16

Earlier today, I asked a programming question on a forum. I phrased the question as "What is the best way to do x?" Someone responded with something to the effect of, "the best way is usually with a tool that is widely used and accepted by established programmers for solving this sort of problem," and recommended a tool. I agreed that this is usually the case and took his advice without a second thought but it made me think about other situations where it's common to take for granted that things that are widely accepted by experts are of high quality, in turn leading to a self-fulfilling prophecy as more and more people accept it on account of its acceptance.

On the one hand, it seems somewhat similar to the bandwagon argument, which is of course fallacious. On the other hand, we're talking about a consensus among experts (for the sake of argument, let's assume they're experts) and not just ordinary people who made a decision to use one thing over another. Does an argument based on this alone have merit or is it totally fallacious? Does it even matter that it's the consensus of a group of experts and not the informed opinion of a single expert (which would simply be an appeal to authority)?

10
  • 23
    No, it is not. This is exactly the case when an argument from authority is informally valid, especially when it is taken as strong evidence rather than absolutely conclusive. Such arguments are called defeasible rather than fallacious. That means that their conclusions can be overridden, but only if one has really good reasons for doing so (e.g. are themselves experts and/or have extra information unavailable to the experts).
    – Conifold
    Sep 26 at 5:08
  • 2
    @Conifold: IMHO, it's the expectation that a community of experts must be infallible that is the fallacious reasoning here. See my answer. Sep 26 at 6:06
  • 2
    There are risks of en.wikipedia.org/wiki/Groupthink cognitive bias in the form of en.wikipedia.org/wiki/Bandwagon_effect and other aspects of mass psychology.
    – CriglCragl
    Sep 26 at 8:36
  • 3
    You ask for "the best" way to do X. But there are many criteria to take into account, and any solution will be a compromise between these criteria. One important criterion in programming is ease of understanding and maintainability. If everyone is used to one way of doing things, and you do it the same way, then your code will be easier to understand. In some situations, we have to pull obscure algorithms from research papers; but in most situations, we prefer to stick to well-known functions from standard libraries.
    – Stef
    Sep 27 at 13:07
  • 1
    You seem to be assuming that because general users take the advice of experts, that therefore experts may only take the advice of other experts in their decision-making. I myself have been an expert in certain narrow technical domains and I assure you that that is not how experts work. They are constantly evaluating and re-evaluating new (and sometimes old) tools, technologies and techniques in the pursuit of better quality and productivity. That is in part what makes them experts. Sep 27 at 13:43
30

Well, if I asked a community of non-experts how to perform key-hole surgery and I also asked a community of doctors, I am more likely to get a better answer from the second group. But of course, these doctors may not be surgeons as so they might plead ignorance. The point is the advice is more likely to be correct. It's a question of probability and not infallibility. It's the expectation that advice from a group of experts is infallible that is the fallacious thinking here.

6
  • 5
    +1: One of the differences between professional experts and others, is that the experts are more likely to say something like "this is not my specific area of expertise, so I am not qualified to answer this question". Sep 27 at 13:46
  • 1
    @RBarryYoung have you met people in the real world? Professional egotist are rife in all profession. You get some people who would rather die than admit they don't know something or cannot do something.
    – Neil Meyer
    Sep 27 at 18:08
  • 2
    To add to this, an internet forum doesn't even screen for experts -- they aren't Software engineering Professors, or esteemed members of the ACM. They may not even be practitioners. Even with a reputation system, a forum is 1 step away from "random guys on the internet". Sep 27 at 22:07
  • @OwenReynolds - academics and "association" members are the last people I'd want to ask about software development.
    – Davor
    Sep 28 at 14:20
  • @Davor A good Software Engineering professor is in contact with Sr. leads and ex-students in many different types of industries. But use your own: "a forum isn't made of true experts, who are _____". Sep 28 at 16:13
9

Ullah's answer gets to the heart of the question. The term fallacious implies the use of a standard of logical conclusion and in that strict sense, it is false to argue that a conclusion by experts can by a link in a syllogism.

I would add that once we are thinking about how reliable experts are, it is interesting to consider how any opinion was arrived at. For example, there is a pernicious problem in the sciences where one incorrect finding is published, and then forever after, multiple experts cite the original finding (or citations of it), implying agreement when there is really not a consensus per se. Good scientists are mindful of this issue.

