5

Not sure if this belongs to CrossValidated, Computational Science, or something else entirely.

I've got some code and data that's going into a publication, but I'd like to make said code and data publicly available so that the analysis can be replicated. If I was having it hosted by the publisher, I'd ignore this and let them deal with it, but for a number of reasons to code and data is being hosted externally on my own site.

I presume I should put some sort of license on said code and data. Is there a Creative Commons or other type of license that lends itself particular well to open data? Something else I should consider? Just not bother?

Fomite
  • 23,134
  • Well, the GPL would probably be a default choice for the code, as it allows for others to improve and contribute to the code while not allowing them to change it and not make the changes available. I am unaware of any open data licenses, but would be happy to be proved wrong. – richiemorrisroe Apr 04 '12 at 10:08
  • Excellent question. I've been wondering about this myself. – conjugateprior Apr 11 '12 at 01:06
  • 1
    For making your data generally and systematically available in citable form to complement the paper, the folk at the Dataverse Network can probably help, often at no cost. (btw I'm not one of them). – conjugateprior Apr 11 '12 at 01:14
  • @richiemorrisroe Does the ability to "make code available while not allowing them to change it and not make the changes available" support the progress of science? Scientific methods are generally made available for both use and modification. MIT, BSD, and Illinois/NCSA licenses are less restrictive. – David LeBauer Apr 11 '12 at 03:55
  • there is a related question about publishing data on academia.se. Your institution may have guidelines or suggestions (a librarian should know). For data - I think public databases are quite valuable, as are standard formats (if available) and meta-data - all of which facilitate more powerful data syntheses. – David LeBauer Apr 11 '12 at 04:01
  • @David i think it depends on the discipline. I like the GPL mostly for ideological reasons, but it would be nice that people who use my code would at least contribute their changes back to me. Hidden code and obscure data analytic strategies are a problem in my discpline at least (psychology) – richiemorrisroe Apr 11 '12 at 11:13

1 Answers1

3

Creative commons zero is what I've seen recommended for open data.

Here's the FAQ page, and here's a list of organizations that use it.. Here's a third-party advocating its use.

cc0 essentially puts your data into the public domain. Compared with the GPL or CC-share-alike, it gives your users more freedom, especially if they're working with hundreds of data sets with conflicting licenses for a meta-analysis or something.

For instance, with cc0, if they augment your data set in some way (e.g. by combining it with their own data), they don't have to share. If you'd object to them holding on to the modified data, then you want a different license. But if you don't, then cc0 will make everyone else's life easier.

David J. Harris
  • 11,653