Many excellent ideas here. Worth underscoring that the problem raised in point 3 (although perhaps indirectly related to many of the points) is actually quite fundamental:
"But people whose coalitional membership is constituted by their shared adherence to 'rational,' scientific propositions have a problem when—as is generally the case—new information arises which requires belief revision. To question or disagree with coalitional precepts, even for rational reasons, makes one a bad and immoral coalition member—at risk of losing job offers, one's friends, and one's cherished group identity."
Scientists in theory are the coalition that embraces new hypotheses and evidence, but in practice remain human. This problem specifically may be one where AI agents can be useful in correcting for our own blindspots — at least for so long as those blindspots aren't incorporated into their own models.
Much of these same ideas have been enunciated by the current head of NIH Dr. Jay Battacharya
in an interview several months ago just after being appointed to head the NIH. The interview was with Jim Robinson of the Hoover Institution at Stanford and can be seen on their website.
Some relevant thoughts on the subject by the late Mihayli Csikszentmihalyi:
"Left with carte blanche, every field naturally wants to control as many of the resources of society as possible, and more. The American Psychological Association would be happy if every school, business firm, and family had its own resident psychologist. The interest of artists is to convince the rest of society that things would be better if everyone became a collector of art, while the interest of dentists is to assure us that we would be happier if we devoted most of our free time to oral hygiene. Each field welcomes any new idea that promises to expand its hold on societal resources. In addition, even if there were no selfish, material reasons involved, each field would still push for the implementation of new ideas in its domain, regardless of long-term consequences. A person who has worked for years within the limits of a narrow specialization naturally believes that new developments in his or her domain are the most important and therefore take precedence over developments elsewhere. It is difficult to convince a physicist who has devoted a lifetime to high-energy physics that advances in nuclear technology should not be supported all the way."
...
"Scientists recoil in horror at the thought that anyone else should decide what is or is not good science. A person who has been awarded a Nobel Prize in physics has almost no choice but to believe that he or she is heir to the only possible way of studying the world. To paraphrase Voltaire, he naturally believes that his is the best of all possible sciences, and therefore that any attempt to question the inevitable unfolding of physics by physicists is an anti-intellectual attack on the integrity of science. Each field expects society to recognize its autonomy, yet each feels in the last analysis accountable only to itself, according to the rules of its own domain. For all of these reasons, it is useless to expect fields to monitor their own creative ideas in terms of the long-range public good."
...
"What is needed is a self-conscious effort to establish priorities and to use something like an 'evolutionary impact analysis' as one of the bases for the social endorsement of new ideas. A policy of this type should not result in any kind of philistine thought-policing. Artists should be encouraged to follow their muse, scientists should be respected for following a hunch wherever it leads them. On the other hand, why expect society to support novelties that are valued within a given field but may harm the commonwealth? The greatest art, East or West, was not produced when the artists set the agenda, but when patrons insisted on certain standards that benefited them. Patrons wanted primarily to be admired by the public, so the art they demanded had to appeal to and impress the entire community. In this sense, medieval and Renaissance art, commissioned by popes and princes, was in reality more democratic than it has become since the art world gained the power to separate itself from the rest of society—as a field with its own peculiar tastes and criteria of selection. "
"It admittedly would be more difficult to achieve a public evaluation of scientific creativity. In most scientific domains the frontiers of knowledge have moved so far beyond the grasp of laypersons that only those within the respective fields can be expected to make any sort of informed decision. But it is probably the case that within each field there are enough individuals with both expertise and a sense of the public good who could be deputized to serve the interests of society."
"Currently research grants are evaluated in terms of either the priorities set by the field or the political agenda of the administration disbursing the funds. Perhaps it could be possible to establish a sort of civil service above party politics and disciplinary fashions, composed of those who aspire to be “good ancestors,” as Jonas Salk called them, and who would be willing to represent the claims of evolution when assessing whether scientific advances should receive social support. Inevitably such a group would be composed mainly of older individuals, and therefore it would be open to criticism from younger colleagues who are more concerned with advancing their own scientific careers. On the other hand, the probability for dispassionate wisdom is greater among those who have had more, and more varied, experience and who can see their expertise in a broader context—and these in turn are likely to be older persons. Yet our society expects very little from its elders. This might be one important contribution of seniors that will benefit everyone."
Csikszentmihalyi, Mihaly. Creativity: Flow and the Psychology of Discovery and Invention (Harper Perennial Modern Classics 1996) (p. 323-6).
The fact that the specific issue Ioannidis pointed out in 2005 is not listed might indicate that we're still on the wrong path and will be writing this again in 20 years.
According to Forbes magazine, the United States has 902 billionaires. How about persuading one to run a test of whether sizeable grants to early-career researchers, enabling them to run their own labs for, say, five years, would generate the innovations that people claim that the bureaucratized NIH does not?
This article comes at the perfect time, considering teh critical need for innovation in science. How can we implement better algorithms for funding truly novel ideas? Your insights are incredibly valuable.
A problem with trying to fund very creative new work is that it will have a much higher failure rate. Of course, bad research generally will also have a higher failure rate. How to choose which projects to fund in light of this--- do your past failed grants count against you?-- is tough.
A Nobel laureate much attacked before his ideas were accepted once told me "only your home runs count."
I kind of agree, but in practice, I think that bad research has a higher success rate, because 1) people pick marginal issues to research where the answer is already fairly clear; 2) people p-hack their way to success; or 3) people commit outright fraud to show success.
