There is still some confusion about the idea of pseudo-science. I have long used the following, and it is close to a more accepted definition. The issue is that there are actually three categories, not two. The missing one is: non-science.
Non-science is an assertion that is based with no reasonable underpinnings. In the extreme this can include revealed truth - truth as revealed by a higher entity, or some form of philosophical axiom. My favourite example is the structure of the heavens as dictated by Aristotle. He asserted that the heavens were built from spheres derived from the five platonic solids nested within one another. Indeed a large part of his metaphysics was derived from a set of axioms that amounted to little more than a desire to reflect naive ideas of mathematical perfection into the structure of the universe.
But non-science comes in other areas. I would argue that homeopathy, chiropractic, new age crystal wavers, and all these, are non-science. They claim a scientific status that is wholly derived from some revealed truth. The truth may not be questioned, and the non-science derives from it.
Pseudo-Science is different. Its hallmark is that it initially looks like ordinary science. But it has a curious habit. It starts with a core hypothesis. But in the face of contradicting evidence, the core hypothesis is not disproven. Rather it is added to. New special cases are added to the theory. So whenever an experiment is done, it either validates the theory, or, rather than disproving it, it is used to extend the theory. Pseudo-Science was the driver for Karl Popper to develop his work on science, and in particular his notion of the falseifiable hypothesis. He was particularly concerned with Freudian psychology and Marxist political theory. Both of these claimed a scientific basis, but both suffered from this key flaw. They would never advance a hypothesis that was ever testable properly, because they would always modify or add to the theory in the face of contradictory evidence.
Which brings us to the classical definition of a true science, as proposed by Karl Popper. To be a true science it must have as its basis, a falsifiable theory. That is, not only must you have a theory, but you need to be able to construct an experiment that can disprove it. In many ways this mean that the hypothesis is predictive. If the hypothesis makes a testable prediction, you test the prediction. If the prediction comes about, the hypothesis is validated. If it does not, the hypothesis is disproven. If you can’t come up with such an experiment, the hypothesis is essentially useless.
Non-science might have a falseifiable hypothesis. But since it is revealed truth it is axiomatic that any contradictory evidence is simply wrong.
I like to describe the difference in the three categories like this. It is all about how much information is in the theory. True science has a very small amount of information in it. The less information the better. For instance general relativity describes the way gravity and space-time works for (nearly) every part of the universe. A theory of space time that only applies on the Earth (like Aristotle’s) contains vastly more information - because it needs to contain all the information that limits is application. True science carries the information that describes its hypothesis, plus the current known bounds of its validity. As research continues these bounds may reduce, but in principle should never vanish. There is always some room for doubt.
Pseudo-Science tends to contain an unbounded amount of information, an amount that is equivalent to the sum of all the experiments done. Each experiment adds new special cases and get out clauses that protect the core hypothesis. In the end it is nothing more than the history of all the things that actually invalidate the hypothesis, plus the core hypothesis. It is not predictive, it can only tell you about what has been tested, not about what you might discover with a novel test. There is no room for doubt.
Non-science contains only the information that represents its revealed truth. There are no bounds to known validation, because it may not be questioned.
