Editor’s Note: This article was originally published on the author’s Substack, Science Is Not The Answer, on September 10, 2024 and is crossposted here with permission.
One reason there is so much Bad Science, as I have said many times, is that there is too much science. Rather, too much activity in the name of science. There are too many people calling themselves scientists, and too much money spent in the name of science.
Except for the rare scientists who labor for themselves or for those who have a direct interest in their output, which is to say, those who have skin in the game, most scientists almost never suffer for poor work. Indeed, they can even put out the most atrocious theorizing, as you and I have seen together these many years, as long as that theory aligns with whatever narrative is current in their individual fields. Indeed, this is the very history of science.
How nice would it be to dock the pay, even retroactively, of scientists whose models and theories turn out to match Reality badly, or not at all? Hit ’em where it hurts. No better political strategy. And where does it hurt the most? In the wallet. Not in the theory. Fear of failure would then be real, and gripping. It would cause a dramatic fall in the rampant over-certainty from which we all suffer.
The National Association of Scholars has some ideas along these lines.
They have put forward a series of suggested Acts that Congress might adopt to kick scientists’ butts and bring them, at least to some extent, back to Reality.
Now as we all know lobbyists write the laws for Congress, the members of that august body being too busy rooting up funds for their next campaigns. Bowing to this truth, the NAS helpfully drafted a slew of Acts to put scientists’ models to the test or else no funding. This is the way.
Take the proposed Mathematical Modeling Reform Act. It has such items as “Preregistered Mathematical Models. The {Centers for Disease Control and Prevention} after {Month Day, Year} may include only preregistered and publicly accessible mathematical models in all forecasts, policymaking, and interventions.” (You have to love how easy they make it for Congresspersons: just fill in the blanks.) What’s a model? The Act gives the definition:
‘Preregistered mathematical models’ shall be defined as mathematical models whose registered report, including associated protocols, prospective validation practices, specified, agreed-upon rules for judging success and/or the need for recalibration, computer codes, data analysis scripts, and statistical analyses are archived on an online digital repository in a manner sufficient for continuing independent inspection, replication, reproduction, and/or verification. This online digital repository shall have archival and accessibility capacities at least equal to those possessed by the Open Science Framework on December 31, {Year}.
Naturally, the phrase prospective validation practices is near and dear to my heart. This is the Predictive Way I have been rattling on about for years, shouting into the void pleading for ears to hear. See the Class for why!
Then there is the Weight of Evidence Act which requires agencies to spell out “‘weight-of-evidence’ judgments with transparent and explicit rules with consistent definitions, consistent inclusion criteria, consistent exclusion criteria, and consistent weighting criteria”. The Reproducible Policy Act mandates federal agencies “shall within five calendar years of {Month Day, Year} rescind all significant regulatory actions and highly influential scientific assessments that the commission described in subsection (4) determines no longer possess sufficient scientific justification.”
The Replication Studies Funding Act would create an Office of Reproducibility Research, which I do not love because of the marriage of Parkinson’s Law and Pournelle’s Iron Law of Bureaucracy. Give them the chance to create a metric that proves they’re doing their job, and they’ll do so, then maximize it and then reify it so that it becomes The Science. Far better to require putting science used in policy to the test and demand its replication.
On the other hand, some metrics can sometimes be used to good effect. Perhaps that is so with the proposed Negative Research Act: “After {Month Day, Year}, the {National Science Foundation} may not fund any research grant program, or provide dedicated funding to any scientific discipline, which reports more than 65% positive results during five consecutive years of funding.”
The idea is that only “good” news is welcomed by peers in peer-review, and it’s about time for some bad. The exact specification makes me nervous, however.
“What about DIE, Briggs? They hit that?”
They did, in the Science Depoliticization Act.
Agency Depoliticization. The {National Science Foundation} may not:
-
- use ideological litmus tests in any administrative policy;
- use discriminatory concepts in any training for staff, partner institutions and individuals, or grant recipients;
- encourage, discourage, require, or forbid staff, partner institutions and individuals, or grant recipients to endorse, assent to, or publicly express a given ideology, political stance, or view of social policy in any administrative policy; or
Et cetera, et cetera. This includes no preferential funding for Victims, which ought to cause a lot of howling.
They have a few more. Take a look. Of course, there seems little chance of passing any of this now. But once Dictator-On-Day-One Trump seizes office, as the left insists with all their might will happen, then Trump will simply dictate science must adhere to Reality.
Image by olly — Adobe Stock — Asset ID#: 77601256
What about the Federal False Claims Act?
It was written during the Civil War to address people who were selling shoddy goods to the US Army but applies to anyone who accepts Federal money (i.e. submits a claim) under false or fraudulent terms. It allows private citizens to file suits on behalf of the government (called “qui tam” suits) against those who have defrauded the government — and get a share of the winnings. See: https://www.justice.gov/civil/false-claims-act
There are accepted research methods and things which one is not supposed to do, and while you can’t get them for having research that can’t be duplicated, you *can* get them for not following accepted means and methods, especially when they fabricate data and such.
The False Claims Act is treble damages — having to repay three times what the Federal Government gave them, and I believe this would include the Research Overhead so this could be five or six times the amount of the grant(s). Hence it wouldn’t just be retiring in disgrace but also bankrupt…
Again, if someone is just wrong or the research is inconclusive (e.g. the climate change hysteria), that’s one thing. Likewise there are sometimes legitimate unknown factors that prevent research duplication — that’s essentially how we discovered penicillin. But if people are fabricating data, P hacking, and other such stuff — engaging in fraudulent behavior — why aren’t they facing the consequences of doing such?
Well, what are the scientific accomplishments of William M. Briggs?