Sunday, August 21, 2016

What do universities, journals, and government need to do to stimulate breakthrough scientific discovery? Tirumalai Kamala


Short answer: Solutions start from accurately identifying the problem. Fact is biomedical research has plenty of innovation 'on paper'. The feeling it isn't enough arises from a bottleneck because most breakthroughs aren't reproducible and hence fail to translate to the clinic. Once we accept the problem is lack of reproducibility rather than breakthroughs per se, we can more accurately envision remedies.

With top weekly science journals like Nature and Science expanding their specialist journal base at breakneck pace, weekly issues filled with breakthrough scientific discoveries, clearly the problem isn't insufficiency. Rather there's too much done the wrong way. Who sifts gold from dross to determine which breakthrough's reliable? So far no one. That leaves data irreproducibility to unfold slowly years down the road, like a disaster movie in slow motion.

Publications the lifeblood for a successful academic career, how often do journals publish papers with negative data asserting breakthroughs weren't reproducible? Question's rhetorical because the answer is almost never. Why would they when scientists submitting the papers and those reviewing them pro bono, both are fully invested in the novelty model which rewards their careers? Is it reasonable journals all the way from Nature/Science to specialist journals uniformly impose the mandate of novelty? In fact, all three, universities, journals and government operate a novelty-mandating ecosystem that can only change if they give reproducibility some parity.

No doubt such a preposterous idea could invite howls of derision. After all, with generations of scientists trained to go for guts and glory at all costs, who in their right mind would devote their scientific career to re-doing others' work? Seems an intractable problem except it isn't really so. For far too many years at a stretch, universities have been graduating far more biomedical researchers than can be absorbed by academia and industry. Forced into one post-doc after another, a large mass of highly trained scientists is looking for meaningful work beyond the tedium and indignity of being merely trained hands. The worthwhile goal of helping clean-up biomedical research might be just their ticket. With a sprinkling of high-profile reproducibility studies, Nature, Science and their ilk could continue to focus on novelty, providing the fodder for reproducibility studies in specialist journals. This would create a mutually reinforcing virtuous cycle that fuels reproducible breakthroughs that successfully translate to the clinic. Frustrated perception of not enough breakthroughs would start to fade. Biomedical research would also gain back its reputation for probity.

Bonus idea: Perhaps uniquely so, in one stroke the US has the opportunity to make this happen another way as well. The NIH, the largest government funder of US biomedical research, splits its budget two ways, 90% for funding external research, so-called extramural and 10% for funding its own internal research. The so-called intramural is purely play-in-a-sandbox-for-life type of funding for thousands of full-time NIH intramural researchers. Set up to do the kind of esoteric, high-risk science it was thought tenured university researchers just wouldn't undertake, clearly decades later, the system doesn't work the way it was envisaged. Instead, it's more of the same. However, it represents  a net opportunity. Simply switch the NIH intramural mandate from breakthrough to reproducibility. The budget, workforce and infrastructure are all there. Secure lifetime funding attracted precisely those who value stable job security and generous benefits, exactly the workforce needed for diligent reproduction, for sifting gold from dross from among the breakneck-paced breakthroughs.


https://www.quora.com/What-do-universities-journals-and-government-need-to-do-to-stimulate-breakthrough-scientific-discovery/answer/Tirumalai-Kamala


No comments:

Post a Comment