I think David Hume said something more or less like this when discussing the likelihood of miracles; that if you witnessed a miracle, you ought to conclude you were insane.

I am not sure I buy into this. For one thing, I see a problem with falsifiability. If there is nothing that I could see to convince me that magic might work, I am not objecting to the reality of magic on rational grounds, but as a sort of knee-jerk. It's like the doubleplus loony creationist types who think the devil planted archaeopteryx.

There are reasons I think magic in the Harry Potter sense is not true, reasons that could be argued against (e.g., show me a plausible medium for magic to be carried in). I don't think it would be very rational to make it sort of... axiomatic that magic is false. That seems to in fact be the attitude Eliezer is criticizing in the character of Harry's father.

So yeah, some probability mass goes to the "hallucination/insane" hypothesis, but not very much. Most goes to the "I don't know what's going on here at all, but she did just apparently turn into a cat" hypothesis.