I read an interesting article earlier today about the merits of moving away from defining something as 'scientific' if it makes falsifiable predictions. After all, astrology, tarot cards and palm readers make falsifiable predictions, but are not considered to be 'scientific'.
The alternative argument is to build up a body of evidence, and use Bayesian mathematics (
http://en.wikipedia.org/wiki/Bayesian_probability) to weight competing theories based on the probability that they successfully explain the observed phenomena. This has the advantage over conventional 'falsification'-based science as (ironically) illustrated by the inventor of the technique, Karl Popper (
http://en.wikipedia.org/wiki/Karl_Popper), by the 'black swan' thought experiment.
The idea is you have a theory that all swans are white - the only way to test that is to look at every single swan, but however many you look at there's always the chance you missed one, so you can never 'prove' the hypothesis. However, a single black swan observation successfully
disproves the theory. The reason this is problematic for science is that there is no single theory in existence which successfully explains all observed phenomena (this would be a 'Theory of Everything'), so it makes more sense to take all the observed evidence (suspected number of swans, number of white swans observed, number of black swans observed, etc. etc.) and mathematically weight the competing theories based on which one most successfully describes the observations. Now if we see one black swan in a million, the theory that 'all swans are white' is clearly not completely correct, but certainly preferable to 'all swans are black' - the distinction is quantified, not merely qualified.
I believe this technique can be applied to the RE/FE debate... discuss.