is that this goes
too far (and might also be ethically questionable). With a wrong eye,
one could even see it as extortion/blackmail.

Lucas



More information about the Debian-med-packaging mailing list