Over on Daniel S. Katz’s blog there’s a discussion of university policies and open source software. The issue of Bayh-Dole came up, and I provided a comment there. I’m reposting here, with links and a few typos and awkwardnesses fixed.
The Bayh-Dole Act requires federal agencies to use a standard patent rights clause in funding agreements with universities, nonprofits, and small businesses. The patent rights clauses are not set out in the law but rather in the implementing regulations. Each time a university receives federal funds in the form of a grant, contract, or cooperative agreement, a federal funding agreement is created that includes a standard patent rights clause. For federal grants, the funding agreement is at 2 CFR 200 and the IP clauses are sections 315 and 316. That’s how Bayh-Dole gets to universities–universities formally agree to terms in federal contracts. Bayh-Dole *does not apply to universities directly*.
As part of the standard patent rights clause (at 37 CFR 401.14(f)(2)), universities agree to require written agreements of research and technical personnel to disclose patentable inventions, sign paperwork to allow patent applications to be filed, and sign paperwork to establish the government’s rights in inventions they make–either assignment or a non-exclusive license. Universities as a general practice do not comply with this requirement and instead demand assignment of inventions to the university.
Bayh-Dole has nothing to do with work that is not patentable and nothing to do with work that is outside the “planned and committed” activities under a grant. Thus, if inventive software is not in the RFP or the proposal or the statement of work, then it’s not part of the planned and committed activities. See 37 CFR 401.1 for the full discussion of scope. The regulations anticipate that contractors will attempt to exclude inventions from standard patent rights clause obligations. Instead, as a general practice, university licensing officers aim to *expand* the scope. Doing so gives them (so they argue) claim to ownership of a wider range of inventions “by operation of federal law.” The Supreme Court in Stanford v Roche threw out the universities’ interpretation of the law, but the universities for the most part only made their claims on inventions (and most anything else) even more demanding. A summary of the key passages of the ruling is here.
The provisions of the standard patent rights clause do not require a university to take ownership of federally supported inventions, or to file patent applications (a university can choose not to file), and if a university does file patent applications, the university is not obligated to license, commercialize, or make money on the patents. The standard throughout Bayh-Dole, its regulations, and the standard patent rights clause is “practical application”–the invention is being “utilized” and its “benefits are available to the public on reasonable terms” (see the definition at 37 CFR 401.2(e)).
Clearly, open source is entirely consistent with Bayh-Dole. A university may obtain patents on federally supported inventions embodied in code and deploy the code under a compliant open source license–Apache 2.0, for instance, or Eclipse or GPL 3.0.
If the university does not take ownership of a federally supported invention, then the regulations permit the federal agency to allow the inventors to retain ownership under 37 CFR 401.9. The inventors have still have obligations, but not nearly so many as those imposed on universities–those drafting the regulations apparently distrusted university involvement! 401.9 does not even require the inventors to file patent applications, though an agency might be able to add that requirement as a special condition. Of course, if the inventors do not elect to retain title or file a patent application and decide not to pursue that application or maintain any issued patent, then the federal agency can request assignment of the invention to the government. So inventors of code-based inventions need to have a talk with their program officers before making a decision about patents.
If the code does not embody a patentable invention, then for federal grants to universities 2 CFR 200.315(b) and (d)-(e) control–copyright and data rights. Neither is incompatible with open source software distribution, regardless of whether the university ends up claiming copyrights in the code. Section 315(e) is especially interesting for open source, as is section 316. Section 315(e) requires the university to use the property (“intangible assets”–including copyrights and patents) for the “originally-authorized purpose.” Section 316 goes further: if the university claims to acquire inventions from researchers supported by federal funds as a *condition of employment*, then the university is acquiring the rights to patent applications and patents *with federal funds obtained under the funding agreement*. Section 316 requires the property “to be held in trust by the recipient [i.e., the university] as a trustee for the beneficiaries of the project or program under which the property was acquired or improved.” If a grant proposal includes an expectation of open source distribution, the university, if it becomes the owner of patents or copyrights in the code, is obligated by the funding agreement to support that open source distribution. As a matter of general practice, however, university licensing personnel ignore these provisions.
In short, Bayh-Dole formally is a non-issue for open source.
The NIH, for instance, has various guidelines on sharing research tools and data that are highly consistent with open source:
exclusive licenses for research tools . . . should generally be avoided except in cases where the licensee undertakes to make the research tool widely available to researchers through unrestricted sale, or the licensor retains rights to make the research tool widely available….
(from NIH’s Principles and Guidelines for…Disseminating Biomedical Research Resources,” Federal Register 72095).
All the complications attributable to Bayh-Dole come from university patent licensing practice. These practices are often supported by university patent policy demands that licensing personnel have also drafted. Their claim supporting such policy demands has been that federal law (Bayh-Dole) requires institutional ownership and efforts to “commercialize” inventions. Commercialization (again, in general University licensing practice) means exclusive licensing to an established company, and if not to an established company, then to speculative monopolist investors, and if they won’t play, then to shell startups in the hope that in the future speculative monopolists will appear to buy the shell companies. Typical university exclusive licenses forbid open source distribution of the licensed code.
So university licensing practice, not Bayh-Dole, is the cork in the keg of open source, exchange of research ideas expressed in code or digital data, collaboration, public access, common platforms, and further code development.