Why Did (or Didn't) it Happen?

20Historically Speaking · June 2003 Why Did (or Didn't) it Happen? Stephen G. Brush When I switched from theoretical physics to history, it was in part because I wanted to find out how andwhyscientificknowledgehad been established . Scientists want to describe the natural world and also find outwhatcauses things to happen in thatworld; presumablyhistorians want to undercover causes as well as facts in the humanworld. Tb what extent is this presumptionvalid ? Is studyinghistorylike studying science? Accordingto oneversion of"The ScientificMethod ," proposed bythe philosopher of science Karl Popper (1902-1994), we should use our hypothesis to predict a fact or event thatwe don'talreadyknow. Ifthe prediction turns out to be wrong, we must discard the hypothesis. Ifthe prediction turns out to be correct, that doesn't prove the hypothesis is correct—since some otherhypothesis might have produced the same prediction. It does mean, however, that we can retain our hypothesis—it has been "corroborated" but not "confirmed"—and prefer it to another hypothesis that has not survived such a test. On the other hand, a hypothesis thatis so flexible it can explain anything but cannot make anytestable predictions is notscientific at all. (Popper placed Marxism and psychoanalysis in this category ofpseudosciences.)1 According to Popper, merely predicting the recurrence ofa known phenomenon is not a real test ofa scientific hypothesis: you should predict something that a personwho doesn't know your theory would not expect to happen. You get no credit for predicting that the sun will rise tomorrow, but ifyou predict, as did Svante Arrhenius at the end of the 19th century, that continued burning of fossil fuels will lead to global warming, then otherscientists may eventuallyconclude that your theoryis valid. One might infer that historians are not scientists because they do notjudge their theoretical explanations by their abilityto make successful predictions, while physicists, chemists, and meteorologists are scientists because they do. And indeed this is exactly the kind ofinference thathas led manypeople to conclude thatthe physical sciences are "harder"—they yield more reliable knowledge —than the biological and social sciences, and certainly harder than history, which is not a science at all. There isjustone problemwith this inference : in many(perhaps most) cases, physical scientists donotjudge atheoryprimarilybyits success in makingpredictions, although that maybe a secondaryfactorinpersuadingthem to accept it. After many centuries ofresearch, geologists have not actually produced a theory of earthquakes that is generally accepted and makes accurate predictions about future earthquakes.2 Yet no one, to my knowledge, has concluded from this fact that geology is not a legitimate science. So perhaps we should askinstead: is The ScientificMethod, as defined byPopper, an accurate description ofhow science works? There is also some confusion about what is meantby"prediction." Physicists generally use thewordtoincludeadeduction ofan empirical fact, whether or not the fact was known before the predictionwas published. Ifitwas not known, one speaks of "prediction in advance" or "forecasting" or (in the discourse ofphilosophersofscience) "novelprediction." Formulations ofThe Scientific Method by Popper and other non-physicists usually require thatatheorymake novel predictions, although some of these statements are ambiguous. The linguisticusage ofphysicists suggestsdiattheydon'tthinkitmakesanydifference whether a prediction is novel or not, and in some cases it is quite clear that they don'tthinksuccessfulnovelpredictionsshould count any more, in evaluating a theory, than successful non-novel predictions (deductions).3 More generally, ifone believes the statements of Einstein, Dirac, Eddington, and others, physicists like theories that are simple , universally applicable, logically coherent , beautiful, and consistent with other established theories and laws ofnature; in the long run a theory is expected to make accurate predictions about empirical facts, but it should notbe rejected ifit doesn'thappen to agreewith the latest experiment. Conversely, a theorythatdoes notsatisfythose criteria of simplicity, etc., may be rejected despite its success in predicting new phenomena. Unfortunately, Popper's version ofThe Scientific Method became widely accepted by philosophers and others who have not actually done scientific research themselves. Itiseasytojudge a theorybythe success ofits predictions even ifyou don't understand its causal explanations or ifit doesn't give any reasonable causal explanations atall butsimply extrapolates empirical correlations. The Popper predictiveness criterion has now been adopted by the U. S. Supreme Court (in the Daubert decision) as one ofits rules for decidingon the admissibilityofscientific evidence. In this way...