October 17, 2024
By Brandon Long
The Paris Observatory was one of the first federally funded research institutes, and it produced serious scientific output. Louis XIV funded a plethora of high-resolution maps, water tables, and astronomic discoveries. The head of the observatory was an Italian mathematician and astronomer named Giovanni Domenico Cassini. Cassini’s ability as an astronomer earned him part of the namesake for the NASA probe investigating Saturn, which he discovered possessed rings. This began a long tradition of federal funding for scientific research in the West. But would Louis XIV have swooned if Cassini had requested funding for… unobservable stuff? Here I will argue public money cannot be justifiably spent on esoteric and experimentally intractable theories with unobservable variables.
Physics funding, however, as Sabine Hossenfelder and others argue, often supports unobservable theories. Billions of U.S. dollars and Euros fund particle accelerators motivated by theories with unobservables, such as string theory and supersymmetry, that may contain entities not observable, even in principle, directly or indirectly. Even positivists like Hempel could not find an empiricist basis to justify such theories as scientific ones, even after revising the empiricist criterion of meaning in 1950.
While I will not go into the details, I believe that Hossenfelder’s arguments against what she calls “aesthetic” criteria in theory and funding choice are off the mark. She frequently lumps criteria of “simplicity” in with “aesthetic” criteria. Serious thought has gone into giving some epistemological basis to simplicity in empirical theories. However, her critique of funding for aesthetic reasons or simplicity reasons without any observable data is quite strong. While I am no theoretical physics expert, string theory seems less empirically testable and less experimentally tractable than supersymmetry. To avoid devolving into an essay on physics, what does any of this mean for funding scientific research?
Well, funding decisions should focus less on claims of human impact (Hossenfelder would say claims of exaggerated bullshit) and more on how testable theories are. Funding decisions consider testability, but as the evidence of funding for string theory programs shows, perhaps not enough. I must admit there are many more wasteful ways to spend money than funding interesting people doing interesting theoretical work, but string theory still pulls grant money, which could be a colossal waste of public funds. While it is hard to pin down exactly how much public funding is spent on string theory, a cursory grant search is eye-opening. Funding decisions should be more skeptical of theories claiming phenomena that are unobservable in principle.
I argue asking what unobservables are “doing” is a key question about whether a specific theory or study is doing science. Other disciplines like economics often receive criticism for being unscientific. A key question from economist James Buchanan's Cost and Choice is whether cost and choice are subjective. If they are, it is hard to understand how economics can be considered a science. Some say yes, it is science, and some say no, it is not science. This debate is too nuanced to get into, and many distinguish sharply between social and natural sciences. However, I argue that the testability of cost and choice provide a good reason to count economics as science, and a science worthy of public funding.
Notice there are unobservable variables in physics that are easily manipulable. Cost, choice, centers of gravity, and black holes are highly experimentally tractable. String theory simply is not and may never be. Running slightly counter to expectations about the experimental tractability of each discipline, federal grant liaisons should not believe that a physics theory is testable just because it is physics. In reviewing economic research proposals, federal grant liaisons should consider how experimentally tractable the unobservable variables are, irrespective of discipline.
First, what theorists claim they are doing science when researching unobservables like choice and cost?. Scholars like Vernon Smith and Bart Wilson’s (2019) Humanomics use Adam Smith’s Theory of Moral Sentiments (TMS) to explain why traditional game theory fails to predict higher defection rates in experimental game theory. Humanomics takes Smith’s “Impartial Spectator” as our interpretation of other people’s sentiments towards our actions, which causally affects our choices. The tension between the desire to objectively measure subjective human choice is very apparent in such a project. Economic research into choice and cost by theorizing about impartial spectators faces the same “unobservable” posits in its theory as I indicated string theory does.
Even granting that the posits of string theory and economical choice and cost are unobservables in principle does not mean economics should be defunded. Many nuances beneath the surface lean favorably towards economics. We have very experimentally tractable ways to ask questions about cost and choice in human subjects. Further, we have good reason to think that our choices and perceptions of cost are causally efficacious (depending on arguments from analogy). Unlike string theory, cost and choice have indirectly verifiable and easily manipulable variables. Thus, cost and choice are more experimentally tractable than some posits in string theory.
My point is that public money cannot be justifiably spent on esoteric and totally experimentally intractable theories with unobservable variables. Perhaps someone should be working on string theory to determine how experimentally intractable it is. Of course, working out the theories' details may lead to manipulable or verifiable experiments, but we have no epistemic reason to think this is the case. I am not opposed to using money on implausible theories.
Other funding methods are available to continue working on string theory. No matter what academics say about Stephen Wolfram, he does not waste public money. Private donors funding such research is more publicly justified than using federal dollars without real public input. We could pull at whether market allocations are publicly justified.
In summation, we need more incredulous federal grant liaisons who are less interested in the proposed impact and more interested in how experimentally tractable theories are. When string theory or economics makes claims about unobservable variables, the immediate next question of what such posits are “doing” needs to be asked. Using a Kuhnian or Lakatosian framework, we can ask if the theory is seated within a discipline in a renaissance or regress. Disciplines in renaissance make bold predictions that pay off with intuitive and unintuitive findings. Conversely, disciplines in regress make predictions that don’t pay off, and are mostly convenient on the theory side, trying to explain why data does not fit the theory.
In other words, renaissance is when everything is coming up string theory, and regress is when experiments are not turning up productive data. A theory claiming unobservable and untestable variables but are convenient for future funding is a reason to doubt whether the theory should be funded. Of course, string theory could be true and useful, but we don't have good reasons to think that. One may object that successful theories posit unobservables conducive to funding. The difference is that such projects are in a renaissance, while string theory is in regress.
Federal grant liaisons should ask these questions using measurable analysis tools and metascientific data. Cody Mosher at UC Merced is working on such tools to identify stalling and advancing philosophical debates, and those tools can be employed here. Are we funding a project with clear progress towards understanding how unobservables help us understand how humans make decisions, or are we floundering in conceptual space trying to buy another few years of research? String theorists make convenient predictions that are safely untestable, indicating a science that is in regress, and public money should be hesitant to buy projects that are continually heading away not only from observable variables but also from testable hypotheses. Economics, on the other hand, is often described as non-scientific and making untestable claims. Public money should be enthusiastic about theories testing cost and choice in economics because such theories, while having unobservable variables, are testable.
Brandon Long is a Ph.D. student in Philosophy at the University at Buffalo, where he focuses on the intersection of Philosophy, Politics, Economics, and Law (PPEL) and Bioethics.