Life of Research
Appearance

Publishing your work

Choosing an open license for your paper, data or code

Updated 5 min read By Life of Research

Short answer

For most papers choose CC BY 4.0: anyone may reuse the work, including commercially, as long as they credit you. For datasets choose CC0, which removes legal conditions and leaves credit to scholarly citation. For code use a software license such as MIT or Apache 2.0, not Creative Commons. Avoid NonCommercial and NoDerivatives unless you have a specific reason, because they block much legitimate reuse. You can only license what you own, so check any publisher agreement first.

What does a license actually do?

A license gives everyone permission in advance to do things that copyright would otherwise forbid, such as copying, translating or redistributing your work. Without a license, the default in most countries is "all rights reserved": people may read your work and quote it under fair use or fair dealing, but may not republish, adapt or mine it without asking you.

An open license replaces thousands of individual permission requests with one public statement. Readers, libraries, text-mining projects and textbook authors can reuse the work immediately, on the conditions you chose. You keep the copyright; the license only grants permissions.

Two properties matter when choosing. First, Creative Commons licenses cannot be revoked: anyone who obtained the work under a license may keep using it on those terms even if you later stop distributing it. Second, a license covers only what you hold rights in, so figures reproduced from other people's work stay under their original terms.

How do the Creative Commons licenses compare?

Every Creative Commons license except CC0 requires attribution (BY). The others add up to two restrictions: ShareAlike (SA), NonCommercial (NC) and NoDerivatives (ND).

LicenseOthers mayConditionsTypical use
CC BY 4.0Copy, share, adapt, use commerciallyCredit the creator, link the license, note changesPapers, preprints, theses, slides
CC BY-SA 4.0As CC BYAdaptations must use the same licenseTeaching material, wikis
CC BY-NC 4.0Copy, share, adaptNo commercial use; credit as CC BYWhere commercial reuse is a real concern
CC BY-ND 4.0Copy and share unchangedNo adaptations, including translationsRarely suitable for research
CC0 1.0AnythingNone: rights waived as far as the law allowsDatasets, metadata

Life of Research offers CC BY 4.0, CC BY-SA 4.0, CC BY-NC 4.0, CC0 1.0, the MIT License and "All rights reserved". The license you choose is shown on the record page and included in its metadata.

Which license should you choose for a paper?

For a paper, preprint, thesis or report, CC BY 4.0 is the usual choice and the safest default. It allows the widest reuse while guaranteeing credit, and it is the most widely used license among open-access journals and the one many funders ask for. Plan S, the policy of the cOAlition S group of funders, requires an open license and names CC BY as its preference.

CC BY lets others translate your paper, include it in course packs, build text-mining corpora and quote figures with attribution, all without asking. Each of those uses is something researchers usually want.

NonCommercial sounds protective but is hard to interpret: it is not always clear whether use by a company-funded lab, a paid course or an advertising-supported website counts as commercial. NC material also cannot be added to projects such as Wikipedia, which accept only licenses that allow commercial use. Choose NC only when you have a concrete commercial use to prevent.

Which license should you choose for a dataset?

For a dataset, CC0 is widely recommended, and some data repositories accept nothing else. Data is combined with other data far more often than text is, and attribution conditions multiply: a dataset merged from hundreds of CC BY sources must credit every one of them in the way each license specifies, a problem known as attribution stacking.

CC0 does not mean you lose credit. Researchers who reuse data are expected to cite it, just as they cite papers, and that expectation is enforced by scholarly norms, journal policies and peer review rather than by copyright. Make citation easy by giving the dataset a clear title, a version and, where possible, a DOI; the citation guide shows the format.

If you need attribution as a legal condition, for example because a funder requires it, CC BY 4.0 is the next choice. The 4.0 Creative Commons licenses also cover the database rights that exist in some jurisdictions, including the European Union, which earlier versions handled less clearly.

Which license should you choose for software?

For code, use a software license, not Creative Commons. Creative Commons itself recommends against using its licenses for software, because they say nothing about source code distribution or patents. Common choices are:

  • MIT License: short and permissive. Anyone may use, modify and redistribute the code, including in closed products, if they keep the copyright and license notice.
  • Apache License 2.0: permissive like MIT, with an explicit patent grant from contributors.
  • GNU GPL 3.0: copyleft. Distributed modified versions must also be released under the GPL with their source code.

Life of Research currently lists MIT for code. If your project uses a different license, keep its LICENSE file in the deposited archive, say which license applies in the description, and contact us so the option can be added.

What should you check before choosing?

Check that you hold the rights before applying any license:

  • Publisher agreements. If a journal has accepted or published the paper, your agreement may have transferred copyright or limited which version you may share and when. Open policy finder (formerly Sherpa Romeo) summarises publishers' sharing policies.
  • Rights retention. Some funders and institutions ask authors to state at submission that a CC BY license applies to the accepted manuscript, which preserves the right to share it.
  • Co-authors. All rights holders should agree to the license.
  • Third-party material. Figures, photographs or data you reused under permission keep their own terms; mark them as excluded from your license.
  • Personal data. A license cannot make it lawful to publish data about identifiable people. Anonymise first or share under controlled access elsewhere.

Can you change the license later?

You can make a license more open later, for example from CC BY-NC to CC BY, but you cannot take back permissions already granted. Copies obtained under the earlier license remain usable on its terms. On Life of Research you can edit a record's license from My records; the new license applies from then on.

Because of this one-way door, it is reasonable to start with the license you are sure of and open it further when you are ready, rather than the reverse.

Key terms