Julia Belluz and Steven Hoffman at Vox have published a new account of how screwed up academic science is. Belluz and Hoffman report on a string of studies and exposed forgeries that suggest that the published scientific literature is anything but reliable. They quote Richard Horton of The Lancet: “Much of the scientific literature, perhaps half, may simply be untrue.” Belluz and Hoffman point to a study by Simon Shorvon and Nattanit Gregoris that found over 70% of papers published on epilepsy over a span of 20 years had “no enduring value.” Among other figures, Belluz and Hoffman point to an estimate made in 2009 by Iain Chalmers and Paul Glasziou that 85% of global biomedical research funding is wasted. That’s $200 billion out of $240 billion.
Think about it. More funding for academic research likely means more bad science, not more discovery and innovation sparking economic growth. No wonder some university administrators have started talking about “research as an industry.” I expect these administrators expect that building an extensive research footprint will bring talent and companies and inventions and venture capital and jobs and taxes and more funding for expanding research infrastructure. But perhaps what happens is a pork industry of taking government money, publishing studies mostly “with no enduring value” with the happy knowledge that next year, there will be more money.
Belluz and Hoffman, however, put a twist on this whole sad story. They argue that science, being human, is naturally going to be flawed and we should all just learn to live with it–or, actually, to use their phrase “embrace” flawed science. “This is how science works” appears to be the message. But if that’s the case, science could also “work” with about $200 billion less funding and likely be much more effective–presumably, a bunch of the academic scientists specializing in flawed, cherry-picked, statistically challenged work of “no enduring value” would find other work, and perhaps then only 50% or 30% of total research funding would be “wasted.”
Here is the opportunity, however, for university technology transfer. Technology transfer has the opportunity to act as a filter on the stream of bad science, acting as an impartial institutional review. Beyond merely determining if some new finding is patentable and ready for sale to an exclusive licensee, a technology transfer program could push disclosed inventions out to peer reviewers in industry for feedback–is the finding replicable? Does the data support the broad claims? If so, then perhaps the disclosed invention is that 1 in 7 that might have enduring value. Of course, this approach is how most university technology transfer worked prior to Bayh-Dole. Send out disclosure reports to industry; find out what people think. After Bayh-Dole, and especially after provisional patent applications, and more especially after first to file, the typical practice is to file first and find out later–if ever.
So think about it. If most published academic research is sketchy bad, then perhaps most reported inventions to university technology transfer programs are also sketchy bad. Requiring the disclosure of all inventions, and demanding institutional ownership of all such inventions, then serves to increase the number of bad science inventions under review in the technology transfer program. The demand to see and own all inventions is based on the premise that most all inventions are based on good science. The data challenge that premise. Instead, technology transfer programs should “embrace” the flawed nature of human science–at least in its present practice–and revert to the pre-Bayh-Dole practice of relying on faculty inventors to decide when an invention is worth exploring commercial interests. Perhaps faculty self-selecting inventions for commercialization is the single best way for an institution to sort out high-value science from the 85% of wasted work. Add showing the self-selected inventions to industry for review prior to embarking on a patenting effort would create a highly selective, focused, and likely much more effective technology transfer program.