Many excellent ideas here. Worth underscoring that the problem raised in point 3 (although perhaps indirectly related to many of the points) is actually quite fundamental:
"But people whose coalitional membership is constituted by their shared adherence to 'rational,' scientific propositions have a problem when—as is generally the case—new information arises which requires belief revision. To question or disagree with coalitional precepts, even for rational reasons, makes one a bad and immoral coalition member—at risk of losing job offers, one's friends, and one's cherished group identity."
https://www.edge.org/conversation/john_tooby-coalitional-instincts?ref=thebrowser.com
Scientists in theory are the coalition that embraces new hypotheses and evidence, but in practice remain human. This problem specifically may be one where AI agents can be useful in correcting for our own blindspots — at least for so long as those blindspots aren't incorporated into their own models.
Much of these same ideas have been enunciated by the current head of NIH Dr. Jay Battacharya
in an interview several months ago just after being appointed to head the NIH. The interview was with Jim Robinson of the Hoover Institution at Stanford and can be seen on their website.
Some relevant thoughts on the subject by the late Mihayli Csikszentmihalyi:
"Left with carte blanche, every field naturally wants to control as many of the resources of society as possible, and more. The American Psychological Association would be happy if every school, business firm, and family had its own resident psychologist. The interest of artists is to convince the rest of society that things would be better if everyone became a collector of art, while the interest of dentists is to assure us that we would be happier if we devoted most of our free time to oral hygiene. Each field welcomes any new idea that promises to expand its hold on societal resources. In addition, even if there were no selfish, material reasons involved, each field would still push for the implementation of new ideas in its domain, regardless of long-term consequences. A person who has worked for years within the limits of a narrow specialization naturally believes that new developments in his or her domain are the most important and therefore take precedence over developments elsewhere. It is difficult to convince a physicist who has devoted a lifetime to high-energy physics that advances in nuclear technology should not be supported all the way."
...
"Scientists recoil in horror at the thought that anyone else should decide what is or is not good science. A person who has been awarded a Nobel Prize in physics has almost no choice but to believe that he or she is heir to the only possible way of studying the world. To paraphrase Voltaire, he naturally believes that his is the best of all possible sciences, and therefore that any attempt to question the inevitable unfolding of physics by physicists is an anti-intellectual attack on the integrity of science. Each field expects society to recognize its autonomy, yet each feels in the last analysis accountable only to itself, according to the rules of its own domain. For all of these reasons, it is useless to expect fields to monitor their own creative ideas in terms of the long-range public good."
...
"What is needed is a self-conscious effort to establish priorities and to use something like an 'evolutionary impact analysis' as one of the bases for the social endorsement of new ideas. A policy of this type should not result in any kind of philistine thought-policing. Artists should be encouraged to follow their muse, scientists should be respected for following a hunch wherever it leads them. On the other hand, why expect society to support novelties that are valued within a given field but may harm the commonwealth? The greatest art, East or West, was not produced when the artists set the agenda, but when patrons insisted on certain standards that benefited them. Patrons wanted primarily to be admired by the public, so the art they demanded had to appeal to and impress the entire community. In this sense, medieval and Renaissance art, commissioned by popes and princes, was in reality more democratic than it has become since the art world gained the power to separate itself from the rest of society—as a field with its own peculiar tastes and criteria of selection. "
"It admittedly would be more difficult to achieve a public evaluation of scientific creativity. In most scientific domains the frontiers of knowledge have moved so far beyond the grasp of laypersons that only those within the respective fields can be expected to make any sort of informed decision. But it is probably the case that within each field there are enough individuals with both expertise and a sense of the public good who could be deputized to serve the interests of society."
"Currently research grants are evaluated in terms of either the priorities set by the field or the political agenda of the administration disbursing the funds. Perhaps it could be possible to establish a sort of civil service above party politics and disciplinary fashions, composed of those who aspire to be “good ancestors,” as Jonas Salk called them, and who would be willing to represent the claims of evolution when assessing whether scientific advances should receive social support. Inevitably such a group would be composed mainly of older individuals, and therefore it would be open to criticism from younger colleagues who are more concerned with advancing their own scientific careers. On the other hand, the probability for dispassionate wisdom is greater among those who have had more, and more varied, experience and who can see their expertise in a broader context—and these in turn are likely to be older persons. Yet our society expects very little from its elders. This might be one important contribution of seniors that will benefit everyone."
Csikszentmihalyi, Mihaly. Creativity: Flow and the Psychology of Discovery and Invention (Harper Perennial Modern Classics 1996) (p. 323-6).
The fact that the specific issue Ioannidis pointed out in 2005 is not listed might indicate that we're still on the wrong path and will be writing this again in 20 years.
According to Forbes magazine, the United States has 902 billionaires. How about persuading one to run a test of whether sizeable grants to early-career researchers, enabling them to run their own labs for, say, five years, would generate the innovations that people claim that the bureaucratized NIH does not?
This article comes at the perfect time, considering teh critical need for innovation in science. How can we implement better algorithms for funding truly novel ideas? Your insights are incredibly valuable.
A problem with trying to fund very creative new work is that it will have a much higher failure rate. Of course, bad research generally will also have a higher failure rate. How to choose which projects to fund in light of this--- do your past failed grants count against you?-- is tough.
A Nobel laureate much attacked before his ideas were accepted once told me "only your home runs count."
I kind of agree, but in practice, I think that bad research has a higher success rate, because 1) people pick marginal issues to research where the answer is already fairly clear; 2) people p-hack their way to success; or 3) people commit outright fraud to show success.