I can't help but note that this question feels like the "question of our age" particularly in the US where "expert opinion" is dismissed by many on issues as wide-ranging as vaccination, medications, global warming, etc, etc.

6
  • I wonder if a doctor is not worth listening when it comes to vaccines who exactly is enough of a expert on medical matters according to antivaxxers?
    – Neil Meyer
    Sep 27 at 17:58
  • 2
    @NeilMeyer most of them, including me, just don't want to be guinea pigs. I get the flu shot every year, all my other vaxes, but the way these covid ones are being forced is wrong. They use brand new technology, developed in record time, and you can't sue J&J et. al. if you get hurt. I get that short-term consequences are rare, and I get that the FDA approved Pfizer's vax. But every commercial on TV that starts with "You may be entitled to a cash settlement" is about a drug that was FDA-approved. So they're mandating drugs with unknown longterm side effects and no chance for compensation.
    – Ryan_L
    Sep 27 at 19:16
  • 2
    @Ryan_L This is an interesting example of having to "act under uncertainty". You have to estimate the costs of the unknown long term side effects vs. the pretty well known benefits if you do get Covid. Of course whether they should be mandated is a separate question to whether you personally decide to take them. Incidentally the AstraZeneca vaccine is a more 'traditional' vaccine, not sure whether it's available in the USA.
    – tgdavies
    Sep 27 at 23:23
  • 3
    "It is hard to imagine a more stupid or more dangerous way of making decisions than by putting those decisions in the hands of people who pay no price for being wrong." -- Thomas Sowell
    – EvilSnack
    Sep 27 at 23:25
  • 3
    @EvilSnack Oh, it's not just imaginable, it's entirely proven fact. I'm speaking as someone who contracted a preventable, regularly-fatal disease (whooping cough) because my mother believed an unfounded scare story about vaccinations running at the time. My sister and I survived, but we had a 1% risk of at least one of us dying. My mother still regrets that, but regret wouldn't have changed anything if we'd been in the 1%. The most stupid, most dangerous way of making decisions is to put them in the hands of people who are unable to make rational decisions about the risks each way.
    – Graham
    Sep 28 at 15:02
7

This is less a question about logic (which the term "fallacy" would indicate) than a question about the theory of science; perhaps it would be better to ask "how reliable is an opinion of experts?"

Typically, of course, the answer will be more reliable than any single opinion because the chance that a majority of any number of experts is wrong is smaller than the chance that one of them you pick randomly is wrong. This principle is the basis of prediction markets or the more evolved Delphi method: While both go beyond simply asking a panel of experts, a panel's (as opposed an individual's) opinion nevertheless is what ultimately makes the prediction.

A panel may of course still be wrong — that is, a panel or even an entire community of experts is no guarantee for correctness, for a number of reasons. Here are a few examples:

  • For some questions, e.g. outside of nature sciences, there may not be a right or wrong at all.

    For example, there are widely diverging ideas about how to treat mental illnesses among psychologists. Even in economics there is often no consensus about very basic concepts: If you ask two experts — Mr. Shiller and Mr. Fama, both Nobel prize winners in economics — about bubbles and prize formation in markets you will receive pretty much opposite answers. Even if you ask a number of physicists what quantum theory really means you may receive very different answers.

  • The selection of experts may be biased.

    If you asked physicists in Germany in 1940 about relativity you would likely receive a majority opinion that it is obvious nonsense: Because it had been conceived by a scientist of Jewish origin it was rejected and derided.

  • You may be asking at an unfortunate point in time.

    If you asked physicists in the late 19th century whether there was much left to discover they would have shaken their heads and recommended you study chemistry: Most of what's to discover seemed to have been discovered.

  • The question appears deceptively simple and is asked in a layman's context, but would in fact require a thorough analysis: Most experts then don't even think an analysis is necessary, so most of them arrive at a wrong conclusion.

    When Marilyn vos Savant wrote in her general interest column about the Monty Hall problem she was wrongly derided by many professional mathematicians who were perfectly capable of solving it correctly, had they only spent the necessary effort.

