Truth in Science: Discovery vs. Fiat
That truth in science is discovered, tested, and refined by repeated independent tests of predictions against data might seem uncontroversial to many traditionally trained scientists. Scientific knowledge and conjecture are commonly represented by proposed causal laws and theories that make empirically testable predictions for new situations. Testing whether these predictions hold in reality help us to revise and correct the underlying theories in light of experience.
Learning about how the world works by comparing predictions to observations has been a hallmark of modern scientific method since at least Galileo. In my lifetime, it has led to scientific discoveries that overturned the then consensus views on cosmology (Palmer 2011), the Standard Model of particle physics (Murayama 2002), and, arguably, the Central Dogma of molecular biology (Koonin 2012).
But the principle that sound science requires testing falsifiable predictions against observations has recently come under attack by influential academics and science-policy workers who advocate replacing traditional, empirically grounded, science with a presentation of consensus beliefs of selected experts as “scientific truths.” These individuals believe that policy-makers and the public should accept such views based on the authority and credibility of those advocating them, and that anyone who refuses to conform should be ostracized for not accepting — even if the asserted “truths” have not survived having their falsifiable predictions tested against data.
This new type of “science” is considered by its devotees a useful tool of rhetoric and political persuasion. It is well expressed in a recent opinion piece in Scientific American entitled “The idea that a scientific theory can be ‘falsified’ is a myth – It’s time we abandon the notion” (Singham 2020). The piece argues that “falsification cannot work even in principle [because] a theoretical prediction is never the product of a single theory but also requires using many other theories. When a ‘theoretical’ prediction disagrees with ‘experimental’ data, what this tells us is that that there is a disagreement between two sets of theories, so we cannot say that any particular theory is falsified. …[However,] Science studies provide supporters of science with better arguments to combat [their] critics, by showing that the strength of scientific conclusions arises because credible experts use comprehensive bodies of evidence to arrive at consensus judgments about whether a theory should be retained or rejected in favor of a new one.”
The proposal that “science” used in regulation and policy-making should rest on the consensus judgments of credible experts has gained traction in the United States and European regulatory communities, notwithstanding long-standing findings from psychologists that such judgments are usually prone to biases that make them unreliable guides to scientific truth (Kahneman 2011, Chapter 22). Computer models stocked with consensus-driven assumptions are now widely used to simulate benefits from increased regulation based on consensus judgments about how regulations should reduce risks, even when empirical (real-world, as opposed to simulated) data repeatedly show that the predicted benefits do not actually occur (Cox 2020).
I am delighted to join the Center for Truth in Science at a time when the struggle between traditional data-driven science and new authority-driven consensus “science” is starting to have major implications on the national and international stage. To me, it seems plain that the philosophy of consensus science is badly misguided, and that testing beliefs against data can and should remain the foundation of sound science and sound policy decisions. This traditional scientific method serves both scientific progress and the public interest more surely than consensus -driven approaches that protect beliefs by not requiring them to be tested against data, but that thereby lose the possibility of learning from data when beliefs are mistaken.