Sometimes I make an argument from analogy, and I deeply regret it. I say, “X is Y for the same reason A is B,” and commenters counter, “But X and A are different!” and I say, “I never said they were the same!” And so it goes back and forth, and into the weeds.
Arguments by analogy are terrible. They never convince anyone who wasn’t convinced to begin with. Never use them. Or so I say. But before I know it I’m using analogies again, because they’re just so darn effective for making a point.
But maybe I’m still right? Perhaps analogies really don’t convince people who aren’t already convinced, it’s just that I have an audience who is already convinced. Come on, readers! Think for yourselves!
I’d like to share my thought process about arguments from analogy, and the best way to do this is to discuss a specific case study with all its messy details. So I came up with a novel analogy for a subject that most readers are familiar with: the tone argument.
Anger, civility, and car horns
Most of you are familiar with the “tone argument” even if you don’t know it by that name. Basically, it goes, “Activists sometimes have the right ideas, but do they have to be so angry? If they were a bit more civil, maybe we could get things done around here.” The counterargument is that activists are angry because they need to be.
The tone argument has been the subject of endless debates in both atheist and social justice movements. However, the tone argument has gone out of fashion in recent years (blame Trump, who provided an obvious justification for anger). That gives us enough distance to make it a good subject for a case study.
So let’s talk about car horns. Most people will agree with me, that the sound of cars beeping at each other is irritating as hell. Why can’t they get over their road rage, and just shut up? In fact, why do car horns even exist? Couldn’t we just ban them? But no, you can’t ban car horns, and in fact regulations (in the US at least) require all cars to have them. And the reason is obvious: car horns prevent accidents. Our collective irritation at car horns is less important than preventing injury and death.
Angry activism is kind of like car horns. Yes we know it’s irritating, but they’re needed to attract attention towards change that helps people. Maybe some instances of angry activism aren’t really helping, just as some car beeps are solely expressions of road rage. But anger is still a really important avenue of expression.
What good is the analogy?
Let’s begin by introducing a bit of terminology. The target of the analogy, is the thing we’re trying to argue about (i.e. angry activism). The source of the analogy is the new topic we’re introducing in order to compare to the target (i.e. car horns). Generally speaking, in order for an analogy to be any good, the source needs to be easier to understand or less controversial than the target.
My second rule, for an analogy to be any good, is that it needs to be more effective than a direct argument.
Every argument by analogy has a corresponding direct argument. For example, instead of analogizing angry activism to car horns, and then explaining why car horns are necessary, I could just write out the argument for car horns, and substitute all the keywords so that it’s an argument about angry activism from the beginning. Let’s try it out.
Most people will agree with me, that angry activists are irritating as hell. Why can’t they get over themselves, and just shut up? In fact, why do we even accept anger as a valid form of activism? Couldn’t we just ignore them? But no, we can’t ignore angry activism. And the reason is obvious: anger gets things done. Our collective irritation at angry activists is less important than changing society for the better.
To my eyes, something is lost in the word substitution after all. That’s an indication that the analogy is producing value. When it comes to car horns, we have some background information, assumptions, and intuition that make the conclusion feel natural. But our intuition about car horns is hard to explain or justify, and thus it’s hard to make an effective argument just by writing it out the argument for car horns and then substituting keywords. Generally speaking, analogies work best when we have some intuition about the source domain that is difficult to justify explicitly.
One domain where arguments by analogy are especially effective, is moral reasoning. Most of our moral beliefs are based on intuition, and most people have difficulty explaining the underlying justifications. Or maybe you can explain the justifications, but your explanation is completely different from mine. Arguments by analogy allow us to transfer conclusions from one domain to another while bypassing our disagreements over the underlying reasons.
But here’s another way of looking at it: the analogy is tricking us! The analogy encourages us to import assumptions from the source to the target, without justifying any of these assumption in the new domain, without stating what the assumptions are, or even admitting that they exist! What do you think?
Analogy failure modes
There are several common ways that arguments by analogy can fail. The first way, is if the arguer and listener disagree on the source of the analogy. A quick internet search reveals that some people believe car horns do not improve road safety, and should be banned. For such people, the analogy falls flat on its face. You might even find people who disagree on car horns, but agree on angry activism, putting them in an awkward position. This is why the source of the analogy must be uncontroversial.
A second failure mode is if the conclusion in the source domain does not obviously correspond to the desired conclusion in the target domain. For example, consider the following argument:
What is our conclusion about car horns? We agreed that car horns should be legal, but at no point did we conclude that car horns are good. The argument practically concedes that car horns are bad in the majority of cases they are used. By analogy, we can agree that it should be legal for activists to express anger (something nobody is arguing with), but we do not agree that angry activism is good. Perhaps by discouraging angry activism as much as possible–but without going so far as to make it illegal–we ensure that it will only be used in actual emergencies.
What’s going on here? In our original analogy, we implied equivalence between “car horns should be legal” and “angry activism should be a socially acceptable form of activism”. But someone else might decide that “car horns should be legal” is actually more equivalent to “angry activism should be legal”. And maybe they’re right! When you invent an analogy, you need to pay special attention to the source conclusion, ensuring that it actually corresponds to the target conclusion. Otherwise you get people agreeing with the analogy, but saying that it the analogy supports whatever it is they believe instead.
A third failure mode is if the assumptions in the source domain do not carry over to the target domain. For example, there’s an assumption that car horns effectively draw our attention to emergency situations. But is it true that angry activism effectively draws our attention to important social issues? This might be precisely the issue in contention! The analogy to car horns does nothing to demonstrate the effectiveness of angry activism; instead, it assumes agreement on that point.
How (not) to argue about analogies
Let’s talk about failure modes that aren’t real failure modes. They’re just bullshit things people bring up to dispute arguments by analogy.
I mentioned such an argument at the beginning of the essay. “But the source and target are different!” Yes, and that is what we call an analogy. Think about what you’re trying to say. What, precisely, is different about the source and target? And why does this difference matter?
People are often good at pointing out differences, but terrible at explaining why the differences matter. For example:
You know how car horns are different from angry activists? Car accidents are easily quantifiable, whereas societal prejudice is not.
But so what? Do car accidents actually need to be quantifiable for the argument to work? Personally, I’ve never read any research quantifying car accidents, but I still feel persuaded that car horns are likely necessary.
Another thing that people complain about in analogies, is when something really evil is compared with something only mildly evil. In the case of car horns vs angry activism, maybe this could go either way–car accidents and racism are both pretty evil. But suppose we were comparing sexual violence and tea (you can look that one up, it’s a popular analogy). Some people might object that sexual violence is much worse than anything you can do with tea. But remember: pointing out a difference between the source and target isn’t enough, you have to explain how the difference is relevant.
An argument by analogy is when you import knowledge about one domain (the source) to the argument at hand (the target). Analogies work best when we agree on the source, but our agreement is based on some sort of intuition that is difficult to explain directly. Analogies have many pitfalls, but some objections are just bullshit, like when people point out differences between the source and target but neglect to explain how the differences matter.
Arguments by analogy are tricky to use both correctly and convincingly. But if I may return to my first point, analogies are effective at convincing people who are already convinced to begin with. Which is to say, analogies may be at their best when not serving the function of an argument at all. Analogies work best as rhetorical hooks, or lenses to explore a topic.