    During the subprime mortgage boom of the early naughts experts almost unanimously thought the collaterized debt obligations were safe investments. The information that they, in fact, were a ticking bomb shell was available to everyone. But as Micheal Lewis describes so vividly, only a handful of people thought it worthwhile to actually read the extensive information that was mandatory with each paper and analyze it. It helped if you had Asperger's and actually liked to stare at a screen full of numbers.

In more abstract terms: If your question is not experimentally verifiable, you have a selection bias, you ask at the end of a reigning paradigm or your question has hidden depth you may receive arbitrary or wrong answers.

3
  • “if you ask a number of physicists what quantum theory really means you may receive very different answers” – yes, but they probably will agree that there isn't any decidably right or wrong answer to what it “really means”, and that the more important question of what it predicts for a concrete experiment can be answered unanimously. To get real disagreement today you should rather ask about something like string theory. Sep 27 at 11:25
  • In the age of social media the reliability of experts could not be more irrelevant because people trust the opinion of idiots most. It truly is a case of the idiot leading the blind.
    – Neil Meyer
    Sep 27 at 18:01
  • @NeilMeyer While that is a sentiment most of us have probably felt at some point, it was not the question -- quite to the contrary ;-). Sep 27 at 18:42
2

This is an appeal to authority.

Whether or not it's fallacious would depend on the details (and whether it's fallacious is also fairly subjective).

In a debate

I would probably say it's fallacious if it's a core part of the point you're trying to make in a debate.

If you were arguing, for example, in favour of veganism, someone might say "meat is necessary in your diet".

  • If they were to add "because experts say so", I would say this is a fallacy.
  • If, on the other hand, they say "it gives you protein, which you can't find elsewhere; you need protein because experts say so", I would say this is NOT a fallacy. (What they said may not actually be true, there are indeed other sources of protein, but that doesn't make it a logical fallacy.)

The key difference is that the veganism debate (generally) includes some basics about food groups and nutrients, but it doesn't include advanced biology or the medical field. Someone engaging in such a debate would be expected to know that humans need protein, but not what effect protein deprivation has on the human body.

Although if you know the basics of biology, you may be able to say something like "a lack of protein increases the risk of disease, reduces muscle mass, causes swelling and slows healing". Then they might ask how you know this happens, and then you're back at needing to appeal to authority. Or you may mention something about protein being broken down into amino acids in the body, but they may ask how you know this happens and you're back at the same problem.

Or perhaps you manage to make your way through the whole biological argument. Now they might say "a lack of protein might kill a human or make them sick, but why is this a problem". Now you get into the philosophical domain of why human life or suffering is significant to us and why we should care about this and we're getting even further away from the topic of veganism. At this point (if not long before) one may just say "that's just assumed to be a problem" if you can't explain that or to avoid going down that tangent.

I imagine if someone decides to question deep enough in a debate, the other person will often need to either appeal to authority or just state something as an assumption. This is partially because a lot or most of what we know comes down to one of those two things and partially because the debate will just take too long otherwise. This doesn't happen too often in debates though (in my experience), because most people are working from roughly the same set of core assumptions about the world and existence. As such, these assumptions aren't questioned and don't need to be stated.

Outside of a debate

If you just say that we should trust experts on some issue in casual conversation or informal discussion, I'd say that isn't generally fallacious.

In casual conversation you generally don't expect the people involved to be experts on the topics they're discussing, thus you wouldn't expect them to be able to fully justify any point of view based purely on their own knowledge of the subject.

It is perfectly reasonable, expected and almost universal (in some respects, at least) to trust an expert when you're not an expert yourself. This is basically how society manages to function. If everyone had to learn enough about engineering to build their own car and construction to build their own house and electronics to build their own computer and medicine to treat their own ailments, we'd still be stuck in the stone age (this might be a bit hyperbolic, but it also likely isn't too far off from where we'd actually be).

Even experts trust the experts who came before them (at least until they have reason to believe or suspect otherwise). If this weren't true, they'd need to learn, verify and/or redo all the work done by countless people over centuries just to get back to where we are in the present day. This alone would take multiple lifetimes of work. And then they haven't actually contributed anything new or gotten around to actually applying that knowledge in practice yet.

