July 12, 2017
Scientific research within a given subfield generally proceeds in three roughly distinct stages. In the least mature stage, scientists follow their collective noses. They make measurements and observations on related systems and try empirically to find correlations among them that may lead to a more coherent understanding of their behavior. Many recent reports of a lack of reproducibility of scientific results are dominated by findings in this early stage of research, when scientists are often not aware yet of all the conditions that affect a system’s behavior. Attempts by independent scientists to replicate results may then suffer from unwitting differences in conditions not yet realized to be relevant.
In the second stage, scientists develop models that can account for a variety of observations and that have predictive power. Experimentalists then make measurements to test those predictions and deviations are used to further tune the models.
In the most mature stage, typified by nuclear and particle physics, those models are combined and perfected under the umbrella of a consensus theory based on a small set of fundamental principles and postulates, which can account for a wide variety of observations based on a modest number of parameters whose values are to be established by comparison with experiment. The theories are, ideally, falsifiable: they make predictions that are so directly tied to their fundamental postulates that reproducible, observed deviations from those predictions are taken as evidence that the theory either requires important modifications, perhaps signaling new scientific breakthroughs, or must be rejected. Even Quantum Electrodynamics — the most successful theory out there, accurately agreeing with experiment to a part per billion or better — is worth testing ever more precisely, because a tiny but reproducible deviation may be a thread that, if pulled, reveals new physics.
Real scientific facts are established during the above research stages by observational results successfully replicated by independent researchers using independent techniques. They can pass the scrutiny of rigorous peer review by independent experts. In the most compelling cases, their interpretation is supported by rigorous theory that explains a wide variety of phenomena in a self-consistent and falsifiable framework. The systematic reliance on replication of experimental findings and on precise testing of models and theories makes the scientific method self-correcting over time. Pseudo-science does not share this feature.
There are inevitably legitimate disagreements among scientists that arise during this process, over the reliability and interpretation of experimental results, and over the content and assumptions of theoretical models. These disagreements are a central part of the scientific method. Debates over these disagreements are carried out in the peer-reviewed scientific literature, occasionally supplemented by a small number of “neutral” blog sites. An example of the latter for climate science issues is judithcurry.com. The debates culminate in agreements regarding crucial additional measurements or improved treatments of model components. Such debates lead over time to correction of earlier errors, to more complete models or theories, to settlement of disputes and to the emergence of consensus about scientific facts. Consensus does not require unanimity — there often remain a small number of holdouts who refuse to be convinced by arguments that compel the vast majority of scientists working in the field.
But there is also an industry of science denial, whose goals are political management of scientific facts and the purposeful creation of confusion in public minds. The deniers use methods — cherry-picking and misrepresentation of data, models and claims — that are not aimed toward settling disputes and reaching consensus. Deniers’ claims are almost never presented in legitimate peer-reviewed scientific literature. They are often coupled with personal attacks against the scientists whose results they aim to refute. They use approaches from a well-worn and recognizable “toolbox” that will be described and illustrated in future blog posts.
The fundamental difference between scientific skepticism and science denial is that skeptics are willing to change their view when presented with compelling evidence, while deniers are willing to misrepresent the evidence to reinforce their original view. This can be illustrated with a couple of examples.
The first comes from climate science, where denialism remains strong and, at this moment, politically ascendant. Professor Richard Muller, a physicist from the University of California, Berkeley used to be one of the most prominent skeptics of climate change among scientists with prestige. He doubted that global warming was real and, subsequently, that human activity was the primary cause. But he followed the scientific method to address his doubts, forming the Berkeley Earth Surface Temperature project to improve on temperature measurements and data analysis techniques, thereby expanding the relevant database and extending it backward to earlier times. Those measurements and analyses convinced him by 2012 that, despite his earlier skepticism, the conclusions of the U.N. International Panel on Climate Change (IPCC) were correct: the planet was indeed warming markedly and the warming was almost certainly due primarily to human activity. He presented an account of his conversion in a New York Times op-ed piece. (He remains skeptical about some claims — e.g., that violent storms are occurring far more frequently as a result of global warming — that are so far not definitively supported by existing data.) However, climate science denial from politically motivated groups such as the Heartland Institute, and from some of Trump’s cabinet secretaries and some prominent members of Congress, remains unmoved by any such scientific evidence.
The second example comes from the field of evolutionary biology. Michael Behe is a professor of biochemistry at Lehigh University and a proponent of “intelligent design” as an alternative to evolution of the species by the mechanisms of random genetic mutations and natural selection. In his 1996 book Darwin’s Black Box: The Biochemical Challenge to Evolution, Behe provided the argument of irreducible complexity, on which creationists have seized to support their ongoing crusade against macroevolution. Behe’s view is that current evolutionary theory cannot account for certain complex structures, particularly in microbiology. On this basis, Behe argues that such structures were “purposely arranged by an intelligent agent.”
As a scientist, Behe has argued that his concept of intelligent design is, contrary to the claims of his opponents, falsifiable. He proposed laboratory experiments that could, in principle, demonstrate that complex microbiological systems, such as the flagellum that bacteria use for locomotion, arise by random mutations and natural selection over generations of development of bacteria beginning without a flagellum. He concluded: “If that happened, my claims would be neatly disproven.” Despite the fact that Behe’s falsifiability logic is flawed — how could one prove that mutations that occurred were random, rather than arranged by some unseen “intelligent” hand? — many other biochemists have indeed produced strong evidence of evolutionary pathways for all of the systems Behe used as examples of irreducible complexity. The flagellum, for example, appears to have evolved naturally from systems that initially served different biological purposes (e.g., secretion) in bacteria.
Furthermore, in his 1994 book Darwinism: Science or Philosophy?, Behe asked “If random evolution is true, there must have been a large number of transitional species between the Mesonychid [a land-dwelling whale ancestor] and the ancient whale. Where are they?” Less than one year after Behe’s statement, no less than three transitional species between Mesonychids and whales were confirmed. In recent writings, Behe makes no mention of the issue of land-dwelling whale ancestors.
So Behe’s theory of irreducible complexity is, indeed, falsifiable and has been falsified. The scientific method works, but it does not convince science deniers. Behe himself has not given up his claim of irreducible complexity, though he has softened it and made it more compatible with Darwinian evolution over the years. It can be psychologically challenging for a scientist who stakes his/her reputation on an alternative theory to admit freely that the subsequent evidence proves him/her wrong. But the proponents of intelligent design continue to simply ignore the falsifying evidence altogether, and to use irreducible complexity as the centerpiece of their continuing denial of macroevolution.
Both the climate science and macroevolution cases will be explored in more detail, along with a number of other issues, in later blog posts.