The idea is that you keep a theory until you can prove it wrong.

That kind of thinking is why the scientific method and modern science is full of crap. "Prove me wrong" isn't a valid argument. In politer circles anyone arguing for the existence of ghosts with "prove me wrong" would get laughed out of the room. You have to prove yourself *right*.

Wrong. If you test the ghost theory, your test will fail and your theory goes away.

You cannot "prove" a theory right. You can provide corroborating evidence that suggests that your theory works. Multiple theories will give rise to the same results of an experiment. By doing more experiments to distinguish between two theories, you can take away theories (even theories unknown.)

For example, let {A

_{1},A

_{2},...A

_{n}} be the set of all potential theories about of a phenomena that satisfy known data. You say "I think theory A

_{1} is the correct theory. Other people's theories range from {A

_{2},...A

_{m}} (m<n). I will perform experiments whose results will be able to confirm A

_{1} as the correct theory." So you then produce results that say A

_{1} predicts much better than any theory in {A

_{2},...A

_{m}}, refuting {A

_{2},...A

_{m}} along the way. According to a zetetic, A

_{1} would be "true" (it matches all observed values).

However, there are other, unknown, theories in the range {A

_{m+1},...A

_{n}}. It is easily possible, and even likely, that one of these unknown theories is correct. However, If A

_{1} predicts the same results as the "true" theory, A

_{true}, under all previous testing conditions, you could never know that A

_{1} is not the correct theory, even though it is zetetically correct.

However, the scientific method recognizes this inability to declare something "100% true". That is why theories must be tested in more and newer ways. Because once someone (by chance) runs into an experiment where A

_{1} gives a different result than A

_{true}, we know that A

_{1} is not the correct, (or at least not the complete), theory. Then, a competing group of theories {A

_{m+1},...A

_{p}} (m<p<n) can be formed and tested, starting the cycle over again. Thus, according to science, A

_{1} would only be the "best theory we have right now."