1

Argument from Authority is generally fallacious if used on its own. But reliance on authorities can sometimes be justified in conjunction with other evidence, which aims to demonstrate that there is more objective evidence for the conclusion than against it, that the experts have considered both sides, that the experts are competent, free, and motivated to recognise bad evidence and reject it, and that there are no factors at work preventing experts who disagree with the consensus from being heard and considered.

You deduce the existence of sufficient subject-relevant evidence by arguing that the experts will have checked for it and would behave differently if it didn't exist.

So for your example of a software tool, you would want to know whether the programmers you're talking about have checked out a range of tools, are competent and motivated to evaluate them fairly, are not constrained by commercial or contractual interests, differential availability or awareness, advertising, educational background, etc. and whether individuals producing better tools will be recognised and adopted.

So there are some software products that are widely used simply because they are widely available. UNIX was offered for free to universities, when other operating systems cost money. So everyone learned UNIX at college. Microsoft Windows and Explorer were distributed at no extra cost on PCs, so many people never bothered to download or buy alternatives. People in big companies often use Microsoft Excel for doing bulk statistical data analysis, because the companies often did a deal to install Microsoft Office on every machine, so it's already there, available, and familiar to the users. If several of the biggest software developers adopt a particular suite, programmers hoping to work there, or who previously worked there, are all familiar with it. Or software may be kept long after something better has been created because of the investment in all the legacy code built up reliant on the old tool, that would have to be binned and rebuilt at great expense.

Just because they're widely used doesn't mean they're the best for the job. Programmers under pressure tend to start with the languages, technologies, and applications they're familiar with, and only explore other options if the difficulties of using their default option are insurmountable. Not always, but often enough.

It is also often not clear whether the criteria for 'best' are widely agreed upon. Programmers very often argue about what language or operating system is better. And the continued existence of multiple solutions suggests it is not clear cut. Is Python better than C++? There is still lots of C++ code being developed, so clearly not everyone has decided that Python is superior and moved everything across. But if C++ was clearly better, why has Python appeared and taken off? Does this behaviour support the belief that programmers choose what tools to use based on objective and universally applicable criteria of quality/correctness?

Experts may be expert on one subject but not on a closely related subject. Experts may be out of date, or have expertise specific only to a particular locale. Experts may be paid to hold an opinion, or be so invested in one side of a controversy as the foundation of their career, reputation, and livelihood that they are not objective about it any more. And established experts in a field may have the power and influence to exclude people who disagree with them from being considered experts. An expert may have relied on Argument from Authority themselves and have learnt everything they know from uncritically accepting the pronouncements of other experts (like their university lecturers), same as you. Nobody has the time to examine the evidence for absolutely everything they believe and rely on. Experts are fallible, corruptible, sometimes lazy. Fundamentally, experts are only human.

Richard Feynman said "Science is the belief in the ignorance of experts." Assuming ignorance is the starting default position. Expert opinion can provide indirect evidence of correctness/quality, but you first need objective evidence that the experts are worth listening to! You can't just take it for granted because they're called 'experts' in their job title. But if there is evidence that the experts are diligently applying the scientific method - constantly challenging consensus and systematically seeking out other points of view - then expert opinion can provide an indirect source of evidence for the existence of justifying evidence on subjects too technical for the layman to examine the specific evidence directly.

2
  • 2
    "There is still lots of C++ code being developed, so clearly not everyone has decided that Python is superior and moved everything across. But if C++ was clearly better, why has Python appeared and taken off?" Granted, this particular case is much more due to one being much better suited to some types of tasks and the other being much better suited to other types of tasks than to lack of consensus.
    – reirab
    Sep 27 at 12:33
  • 1
    @reirab: This particular case is based on misapplication of the Law of the Excluded Middle (or alternatively, a bad assumption that there is a strict ordering). The evidence presented against the comparison in both directions is correct, and the correct conclusion is that the ordering is not strict.
    – Ben Voigt
    Sep 27 at 15:40
1

The superlative 'best' is an extension of the term 'better', which compares values on some measurable dimension or dimensions. A community of experts has access to acumen, skills, tools, and methods for making such comparisons analytically; as such, their assessment of 'best' within their particular domain will be far more useful and accurate than the assessments of non-experts. In specific, expert programmers will have criteria for code — efficiency, simplicity, aesthetic qualities, etc — that only they can properly assess, so only they can make proper comparative evaluations.

