I'm a computer scientist. I'm very fond of free software as it helps
me to use my machine to the fullest possible extent. It is also a bless
when writing research code, free software will almost always cover the
basics for you. A computer scientist can handle most of the issues
that would be difficult to accomplish in a proprietary operating system
reliably and rapidly.
It is a frequent extrapolation in the free software community to liken
research to free software: all research software should be released
as free software.
In this article, I argue that research code has very different
requirements from free software (ie GNU/Linux systems) and propose
that a license suited to the needs of a researcher should be crafted.
Now that my Msc. thesis is completed I wonder which parts of it
I should release as free software. I need a license that will
help me keep the research open. I wouldn't like my research being
worked on behind closed doors.
In fact, that is what makes research different from commodity software.
GPL is designed for generally useful software thus allows for private
modifications of all sorts. That makes a lot of practical sense, as
many people will need to modify the software for their own use.
Nevertheless, if I GPL a research code somebody will be able to
1) Extend my work in a journal paper, and not even talk about the code
or the license.
2) Use my work as a service, say on the WWW without ever making the
code freely available.
Moreover, my software is not generally useful software. It isn't an
operating system, a command line shell, a compiler, or a web server.
It is a very specific piece of code that can only be used and understood
by experts. And I intend to keep it that way for a long time (say 3 years)
It is an orthogonal matter whether a work in computer science may
have any practical use. As Dijkstra said, computer science is no more
about computers than astronomy is about telescopes. Much research software
does not even have a slight connection to the real world.
That said, researchers have unique requirements that we should consider.
First of all, a researcher does a piece of work for recognition. Second,
it's the results that matter not the software itself per se.
Third and most importantly, being able to maintain scientific practice
is beneficial for both the scientist and science in general. Some scientists,
like me, believe that is possible through making scientific process more
I'm cutting the article here, as my goal is to collect feedback from
fellow hackers. What kind of a license would you deem appropriate for
the requirements of a researcher who favors openness in science?
I have beeing developing software for bioinformatics
last two years, and solid state physics software the two
years before, and trying to convice my advisor to
release it as free software. From the point of view
of a person that belives on free software and advocates
it, you can find in a paper that I link here my personal vision of the
"state of art"
of free scientific software, what is a way to research
on free software.
The paper was published on Informatik/Zeitschrift
schweizeizerischen Informatikorganisationen on
its number of december of 2001 -it is written in
english, don't worry-, but was written one
year before. You can find the paper here.
My views are:
The free software is the natural way to develop software
to a scientific, due to software is knowledge, and knowledge should
be free; but
The situation is bad.
The situation is getting worse.
There should be some INDEXED paper centered on free software
proyects -needed to be able to keep our research grants and to get
tenure positions to free software researchers.
Govern paied proyects, govern developed proyect should be free
under a free licence. (MS, PhD grants, government-founded research)
The oxonorious advertising clause and the paper citation. This
is a complex point, and I have had lots of e-mail exchanging about it
with free sofware developers after the publication of the paper. My
mind has shifted a little bit, but I know that some kind of enforcing
the citation of a particular paper if you research and publish using
code it is something against the GPL, and something that people like
RMS dislikes -old BSD licence case-. But it is a must-be for the most
of the "old dinossaurs" of the departments before releasing code under
any free licence. This is a complex issue, and I am not complety
convinced that there is a way to satisfy all people.
Please feel free to do any coments about the paper or about this
reply, so far this is an area in which I am very interested. My email
is irbis at orcero dot org.