Licences, Intellectual Property, and Publication

Overview¶

This lecture covers the legal and publishing side of open research: what intellectual property (IP) protects in a research project, who owns it, how licences make a protected work open, and what publishers and funders now expect when it comes to sharing articles, data, and code. It complements two other Polytechnique courses: CAP7210 (IP and patents) and CAP7005 (copyright and open access).

What We’ll Cover¶

• The forms of IP relevant to research, and what copyright does (and doesn’t) protect

• How to figure out who owns a research output

• What a licence is, and how to read one

• Creative Commons licences for text, figures, and data; open-source licences for code

• How openness and valorization can coexist

• Data availability statements, ORCID, and open access requirements

Intellectual Property in Open Research¶

Four ideas are easy to mix up: IP (what rights protect what?), ownership (who holds them?), open dissemination (on what terms can others reuse the work?), and valorization (how is value created from research?). The what determines the how: before choosing a licence or a valorization path, you need to know what you hold, whether it’s protected, and who owns it.

Open science lives almost entirely in the copyright row. Copyright protects the concrete expression of a work — never the underlying idea, algorithm, or raw fact. Raw data (measurements, elevations, flows) belong to no one, but their form (file structure, selection, analysis scripts) can be protected, and access can still be restricted by contract. Personal information can’t be owned either, but it is governed by privacy law and ethics (see Lecture 4). And under current Canadian law, originality requires a human author, so purely AI-generated code is generally not protected — though case law is still evolving.

Who Owns a Research Output?¶

By default, the author is the first owner of a work. Several other layers can change that, and it’s worth checking them in this order: (1) the law — is there employment, or several authors? (2) funding and partnership contracts; (3) your institution’s IP policies, which often differ for students and employees; (4) specific agreements with your supervisor or collaborators; and (5) the licences of third-party components you build on (libraries, datasets, pre-trained models).

Licences¶

A licence is written permission from the owner, with conditions. It does not transfer ownership — you remain the owner of code you release under an open licence.

When reading a licence, ask: who grants the rights, what is allowed (use, modify, distribute, commercial use), on what conditions (attribution, share-alike), what warranties (usually none, “as is”), and for how long. A licence also doesn’t cover trademarks: a project’s name and logo stay protected unless stated otherwise.

Creative Commons Licences¶

Creative Commons (CC) licences suit publications, figures, videos, and datasets — not code. They combine building blocks: BY (attribution), SA (share-alike), NC (non-commercial), and ND (no derivatives). CC0 waives all rights, including attribution. Use the current version (4.0) and state the licence clearly on the work itself. The FRQ’s open access policy strongly encourages CC BY 4.0, with CC0 also accepted.

Licences for Code¶

Code licences fall into three families: proprietary (no reuse without permission), copyleft (e.g., GPLv3: modification and redistribution allowed, but distributed derivatives must keep the same licence), and permissive (e.g., MIT, BSD, Apache 2.0: few restrictions). When you reuse code, check whether its licence constrains yours — the GPLv3 library in the GéoCrue case does. When you wrote it, you choose. For maximum reuse, MIT for code and CC BY for everything else is a simple, widely understood combination. Also note that some platforms automatically license what you post (e.g., Stack Overflow).

Openness and Valorization¶

Opening your work doesn’t prevent creating value from it — it changes where the value lies. “Open” is about permissions, not price: open licences usually allow commercial use. Valorizing doesn’t only mean patenting — much of it happens through services, expertise, contracts, and trademarks. And publishing means losing what you disclosed (e.g., patentability), not necessarily the value.

In the GéoCrue case, the code is GPLv3, the maps are CC BY, and a company sells what is scarce and hard to copy: local calibration expertise, engineer-sealed reports, a hosted monitoring service, training, and a trusted brand name. The same pattern shows up in academic labs: when a lab’s open-source MRI analysis tools become widely used, organizations running clinical trials hire the lab’s expertise to analyze their data or design acquisition protocols — not the software. Openness created that demand in the first place.

Publications and Reproducibility¶

Data Availability Statements¶

Since the mid-2010s, most publishers require or strongly encourage a data availability statement (DAS). At a minimum, it says where the data and code are, and under what licence — e.g., “Data and code associated with this article are available at https://doi.org/... under CC BY 4.0 (data) and MIT (code).” Prefer a DOI over a Google Drive link, and avoid “available upon request,” which rarely works in practice graf2020. Articles whose DAS links to a repository are also associated with more citations colavizza2020colavizza2024. More broadly, DORA pushes to recognize data and code as research contributions in their own right; Polytechnique signed it in January 2026.

An ORCID iD — a free, persistent 16-digit identifier — follows you across institutions and distinguishes you from researchers with similar names. Publishers and funders increasingly ask for it: create one early and keep it up to date.

Open Access¶

Open access means peer-reviewed research is freely available online, without a paywall. Current requirements:

• Tri-Agency (NSERC, CIHR, SSHRC): under the 2026 Open Access Policy on Publications, articles from grants awarded on or after January 1, 2027 must be openly available immediately (the previous policy allowed a 12-month delay).

• FRQ: publications under CC BY (or equivalent), theses freely available immediately, for both grants and scholarships.

• Polytechnique: no institutional open access policy yet; one is under review.

There are two main routes. Gold: publish in an open access journal, usually for a fee. Green (self-archiving): deposit your peer-reviewed manuscript, usually for free, in a disciplinary archive (e.g., arXiv) or an institutional repository (e.g., PolyPublie), after any publisher embargo. Check a journal’s policies before submitting, e.g. with Open Policy Finder.

Beyond the PDF¶

Publications are evolving from static PDFs into interactive, executable objects that bundle text, code, data, and computing environment — reproducible by design. Examples include the qMRLab MOOC, NeuroLibre preprints, Aperture Neuro, and this course book itself. What gets reviewed, cited, and licensed is no longer just text, and each component may carry its own licence.

Further Resources¶

• Choose a License — Plain-language guide to picking an open-source licence

• Creative Commons licence chooser — Interactive tool for CC licences

• Tri-Agency Open Access Policy on Publications — Federal open access requirements

• Open Policy Finder — Publishers’ self-archiving and embargo policies