What Is Science, Anyway?

Dr. Thomas R. Holtz, Jr.

>If we stuck to the facts - just the facts, as Stephenson/Joe Friday would
>have it - then all any paleontologist could do is describe the fossils...
>NOT the creature they came from, but the fossils themselves; morphology,
>composition, placement in the bed, etc. - because these are observable,
>repeatable etc. None of the threads on bolide impacts, stealthy
>egg-predators, swimming mammoths, tripping-up tyrannosaurs, sauropod feeding
>habits, would have any place in such a rigorously scientific discussion.
>One might well ask - what's left to discuss?

In point of fact, the swimming mammoth, tripping tyrannosaur, and sauropod feeding topics have all been the subject of one or more rigorously scientifically discussed papers!

When some member of the net (myself included) show some hostility towards "speculation", what we are really hostile towards are "untested, or untestable, hypotheses".

Science is NOT simply about describing the dimensions and properties of natural things. Science is the practice in which we employ falsifiable hypotheses to examine the natural world. For a hypothesis to be scientific, there must a) be a manner in which we can clearly demonstrate it is UNtrue and b) we must then TRY and prove that hypothesis wrong. (The academic practices of history and linguistics do the same with hypotheses of human events and language).

Hypotheses themselves are a form of speculation (could mammoths swim to Santa Catalina? what would happen if a T. rex tripped at 20 m/s? did sauropods rear up on their hindlegs to reach the tree tops, and then rake the leaves and needles off with their teeth? [and a personal favorite] did tyrannosaurids share a more recent common ancestor with allosaurids or with small theropods?). However, to be scientific, we must start off with some evidence to suggest the possibility (i.e., thinking from the start that mammoths reached Catalina by teleportation isn't particularly scientific, because teleporation itself has not been demonstrated in the supraatomic world).

Pseudoscience and some less-than-rigorous science stop at this point: they present an idea, and some evidence to suggest that idea is true. To be analyzed in a scientific manner, however, you must go to the next step:

Then, we must frame our hypothesis in such a manner in which it could be disproven. This step is also called "experimental design." In paleontology, comparative anatomy, functional morphology, biomechanics, microscopic analyses, phylogenetic analyses, footprint analyses, etc. are the means by which experiments are set up in order to falsify our hypotheses. If it were found that modern elephantids could not swim, or that tyrannosaurids could not stand the acceleration from tripping at 20 m/s, or that rearing would exceed sauropod structural limits and that their teeth showed no sign of wear, or that the most parsimonious distribution of derived characters in a particular data matrix found tyrannosaurids shared a more recent common ancestor with allosaurids than with coelurosaurs, then we would reject the appropriate hypothesis.

If, on the other hand, the results do not contradict the hypothesis, then it is (temporarily) accepted. However, it is then open for anyone else to reanalyze with new data or techniques.

Dinosaurian research has attracted a lot of speculation without attempts at falsifiability (or attempts at falsification). Some particular extinction hypotheses have been accepted even when the speculated events do not match the actual patter of fossil distributions, for example. Of somewhat less significance, some speculation is potentially falsifiable, although the data to test the hypotheses may not be immediately forthcoming (i.e., feathers in dromaeosaurids, which are equivocal on a phylogenetic basis (unless a taxon more distantly related to birds is shown to have feathers) and on a taphonomic basis (no dromaeosaurid has yet been found preserved in sediments capable of retaining feather impressions)).

I hope this helps to clear things up.