A recent tweet called out the Duke University technology transfer program for starting 12 companies so far in 2022. That’s a good number of startups for a university. We should applaud. But the tweet went on to attribute the startups to Bayh-Dole:
#BayhDole makes all of this possible, getting the research from the lab to the consumer.
We should note that (i) Bayh-Dole does not make starting companies associated with universities possible, (ii) that starting a company does not mean that a product based on a licensed subject invention will ever get to “the consumer”, and (iii) that utilization in the form of practical application, not consumer products, is the stated purpose of Bayh-Dole’s restrictions on patent property rights for inventions made in federally supported work.
But given these observations, I thought I’d spend an afternoon of my mortal existence taking a look at Duke’s startups. Duke has a nice web page that lists a bunch of them. The information there could use a refresh and an edit but it can give us a rough estimate of the place of Bayh-Dole.
I looked through about 90 companies started at Duke from 2017 to 2022–nearly six years. For each I looked at the inventors, and checked their patent work at Duke using the new (sigh) Public Search tool at the USPTO. From that I made an evaluation of what patent(s) were licensed to the startup, and checked those for a government interest statement. For the most recent startups (2020-22), patents may not have issued (and may never, for that matter) and patent applications haven’t necessarily published, so I don’t have access to that information, though Duke does and certainly if someone there wants to supplement or correct my work, that’s what the comments section is for–or Duke folks, contact me directly, I’d appreciate it.
Here’s the basic sketch. Of the 83 companies that I reviewed that started 2017 or later, 20 cited federal funding, 25 did not, another 25 were not patent-based startups, and 13 did not have an issued patent and so could be any of patent (federal, not federal) or no patent. Of the startups without patents, many were software systems or offered services. Think of that, 25 of 70 (35%) startups at Duke had no patent-based inventions at all. Yet they were able to start, and make a go of it. Similarly another 35% of Duke startups were patent-based, but weren’t encumbered by Bayh-Dole restrictions. That means that of the startups I could find decent information on, 50 of 70–70%–of them were not creatures of Bayh-Dole. We might say that Bayh-Dole takes its share of startups but clearly is not the beast that makes startups possible. Companies can start without federal research, can start with or without university-held patents.
In a number of cases, the Duke startup page lists the inventor as the company’s “management.” In these cases, we should question the idea of “technology transfer.” The inventor already has not only the invention but also all the tacit knowledge around the invention–how to characterize it, where it may be off or need to be improved, data sets that demonstrate its function and performance. The loop that forces the inventor to assign the invention to Duke and then have Duke license it back to the inventor’s company is purely bureaucratic. There’s no benefit to the inventor or startup for this loop, other than that Duke fronts the cost of the patent application. It would be a whole lot more direct for Duke to offer to pay the startup’s patenting costs in exchange for a future payback plus something else of value, like a cash equivalent of 1% of the startup at the first significant liquidity event. Simple. No license involved. No bother with patenting–leave that to the company and inventor to mess with.
This direct approach is even possible with Bayh-Dole, though university technology transfer officials usually stick their heads in the sand and say it isn’t. Under Bayh-Dole, a university contractor can make potential inventors parties to the federal funding agreement (see 35 USC 201(b))–by any assignment, substitution of parties, or subcontract of any type. A party to the funding agreement is, by definition (35 USC 201(c)) a contractor. Thus, in this pathway, when a university inventor-cum-party to the funding agreement-cum-contractor invents, the invention is “of the contractor” and therefore a subject invention disclosable under Bayh-Dole’s standard patent rights clause, but the burden of disclosure runs with the inventor, not the university. The university is essentially out of the picture.
The inventor’s disclosure obligation to the federal agency triggers when the inventor discloses the invention in writing to the inventor’s patent personnel–such as patent counsel. Everything else follows, with the crucial difference being that the inventor-as-contractor is subject to a subset of the small business patent rights clause, not the nonprofit patent rights clause. The small business patent rights clause does not restrict how royalties and income earned can be spent, does not require small business preferences, and does not restrict assignments of subject inventions. In all, the inventor patent rights clause–37 CFR 401.9–is way more favorable to a university startup than is forcing the startup to take an exclusive license (=assignment) from the university under Bayh-Dole’s nonprofit patent rights clause.
Of course, if there’s no patent involved at all, as in the case of much software, data sets, video instruction, and services, then the basis for requiring a university license is even less persuasive. Sure, there can be copyrights and trademarks and information sets–and there may be educational-use-only, waive that fair use clause licenses that the university has accepted that must be navigated, but the rationale is pretty thin for requiring the university to get involved, assert an ownership interest, take assignment, just to license exclusively (expressly a transfer of ownership under US copyright law) back to the startup. At best, the university serves as a rights aggregator for a work of many individuals, holding the rights as it were in escrow until a decision is made on how the work might be deployed. In such a case, it matters just as much to aggregate rights (assignments, licenses-in) if the work is going to be released free of controls or non-exclusively with controls or licensed exclusively to a publisher or startup. But in such instances, also, it is as much a matter of a developer of a non-patent asset requesting that the university play this role than any policy asserting that the university must play this role–which is a make-work-for-administrators sort of thing where people really do have the wherewithal to work things out for themselves, especially if they use Creative Commons or open source/free licensing tools.
I didn’t find all 12 Duke startups for 2022–just 5–accounted for on their startup page. Maybe there will be an update. And I haven’t had time to chase down what has happened with the Duke startups since 2017. Some have raised millions of dollars in venture financing–very cool–and some still have stub website pages that make it look like they are paper companies looking for a good origami artist to give them shape. There are a few faculty researchers at Duke that in the past six years appear to be central to multiple startups each. Wow for that. There are plenty of stories to discover. I muchly want to learn about the worm therapy, and whether the tire tread sensor company (2016, so before the counts above) really has designed out from under its Duke patent license. But these things are for later!