Like almost all fallacies, it is related to a legitimate form of reasoning. And this particular case is especially tricky, because we very often should rely on authorities. How many people live in France? We’re certainly not going to go count them ourselves. But we can look up the number, and it is absolutely reasonable to trust the numbers we find in reputable sources. (Indeed, even Wikipedia, which is not a reputable source, will very likely suffice for a simple fact like this.)
The appeal to authority is illegitimate (and thus a fallacy) when we shouldn’t rely on the judgment of a particular person. There are several situations where this might be the case, but most cases of the fallacy fall into one of two groups:
1. Appeal to irrelevant authority. This is the one that most logic texts focus on. The key point here is that someone’s being an authority in one area does not make her an authority in other unrelated areas. Einstein is legitimately cited as an authority on theoretical physics, but it would be a fallacy to insist that we should accept him as an authority about the political solutions to international conflicts. We have no reason to think that his knowledge or judgement in this area is any better than anyone else we might be inclined to appeal to.
One sees this fallacy at work in denialist lists of people who reject “Darwinism” or who question the dangers of climate change. These lists are often populated with people who have some sort of credentials, but not the sort that are relevant to the question at hand. When a dentist disagrees with someone who has a PhD in biology about evolution, it would be wrong to trust the dentist. It’s reasonable to trust an engineer about how sturdy a bridge is, but not about whether global temperatures are likely to rise.
Let’s make the form of this fallacy explicit: “A is an authority on topic X, therefore you should trust her pronouncements on topic Y.” This is obviously illegitimate unless you somehow establish that expertise in X translates to expertise in Y.
2. Ignoring countervailing authority. This one is not discussed as frequently, but it is also important. We should not be swayed by someone’s authority if we’re trying to decide between her position and an opposing position of equal authority.
It’s reasonable to trust Einstein’s judgment about physics in general, but not if we’re trying to decide between his view of quantum mechanics and Bohr’s position. Bohr is also an authority on the matter, so when there is a conflict, we can’t suppose that Einstein’s authority settles the matter.
The version of the fallacy is particularly bad if we’re trying to argue against someone who is herself an authority on the matter. If we’re talking to a theoretical physicist who is telling us that Einstein’s theory of general relativity is flawed, we cannot legitimately appeal to Einstein’s authority to rebut her. Why not? Because we have no reason to think that Einstein’s understanding of the situation is superior to hers.
Appeals to authority are legitimate when we’re trying to convince someone (including ourselves) who lacks either the information or intellectual skill required to grasp the argument for a conclusion. But when we can and do turn to evaluations of the information and inferences, then authority becomes irrelevant.
This second point also highlights the importance of recognizing scientific consensus. When an expert is reporting the position that the great majority of researchers in a field accept, we have very good reason for accepting that conclusion. If, however, there is significant ongoing debate on a topic, then there is no clear legitimate authority we can appeal to.
More on this at this earlier post.