Karl Popper famously said that the criterion demarcating science from nonscience was falsifiability. Scientific theories and hypotheses are falsifiable; it is always possible to cite specific observations that would prove them wrong. Dinosaur fossils found in Tertiary strata would falsify the claim that the K/T extinction event gave dinosaurs the coup de grace. Nonscience, on the other hand, makes unfalsifiable claims. For instance, daily “sun signs” astrological claims are so loosely expressed that it is hard to see what would count as a decisive falsification. “You will have a day full of opportunity tomorrow” is compatible with almost any kind of day in which you are neither dead nor comatose.

The discussion in the comments section of another post started on the subject of Popper and falsifiability, but, alas, has become rancorous. In an attempt to shed light without (I hope) adding to the heat, let me quote from the sixth chapter of my book It Started with Copernicus (Prometheus Books, 2014, pp. 293-295):

Does falsifiability give us a sound demarcation criterion for distinguishing science from nonscience? For instance, when something like YEC [young-earth creationism] comes along that challenges established science and claims scientific credentials for itself, can we deploy the falsifiability challenge to discredit it and consign it to the trash can of pseudoscience? Here it is important to make a distinction between the intransigence of scientists and the unfalsifiability of a theory.  It may well be that proponents of YEC will dismiss, distort, or ignore all contrary evidence.  They will thereby defy Popper’s characterization of the true scientific attitude whereby bold conjectures are made and equally bold attempts are made to refute them.  Defenders of YEC seem to be interested in buttressing, by any means possible, a dogma dictated by their fundamentalist convictions.  To be fair, though, even the most indisputably genuine scientists can be awfully pigheaded and just as inflexibly attached to their own theories.  There is a story often told about Einstein that he was informed by an eager graduate student that his theory of general relativity had been strongly confirmed by observation.  Einstein reportedly shrugged.  When asked by the student what he would have done had the observations gone against his theory, Einstein supposedly said: “Then I should be sorry for the Dear Lord.  The theory is correct.”

Falsifiability has to be a logical, not a psychological criterion.  It must be a standard that scientific theories, and only scientific theories, can pass.  The fundamental problem with using falsifiability as a demarcation criterion is that it is impossible to formulate that criterion so that it is not either too loose or too strict.  A demarcation criterion is supposed to be a gatekeeper, like one of those big guys with the velvet rope at an exclusive gala event.  He is to let in only the invited celebrities in and keep out the hoi polloi like you and me.  A demarcation criterion must admit only genuinely scientific theories and keep all others out.  However, if “falsifiable” is construed too loosely, all sorts of pseudosciences might pass the test.  A proponent of YEC, for instance, could say “Sure, I will believe in evolution if you produce a missing link that is precisely in the middle between apes and humans in all of its anatomical features.”  Evolutionary theory does not predict such half-and-half missing links; transitional creatures are mosaics with some derived features and some ancestral ones.  Still, the creationist would meet Popper’s criterion, loosely construed, fair and square.  He has mentioned a possible discovery that would conclusively falsify his evolution-denying theory, so his theory would have to count as scientific.

On the other hand, if we make the conditions for falsifiability too strict, we will rule out some theories and activities that are unquestionably scientific.  Practically every theory clashes with some evidence or observation.  Newtonian celestial mechanics famously clashed with two well-known observations: The orbit of the planet Uranus was not what would be expected on the basis of Newtonian theory.  Also, there was the problem with the precession of the perihelion of Mercury.  A planet’s “perihelion” is the point of its elliptical orbit where it is closest to the sun.  Mercury’s perihelion “precesses,” i.e. it moves from one point along the orbit to another.  According to Newtonian theory this should not happen.  These anomalies were eventually understood.  The perturbations in Uranus’s orbit were due to the presence of another planet, Neptune, which was eventually discovered.  The precession of Mercury’s perihelion was later explained by Einstein in terms of relativity theory.  However, during the many years that these anomalies went unexplained, should Newtonian celestial mechanics have been regarded as unscientific because it was falsified by observation?  Surely, if any theory has ever been scientific, Newtonian celestial mechanics was.

A criterion that cannot walk the fine line between being too strict and too lenient is not a useful tool for demarcation.  In fact, almost all philosophers of science now agree that there is no single criterion that distinguishes science from nonscience.  Science is just too multi-faceted and complex an activity to characterize or categorize simply.  I think that all we can really say is that a good theory will have many virtues and few vices and a bad theory will be the opposite.  Sometimes a theory has such egregious vices and so few and minor virtues that it is rightly stigmatized as “junk science,” or “pseudoscience.”