Monday, January 02, 2017

How to use an "argument from authority"

I spent the holidays playing with the video animation software. As a side-effect, I produced this little video.

If you'd rather read than listen, here's the complete voiceover:

It has become a popular defense of science deniers to yell “argument from authority” when someone quotes an experts’ opinion. Unfortunately, the argument from authority is often used incorrectly.

What is an “argument from authority”?

An “argument from authority” is a conclusion drawn not by evaluating the evidence itself, but by evaluating an opinion about that evidence. It is also sometimes called an “appeal to authority”.

Consider Bob. Bob wants to know what follows from A. To find out, he has a bag full of knowledge. The perfect argument would be if Bob starts with A and then uses his knowledge to get to B to C to D and so on until he arrives at Z. But reality is never perfect.

Let’s say Bob wants to know what’s the logarithm of 350,000. In reality he can’t find anything useful in his bag of knowledge to answer that question. So instead he calls his friend, the Pope. The Pope says “The log is 4.8.” So, Bob concludes, the log of 350,000 is 4.8 because the Pope said so.

That’s an argument from authority – and you have good reasons to question its validity.

But unlike other logical fallacies, an argument from authority isn’t necessarily wrong. It’s just that, without further information about the authority that has been consulted, you don’t know how good the argument it is.

Suppose Bob hadn’t asked the Pope what’s the log of 350,000 but instead he’d have asked his calculator. The calculator says it’s approximately 5.544.

We don’t usually call this an argument from authority. But in terms of knowledge evaluation it’s the same logical structure as exporting an opinion to a trusted friend. It’s just that in this case the authority is your calculator and it’s widely known to be an expert in calculation. Indeed, it’s known to be pretty much infallible.

You believe that your friend the calculator is correct not because you’ve tried to verify every result it comes up with. You believe it’s correct because you trust all the engineers and scientists who have produced it and who also use calculators themselves.

Indeed, most of us would probably trust a calculator more than our own calculations, or that of the Pope. And there is a good reason for that – we have a lot of prior knowledge about whose opinion on this matter is reliable. And that is also relevant knowledge.

Therefore, an argument from authority can be better than an argument lacking authority if you take into account evidence for the authority’s expertise in the subject area.

Logical fallacies were widely used by the Greeks in their philosophical discourse. They were discussing problems like “Can a circle be squared?” But many of today’s problems are of an entirely different kind, and the Greek rules aren’t always helpful.

The problems we face today can be extremely complex, like the question “What’s the origin of climate change?” “Is it a good idea to kill off mosquitoes to eradicate malaria?” or “Is dark matter made of particles?” Most of us simply don’t have all the necessary evidence and knowledge to arrive at a conclusion. We also often don’t have the time to collect the necessary evidence and knowledge.

And when a primary evaluation isn’t possible, the smart thing to do is a secondary evaluation. For this, you don’t try to answer the question itself, but you try to answer the question “Where do I best get an answer to this question?” That is, you ask an authority.

We do this all the time: You see a doctor to have him check out that strange rush. You ask your mother how to stuff the turkey. And when the repair man says your car needs a new crankshaft sensor, you don’t yell “argument from authority.” And you shouldn’t, because you’ve smartly exported your primary evaluation of evidence to a secondary system that, you are quite confident, will actually evaluate the evidence *better* than you yourself could do.

But… the secondary evidence you need is how knowledgeable the authority is on the topic of question. The more trustworthy the authority, the more reliable the information.

This also means that if you reject an argument from authority you claim that the authority isn’t trustworthy. You can do that. But it’s here’s where things most often go wrong.

The person who doesn’t want to accept the opinion of scientific experts implicitly claims that their own knowledge is more trustworthy. Without explicitly saying so, they claim that science doesn’t work, or that certain experts cannot be trusted – and that they themselves can do better. That is a claim which can be made. But science has an extremely good track record in producing correct conclusions. Questioning that it’s faulty therefore carries a heavy burden of proof.

So. To use an argument from authority correctly, you have to explain why the authority’s knowledge is not trustworthy on the question under consideration.

But what should you do if someone dismisses scientific findings by claiming an argument from authority?

I think we should have a name for such a mistaken use of the term argument from authority. We could call it the fallacy of the “omitted knowledge prior.” This means it’s a mistake to not take into account evidence for the reliability of knowledge, including one’s own knowledge. You, your calculator, and the pope aren’t equally reliable when it comes to evaluating logarithms. And that counts for something.


akidbelle said...

Hi Sabine,

without any authority in that matter I wish a happy new year to you, your family and friends.


and I also hope to read Uncle Al's... maybe I can even decode his words - this time;)

Maro said...

In the video you say "an argument from authority is necessarily wrong", whereas the text says "an argument from authority is NOT necessarily wrong". I think you misspoke and left out the "NO". It's quite confusing :)

Sabine Hossenfelder said...


What I say is "Isn't necessarily wrong". I'm very sorry for the unclear pronunciation, I noticed this too late.

TheBigHenry said...


“Can a circle can be squared?” should read "“Can a circle be squared?”

Sabine Hossenfelder said...

Thanks, I fixed that!

TheBigHenry said...

"You, your calculator, and the pope and your calculator ..."

Are you implying that the Pope is more skillful in using my calculator than I am?

Unknown said...

Regarding examples of the mechanic claiming you need a new crankshaft, or a doctor saying you need a heart surgery, it would be naive & foolish to not get an independent 2nd or even a 3rd "opinion".

The more expensive the task -- not just in upfront $, but also in difficulty or time, or the consequences of a wrong opinion -- the more essential to get additional independent "opinions".

-- TomH

Matthew Rapaport said...

