We already know how to dramatically accelerate science: by getting out of the way.
Scientific progress has decelerated significantly in recent decades.1 Some of the reasons have become glaringly obvious. The institutional environment for academic science has changed to discourage good work, to reward bad work, and to waste much of researchers’ time on activities that have no scientific value or relevance. Off the cuff, I’d estimate more than 90% of resources are wasted now.2
Of course, research is inherently uncertain; I don’t mean 90% “fails.” It would be better if much more research did “fail”! Under current incentives, researchers have to ensure that everything they do “succeeds,” typically by doing work whose outcome is known in advance, and whose meager results can be stretched out across as many insignificant-but-publishable journal articles as possible. By “wasted,” I mean that often even the researchers doing the work know it’s of little value. Often they can name better things they would do instead, if they could work on what they believe is most important.
Most resources go into what Richard Feynman called “cargo cult science.”3 We perform rituals that imitate science, but are not science.4 We are just going through the motions, and that doesn’t deliver progress.
Cargo cult science means conforming to misaligned incentives. For academics, it optimizes the proxy objective “publish journal articles,” which has increasingly diverged from the actual objective, understanding natural phenomena.
The safest and most effective way to accelerate progress will be to improve incentives. If 90% of resources are currently wasted on cargo culting, eliminating that might increase research productivity ten-fold.
If we think superintelligence would be immensely valuable, it would also be sensible to make better use of the most extreme intelligences we have now. On the one hand, many of the best science and engineering students get hired by hedge funds and Mooglebook for activities with small or negative net value. On the other hand, those who go into academia face increasing obstacles to getting research done, and an increasingly dismal career path. In the past few years increasingly many of the best have left.
Reversing this will require dramatic changes in incentives. It implies providing outstanding researchers with career paths as rewarding as Bridgewater or Mooglebook, and letting them get on with plate wobbling or tick saliva or zeptobots or whatever other incomprehensible obsession they are on about.
As Stuart Ritchie writes:5
Changes that would make dramatic improvements to the quality of research are right there—but, although they’re often available, most scientists haven’t even begun to pick them up…. The burst of meta-science that we have seen since the replication crisis mustn’t be squandered: Pushing for the funding of much more such research should be a major priority for anyone who wants to improve science, and wants to do so using hard evidence.
Some significant changes are already under way. The Open Science and Replicability/Credibility movements, led by scientist-activists, have succeeded in changing some government, university, and academic journals’ policies. This has been slow and uphill, but seems to be accelerating, and could reach a tipping point.
Alternatively, impediments may be so entrenched in academia that adequate improvement has become infeasible there. Accordingly, creating alternative, better scientific institutions—funding mechanisms, workplaces, communication channels, social norms—is now important and urgent. Fortunately, this process is also under way.
I recommend a review article by Matt Clancey, “How to Accelerate Technological Progress”;6 the work of José Luis Ricón on progress studies and reforming scientific institutions;7 Stuart Ritchie’s “Rebuilding After the Replication Crisis”;8 Nadia Asparouhova’s post on “Understanding science funding in tech, 2011-2021”;9 and for a deep exploration of possibilities for institutional reform, with many specific proposals, Michael Nielsen and Kanjun Qiu’s “A Vision of Metascience: An Engine of Improvement for the Social Processes of Science.”10
- 1.“Progress” is not easily measured, so this is a somewhat subjective judgement. There are vastly more scientific journal articles published now than in the past, but most seem to have negligible or negative value. In an extensive tweet thread (twitter.com/mattsclancy/status/1612440718177603584), Matt Clancy reviews the evidence for slowing, with citations and graphs. Park et al.’s “Papers and patents are becoming less disruptive over time” (Nature 613, 138–144, 2023) is a recent quantitative study, summarized in Max Kozlov’s “‘Disruptive’ science has declined — and no one knows why” (Nature 613, 138–144, 2023). Classic works are Patrick Collison and Michael Nielsen’s “Science Is Getting Less Bang For Its Buck” (The Atlantic, November 16, 2018), Patrick Collison and Tyler Cowen’s “We Need a New Science of Progress” (The Atlantic, July 30, 2019), and Bloom et al.’s “Are Ideas Getting Harder to Find?” (American Economic Review 2020, 110(4): 1104–1144). Since those were written, awareness of replication crises across many sciences has made the case even clearer. A recent non-technical statement is Derek Thompson’s “Science Has a Crummy-Paper Problem,” The Atlantic, January 11, 2023.
- 2.After writing this off-the-cuff estimate, I came across John P. A. Ioannidis’ “How to Make More Published Research True,” which estimates 85% of research resources are wasted. PLoS Med 11(10): e1001747 (2014).
- 3.“Cargo Cult Science,” 1974 Cal Tech Commencement address.
- 4.Already in 1956, in an outstanding paper on scientific research management, John L. Kennedy and G. H. Putt wrote that “Research has come to be as ritualistic as the worship of a primitive tribe, and each established discipline has its own ritual. As long as the administrator operates within the rituals of the various disciplines, he is relatively safe. But let him challenge the adequacy of ritualistic behavior and he is in hot water with everyone.” That’s in “Administration of Research in a Research Corporation,” Administrative Science Quarterly, Vol. 1, No. 3, (Dec., 1956), pp. 326–339.
- 5.“Rebuilding After the Replication Crisis,” Asterisk, no date.
- 6.New Things, Aug 06, 2021.
- 7.Both are on his site Nintil.
- 8.Asterisk, no date.
- 9.On her site nadia.xyz.
- 10.On scienceplusplus.org.