The Meta Philosophy of Science

I began to form some hypotheses about what was wrong with these other bodies of knowledge [outside astronomy], and why. I particularly noted a regular practice of not re-examining the fundamental assumptions underlying a theory once it gained "accepted" status, almost no matter how incompatible some new observation or experiment might be. And I saw powerful vested interests in a "status quo" develop around certain accepted theories.

It gradually became clear that a lot of people had a lot to lose if an accepted theory or practice were challenged: the authors of the original theory, whose names had become well-known; all those who published papers which reference or depend on the theory; journal editors and referees who have made decisions or criticized other works based on a theory; funding agencies which have paid for research which pre-supposes a theory; instrument builders and experiment designers who spend career time testing ideas which spring from a theory; journalists and writers whose publications have featured or promoted a theory; teachers and interested members of the public who have learned a theory, been impressed by the wonder of it, and who have no wish to have to teach or learn a new theory; and students, who need to find a job in their field of training.

It has been my sad observation that by mid-career there are very few professionals left truly working for the advancement of science, as opposed to the advancement of self. And given enough people with strong enough interests, professional peer pressure takes over from there. Peer pressure in science, as elsewhere in society, consists of alternately attacking and ignoring the people who advocate a contrary idea, and discrediting their motives and/or competence, in order to achieve conformity. Even when it is not effective directly, it is usually successful at ensuring that the contrary person or idea gains few allies, and remains isolated. In short, those who may suspect the need for a radical change in an accepted theory have no interests or motivations as strong as those supporting the status quo. And members of the former group usually lack the background and confidence to challenge the latter group, who are the "recognized experts" in the field and well-able to defend their own theories.

As if there weren't already enough inertia to major changes of models, I see yet another phenomenon -- new to our era of rapid progress in science -- which militates against change even in the face of overwhelming need for it. Few scientists consider themselves qualified very far outside their own areas of expertise. Since each expert can account for only a small portion of the data dealing with a model, he defers to the other experts to support the model in other areas. Few, if any, scientists have the breadth of knowledge to see the full picture for a given model. So the model remains supported because many individual authorities support it, none of whom have the expertise to criticize the model overall, and all of whom have the utmost confidence in the others collectively. Authorities can continue to multiply indefinitely, with no one taking responsibility for integrating all their combined knowledge. As a result, the existing models get perpetuated regardless of merit or the extent of counter-evidence, because "so many experts can't all be wrong." Thus each expert is persuaded to force-fit his own data into the accepted model, oblivious that the others are doing the same.

However, I had learned by then to start being more open-minded toward new ideas, no longer dismissing them out of hand without strong enough reason that even the idea's proposer could understand. Whereas before it was rarely "worth my time" to deal with proposed new ideas, I now felt quite the opposite. This was chiefly because even in the process of proving that a new idea was false, I learned a great deal about the fundamentals underlying the challenged theory. I came to see the soft underbelly of many theories with a tough outer shell. I found a lot of unsuspected weaknesses.

The first challenging new idea which I entertained as seriously viable was P.A.M. Dirac's proposal of the variability of the universal gravitational "constant." I performed a test of the idea using observations of the Moon's orbital motion around the Earth, and obtained results which supported Dirac's theory and seemed to be statistically significant. This experience led me to realize how fragile were the assumptions underlying the Big Bang and other theories of cosmology, when even the constancy of gravitation, the most important force in shaping the large-scale structure of the universe, had been called into question. And I saw that very few of my colleagues were taking seriously the idea that anything could be wrong at such a fundamental level. Their attitude was understandable, but unscientific.

From my disturbing experiences with the insubstantiality of fundamentals in other fields, I learned how I could sometimes spot the bad accepted theories from a combination of their strangeness, a certain lack of providing true insight into the underlying phenomena, and a continuing need to do "maintenance theorizing" to patch the theory in ever stranger ways as new data became available. I later added "derivation from inductive reasoning" as additional grounds for holding a theory suspect. Many of the accepted astronomical theories in use today are "suspect" by all these criteria. I also learned how to proceed when one encounters such a theory: Revert to the principal investigative tools of science and scientists, by means of which we try to separate good theories from bad ones.

These are embodied in the Scientific Method, a process that involves competitive testing of all ideas. Most scientists understand, at least abstractly, the importance of testing. The part they have forgotten, or were never taught because too many major theories in too many fields would be called into question if it were, is controls on testing. This is the step in which the test is designed in such a way that the expected outcome, also called the “bias of the experimenter”, cannot influence the actual outcome. Instead, it has become common practice to question or challenge data that leads to an unexpected outcome while not even checking data or procedures that give the expected result. Even more common is an ad hoc patch to the idea being tested to accommodate the outcome. Naturally, such a patch completely invalidates the test, and requires some independent test with new data. But all too commonly, the result of the original test is cited as evidence supporting the patched idea. Such is the state of mainstream science today.

2002/05/05