You are waxing philosophical Dr. H! Good article. May I share your Google+ post with my "armchair philosophy" forum?

Haelfix said...

During the inquisition and for instance during the trials of Galileo, there was a lot of logical arguments by the clergy along those lines. Galileos arguments weren't always perfect and sometimes possessed logical problems, the evidence for what he was claiming was also frequently imperfect (and the clergy and their host of scholars often had reasonable points disputing Galileo's claims).

This then allowed the clergy to state that b/c there was some fog clouding the whole affair, that Galileo should trust the weight of communal knowledge and the evidence based upon thousands of years of philosophy/religious debate.

You see the problem of course... Galileo shouldn't have had to entertain any thoughts about priors and how much he trusted their expertise or not. Galileo simply knew the right answer b/c he understood the correct physics.

It's like the first time in highschool or wherever where you see Newtons laws, and realize that of course it couldn't be any other way. The rest is blah blah blah.

TheBigHenry said...


Galileo's arguments

Newton's laws

The possessive requires an apostrophe.

Sabine Hossenfelder said...


Sorry I had missed your comment. I've removed the 2nd calculator, thanks for pointing out. I had printed the transcript and revised it by hand. Evidently, I forgot to make the changes also in the text-file.

Maurice said...

So what's your punchline: should Bob "believe" in the answer of his calculator?

Sabine Hossenfelder said...

If I told you, should you "believe" me? ;)

M_Malenfant said...

A very gould article, I appreciated it.
But I see the problem, that if the basis of knowledge is too small, judging authorities doesn't work very well. Without some idea of what science means the publicly communicated results obviously are indistinguishable from politics and all the volatile everyday reasonings.
Add in some half-digested understanding of theories and scientific proof, highly sponsored campaigns for 'competing' or simply interest-driven 'science' and the strange trust in rather arbitrary claims found in the internet. With this background the selection of authorities for many scientific questions doesn't lead all to scientists.
If you have some solid scientific education in any field you are likely to get an idea what it means to investigate a topic in depth and to whom to look for reliable answers - and to judge how reliable the answers are, even if you have some own ideas of specific details.
I'm afraid part of the society is loosing the critical level of pre-understanding or trust in science as such. I can only hope this trend will change soon.

Alex said...

"But unlike other logical fallacies, an argument from authority isn’t necessarily wrong."

Isn't this true of all logical fallacies?

I agree that in everyday life we have limited time to check the validity of certain arguments, but I'm suspicious of the attempt to water down this logical fallacy. There already are too many institutions that argue from authority when in my opinion they shouldn't be. I've seen too many physics papers rejected on the basis that they disagree with some other *recently* published paper without the referees actually pinpointing where the argument is most likely incorrect. Referees should have ample time to evaluate journal submissions so I think it would even be to the benefit of the referee to actually figure out where the argument is wrong.

Following this line of thought, couldn't Bob just learn multiplication and raise 10 to the power of 5.544 to check the answer?

Unknown said...

The eastern end of the Panama canal is on the Pacific Ocean, and Reno Nevada is west of Los Angeles, California. Do you believe an Anonymous source?

TheBigHenry said...


Those assertions are easily confirmed by inspection. Belief doesn't enter into it.

Chris Mannering said...

Appeal to Authority has NOTHING to do with questions of experts and quotation of expert opinion, and whether or when it is legitimate to substitute an expert positioning, for a bespoke argument, in the course of debating.

Likewise, Arguing authoratively using fully referenced expert opinion in quotations by it's very definition is NEVER an 'Argument from authority'.

.....because it isn't just assertion, it's a referenced expert opinion. Which is obviously both legitimate and heavily encouraged in discourse as an exemplifification of scholarly de'rigour in civilized discourse WOT WOT

APEppink said...

Give the Pope a break. He didn't miss it by much.

Nobody said...

Hi everybody!
Whoever is an authority or not, when his/her argument fails the test of logical consistency on a fundamental level, it simply proves its ignorance (better I would say ignorance is replaced with abstract maths that do not follow reality). I would not agree with the "Greek rules do not always work". Why? Because the arguments used by the Greeks were referred to ultimate fundamental engaged information, something that Western culture have totally misunderstood. Re-Embracing logic in quantum physics will lead us where we should be (today is an illusion).

Sabine Hossenfelder said...


You can use logic to deduce consequences only if you are dealing with mathematically clearly defined structures. In reality this is never the case. Mathematical relations are the only cases for which we can say "that's right" and "that's wrong" with 100% certainty. What the Greeks didn't take into account is that when referring to the real world all we can speak of are probabilities of something being correct.

I have no clue what you think this has to do with quantum physics.

Plato Hagel said...

An argument from authority? What should it look like?

Well would there be deductive/inductive/abductive reasoning to suggest "a first principle," and then we recognize "the idea" of the proposal from such an authority as self evident? We would move on from there, and do our work?

Ambi Valent said...

When the scientists working on quantifying the influence of greenhouse gases on global warming, they were faced with the counterargument that global warming would be caused by solar variability. They did not simply dismiss that argument, but quantifiwd both greenhouse gas signal and solar variability signal, and found that the greenhouse gas signal was the dominating cause of global warming.

Regarding General Relativity, my concerns with the classical model of black holes (still found in magazines and even with many science popularisers) were taken into account in the work of Liu and Zhang (2009). I recognize where I was wrong. Sorry for making such a mess here.

(BTW, as a GR expert, what do you think about the work of Liu and Zhang?)

Nobody said...

Hi Sabine!
I sent another post about a day or so ago but did not appear on this discussion. What is going on?

Sabine Hossenfelder said...


I don't know. I only got one comment from you, four days ago, and it appears above. There's no other comment from you in the queue.