Re: [libreplanet-discuss] Are proprietary software and science compatible?

[Mike Gerwitz]

On Sat, May 14, 2016 at 09:11:11 +0200, Fabio Pesari wrote:
> Some scientists go to great lengths to make sure their experiments are
> reproducible and can be peer reviewed, but then often use proprietary
> programs to achieve their results.

On that note, this article discusses issues related to the problem of
source code and reporoducibility from a scientific perspective.  We may
not (or may) agree with all of it, but it's a useful perspective:

  http://www.nature.com/nature/journal/v482/n7386/full/nature10836.html

(Note that the first comment contains more information from the authors
that they could not fit in the article due to word limits.)

> It seems a bit contradictory to me. For starters, the proprietary
> programs themselves aren't peer reviewable, so the researchers (and
> other scientists and the general public) just have to blindly trust what
> those programs say.

...but it stops very short: under the "Limited Access" heading, it says:

  Researchers may not have access to at least some of the software
  packages that are used for development. We suggest that this would not
  be a problem for most researchers: their institutions would normally
  provide such software. If it were to be a problem, then a journal
  could mark a publication as ‘Partial source code’. The release of the
  code, even without the software environment required for compilation
  and execution, would still be valuable in that it would address issues
  such as dissection and indirect reproducibility (see above) and would
  enable rewriting using other programming languages.

Which completely misses the point.  And I think this is the largely
accepted view, and somehow dismisses the very issues that the paper is
raising by either assuming that the black box of the proprietary library
is somehow immune, or just negligently neglecting consideration on the
matter.

I meant to write the authors a while back but forgot (for this reason
and a number of others); I think I'll do so.

> So, in the end, I don't think science and proprietary software are
> compatible at all, and research conducted through proprietary software
> should be considered unreliable.

The simple answer is: proprietary software isn't compatible with _any_
reasonable philosophy.

Short of that, yes, for the many good reasons you raised, and many
others.

> For this reason, I think proprietary software's unscientific nature
> should be stressed more when promoting free/libre software to scientists.

Agreed.

Are you in a scientific field, by chance?

-- 
Mike Gerwitz
Free Software Hacker+Activist | GNU Maintainer & Volunteer
https://mikegerwitz.com
FSF Member #5804 | GPG Key ID: 0x8EE30EAB

signature.asc
Description: PGP signature