Science and Replicability
by Christian Y. Cardall
In a discussion of natural and unnatural methods of childbirth, it was asserted that “Scientific data is intended to be replicated—indeed, must be replicable—and objective”. I don’t have anything to say about childbirth, but I would like to point out that depending on what is meant by “replicated,” the above formulation has the potential to give an overly narrow or even misleading notion of science.
Being able to set up and perform an artificial experiment at will is less important to the definition of science than that data be publicly available and that there be sufficient objective control on hypotheses. It is great when a repeatable artificial experiment is possible, but such historical disciplines as geology, paleontology, in some cases astronomy and astrophysics, and so on depend in part on events (and sometimes even observations) that are not repeatable: we will not observe a real-time magnetic field reversal in our lifetime, dinosaurs will not go extinct again, SN1987A will not explode again, and so forth. Nevertheless the public shareability of recorded observations of these events and adequate cross-examinations to control interpretations justify calling their study ‘scientific.’
Statements like ‘many different people can analyze relics of the event and come to the same conclusion’ or ‘more dinosaur bones could be found’ are arguably instances of ‘replication’ in a different sense than repeatable, artificially controlled experiments; but I resist considering these as alternative ways of satisfying some putative overarching scientific ‘replicability criterion,’ for the first may simply collapse to the more useful notion of public transferability, and the second is obviously too restrictive: the study of dinosaur bones would remain scientific even if no new ones were ever found.
Finally, there is a danger in overemphasizing repeatability without sufficient attention to public shareability and objective controls (in this connection I appreciate the use of the word “objective” in the above formulation, I just think it deserves more emphasis—hence my present ascent to the soapbox). For example, it might be observed that instances of Mormon testimony are eminently repeatable (like the McDonalds’ slogan, Millions and Millions served), while the generation of Earth’s biodiversity by evolution cannot be replicated in the size or time scales of human laboratories and lifetimes. With an overemphasis on replicability some might therefore be tempted to call Mormon testimony ‘scientific’ and the study of evolution ‘unscientific.’ (I hasten to say that the commenter I quoted above didn’t say anything remotely like this; perhaps there are very few Mormons who would consider testimony a form of scientific knowledge, though Alma 32 might tempt some to say so. I am simply using her statement as an occasion to take off on an unrelated tangent that may be of little interest to anyone but myself.)
The problem here is that the more important consideration than replicability to the question of whether testimony or evolution are scientific is whether the experiences motivating belief in the respective principles are publicly transferable and controlled by objective data against plausible alternative explanations. In the case of testimony the answer is ‘no,’ and it must therefore be considered unscientific, notwithstanding its wide (if far from universal) replicability. To the extent the answer is ‘yes’ to various aspects of evolution, those aspects are scientific, in spite of the fact that we have only a single precious example of the historical unfolding of a great pageant of life.