The operative phrase, though, is 'within their particular domain'. Some people want to evaluate on dimensions different from the comparative dimensions experts use, and in such cases expert opinions are no better and no worse than anyone else's. For instance, if the primary concern of an end-user is ease-of-use or a pleasing, beautiful interface, expert programmers have no particular skills or acumen for assessing these dimensions, and so their idea of what is 'best' in these cases may or may not be reliable. We need to keep in mind what experts are (and are not) expert at before relying on their opinions.

0

If you assume perfect determinism (and having the necessary knowledge to confirm the question at hand), then your claim say is correct. Opinion does not matter.
The simplest example of such a case is mathematics. No matter how many experts agree that 1 + 1 = 3, it doesn't matter until there is irrefutable proof.

However, real life is usually not as cut-and-dry as we would like. Therefore, we have to give way to imperfections. These can take several forms, for different reasons:

  • Because we don't know conclusively - In absence of a conclusive answer, a reasonable quorum among leading experts is statistically the most likely answer. If a pro tempore answer is needed before a conclusive answer has been found, the quorum tends to yield the best available answer.
  • Because of local context - The given answer might not be universally correct, but for the given question in its given context, it suffices. For example, we calculate the circumference of a circle using pi, even though we don't yet know it's exact value. We use an approximation, one that is proportionately accurate based on what we need right now (= the context).
  • Because it is not universally deterministic - Your software engineering example applies here. "Best" is subject to implicit considerations such as time, effort, complexity, maintainability, volatility, personal experience of the speaker, ... but these quantifiers are omitted in favor of giving what is genuinely assumed to be a commonly agreed upon balance of the omitted considerations. The true "crime" here is doing away with these pedantic quantifiers in favor of keeping the conversation lighter and easier to grasp. Could these considerations have been expressed? Sure. Would the conversation's participants appreciate it, or have need of it? No. This is a variation on the second bullet point, where the conversation's participants mutually agree on a local context which precludes the need for explicitly mentioning strict modifiers.

I don't agree that these cases should be considered fallacious, as long as they don't outright and explicitly claim to provide the final and complete, yet unproven or incorrect answer.

For example, if a teacher were to state:

"Pi is equal to 3.14"

If this is said during a beginner's class on geometry, I wouldn't call that a fallacious statement.This falls under the second bullet point: it suffices for the current context.

If that teacher were to state:

"Pi is [all currently known digits of pi]"

That wouldn't be fallacious either, as the teacher would adhere to the first bullet point: giving the best available information.

However, were that teacher to say:

"Pi is exactly equal to 3.14"

I would call that fallacious, as it makes an outright and explicit claim that is unfounded (at best) or a conscious lie (at worst).


To summarize
Your claim holds true in a setting of formal determinism, but casual human speech does not meet such a standard of formality. People omit quantifiers when they are contextually unnecessary, leading to formal imprecisions but as formality is not the topic at hand, none of the conversation's participants suffer from the imprecision.

I could (tongue in cheek) turn this on you and claim that the true fallacy here is your expectation of a casual conversation conforming to a formal standard of universal determinism, which is hardly if ever the case.

0

No, it usually wouldn't be fallacious.

A fallacy is a mistake in reasoning. It occurs when the rules used to form a conclusion from a set of premises don't logically necessitate a true conclusion.

If I say, "experts say X is true, therefore X is true," then that is a mistake in reasoning because the conclusion doesn't necessarily follow from the premise. Experts sometimes say things that are false, so the conclusion (whether true or false in any individual instance) doesn't follow.

On the other hand, if someone says, "experts say X is true, therefore X is probably true," then no such error has been committed, provided that "experts usually say true things" is taken as an implied premise.

Note that a fallacy isn't committed if the premises are false, because a fallacy is only a mistake in reasoning. So in this case, even if it's not true that experts usually say true things, the reasoning from that premise is good, so no fallacy has been committed.

Likewise, if we added "experts only say true things" to the premise of the first example, no fallacy would have been committed there either, the argument would just be wrong.

Not the answer you're looking for? Browse other questions tagged or ask your own question.