Is science too risk-averse?

Many leading science funders have a stated goal of supporting ambitious, “high-risk, high-reward” research, or at least specific programs dedicated to that cause — to give a few examples: “ARIA was built to… go after ideas that may seem far-fetched, but could unlock world-changing capabilities,” “[ARPA-H] provides leadership for high-risk, high-reward biomedical and health research,” “[VolkswagenStiftung] supports groundbreaking and risky research ideas,” and “NSF’s commitment to fund high-risk, high-reward ideas strengthens the U.S. economy...” At the same time, there is relatively widespread concern in the metascience community that science is becoming too risk-averse, conservative, and incremental, and there is growing research evidence to back up the case. What gives? Part of the answer is institutional: Franzoni and Stephan argue that standard review processes, regardless of stated goals, are not structured to properly evaluate risk; and Azoulay and Greenblatt find that risky NIH grants are indeed renewed at lower rates, especially for novel research areas and new investigators. But incentives for/against risk are also social and individual: risk aversion can be rational at the level of the individual scientist, even if it’s collectively suboptimal. So is the scientific enterprise too risk-averse? This week Johns Hopkins hosted a debate on exactly this question, with Tyler Cowen and Brandon Ogbunu arguing for, and Kate Biberdorf and Sethuraman Panchanathan arguing against. In the end, the “too risk-averse” side won more converts among audience members, but both sides agreed that expanding support for science is one of the most important levers the enterprise could pull to increase innovation.