The canonical conception of knowledge is that it is “justified, true belief”.  As canonical as this view is though, it’s also pretty at this point that it’s wrong; this definition of knowledge has some counter-examples.  Nevermind the question of whether or not the concept of knowledge needs a definition that has no counter-examples, just bear with me; for now at least I can say it’s worth a good try.

The Gettier problem (or problems, there are two subtypes of it) is the best known such counter example.  Personally, I have a working idea of how to deal with this one.  But there’s a special case of it that I’m all together stumped on.  Let’s say that you go to the doctor if you have a certain disease.  He looks at his clipboard tells you no, and you would be perfectly justified  in believing him, wouldn’t you?  It’s perfectly reasonable to take you doctors word on something like that, so you should be justified in believing him.  But let’s say that your doctor’s an asshole.  Let’s say the clipboard is blank, and he’s just guessing.  He’s obviously not justified in believing you’re healthy.

So let’s say you actually don’t have the disease, and you believe him.  You’re belief that “I don’t have this disease” is then true, and you believe it.  It would then follow from the assumption that you’re justified in trusting your doctor regardless of whether or not he himself is justified that you *know* you’re healthy.  Even though it’s complete dumb luck that you happen to be right, and you’re doctor himself (the alleged reason you’re justified in believing) doesn’t know.  This should mean you don’t know in this case, even though the canonical definition of knowledge says you do.

So either you’re only justified in trusting testimony when the person giving it is justified, or the traditional definition of knowledge is wrong.  Pick one or the other, or explain why an another assumption of my argument is wrong.