Offense versus harm minimization


Scott Alexander

Imagine that one night, an alien prankster secretly implants electrodes into the brains of an entire country - let's say Britain. The next day, everyone in Britain discovers that pictures of salmon suddenly give them jolts of painful psychic distress. Every time they see a picture of a salmon, or they hear about someone photographing a salmon, or they even contemplate taking such a picture themselves, they get a feeling of wrongness that ruins their entire day.

I think most decent people would be willing to go to some trouble to avoid taking pictures of salmon if British people politely asked this favor of them. If someone deliberately took lots of salmon photos and waved them in the Brits' faces, I think it would be fair to say ey isn't a nice person. And if the British government banned salmon photography, and refused to allow salmon pictures into the country, well, maybe not everyone would agree but I think most people would at least be able to understand and sympathize with the reasons for such a law.

So why don't most people extend the same sympathy they would give Brits who don't like pictures of salmon, to Muslims who don't like pictures of Mohammed?


I first1 started thinking along these lines when I heard about Everybody Draw Mohammed Day, and revisited the issue recently after discovering

I have to admit, I find these funny. I want to like them. But my attempts to think of reasons why this is totally different from showing pictures of salmon to British people fail:

• You could argue Brits did not choose to have their abnormal sensitivity to salmon while Muslims might be considered to be choosing their sensitivity to Mohammed. But this requires a libertarian free will. Further, I see little difference between how a Muslim "chooses" to get upset at disrespect to Mohammed, and how a Westerner might "choose" to get upset if you called eir mother a whore. Even though the anger isn't being caused by alien technology, it doesn't feel like a "choice" and it's more than just a passing whim. And if tomorrow I tried to "choose" to become angry every time someone showed me a picture of a salmon, I couldn't do it - I could pretend to be angry, but I couldn't make myself feel genuine rage.

• Muslims' sensitivity to Mohammed is based on a falsehood; Islam is a false religion and Mohammed is too dead to care how anyone depicts him. I agree with this statement, but I don't think it licenses me to cause psychic pain to Muslims. I couldn't go around to mosques and punch Muslims in the face, shouting "Your religion is false, so you deserve it!".

• It is necessary to draw pictures of Mohammed to show Muslims that violence and terrorism are inappropriate responses. I think the logic here is that a few people drew pictures of Mohammed, some radicals sent out death threats and burned embassies, and now we need to draw more pictures of Mohammed to convince Muslims not to do this. But it sounds pretty stupid when you put it in exactly those words. Say a random Christian kicked a Muslim in the face, and a few other Muslims got really angry, blew the whole thing out of proportion, and killed him and his entire family. This would be an inappropriately strong response, and certainly you could be upset about it, but the proper response wouldn't be to go kicking random Muslims in the face. They didn't do it, and they probably don't even approve. But drawing pictures of Mohammed offends many Muslims, not just the ones who send death threats.

• The slippery slope argument: if we allow Muslims' concerns to prevent us from drawing pictures of Mohammed, sooner or later we'll have to accept every two-bit group with a ridiculous superstition and we'll never be able to get anything done. I take this more seriously than the previous three arguments, but I've previously argued that granting large established religions special rights is relatively immune to slippery-slope. And anyway, drawing pictures of Mohammed is such an unusual thing to do that we can stop doing it without giving up our right to keep doing something else that's actually useful if the situation comes up later.

None of these excuses really does it for me. So my provisional conclusion is that yes, people who draw pictures of Mohammed where Muslims can see them are bad people in the same way that people who go around showing photos of salmon to Brits are bad people.

So the big question is: why is this so controversial in the Mohammed example, when it seems so obvious in the salmon example?


I think several features of the salmon example trigger consequentialist moral reasoning, in which the goal is to figure out how to satisfy as many people's preferences as possible; several contrasting features of the Mohammed case trigger deontological moral reasoning, in which the goal is to figure out who is a good person or a bad person and to assign status and blame appropriately. These two forms of reasoning give different results in the two different cases.

The word that comes up a lot in discussions of this sort of issue is "offensive". When someone draws Mohammed, it is considered offensive to Muslims. When someone writes a story where all the sympathetic and interesting characters are male, it is considered offensive to women.

For me, the word "offensive" brings up connotations of "It was morally wrong to say this, and you are either inexcusably ignorant of this fact or deliberately malicious. You must immediately apologize, and it is up to the group you have offended to decide whether they accept your apology or whether they want to punish you in some well-deserved way."

This means that ever admitting you were offensive is a huge status hit implying you are some combination of callous, ignorant, and racist. Sometimes people may be willing to take this status hit, especially if upon reflection they believe they really were in the wrong, but since most people's actions seem reasonable to themselves they will not be willing to accept a narrative where they're the villain.

More likely, they will try to advance an alternative interpretation, in which their actions were not legitimately offensive or in which they have the "right" to take such actions. Such an interpretation may cast the offended party as a villain, trying to gain power and control by pretending to be offended, or unduly restricting the free speech of others.

The controversy over drawing Mohammed has several factors that predispose to this sort of interpretation. There is already a history of misunderstanding and some enmity between Muslims and non-Muslims. Muslims' status as a minority makes ideas of "political correctness" readily primed and available, making people likely to miss the trees for the forest. Muslims are often of a different race than Christians, so conflicts with them risk tarring a person with the deeply insulting label of "racist". And because there are reports of Muslims rioting and hurting other people because of Mohammed drawings, they are easy to villainize.

This risks embroiling everyone in an unproductive argument about whether an action was "legitimately offensive" or not, with much status riding on the result.


The British salmon example, on the other hand, was designed to avoid the idea of "offense" and trigger consequentialist notions of harm minimization2.

The example specifically refers to the displeasure that salmon cause the British as "psychic pain", priming ideas about whether it is acceptable to cause pain to another person. The British are described as politely asking us to avoid salmon photography as a favor to them, putting themselves in a low status position rather than demanding we respect their status. British are white and first world, so it's hard to think of this as a political correctness issue and wade into that particular quagmire. And because the whole salmon problem is the result of an alien prankster, there's no easily available narrative in which the British are at fault.

A consequentialist reasoner would consider how much disutility it causes not to be able to use pictures of salmon where the British might see them, then consider how much disutility it causes the British to see pictures of salmon, and if the latter outweighed the former, they'd stop with the salmon pictures. There's an argument to be made about slippery slope, but in this case the slope doesn't seem too slippery and other cases can be evaluated on their merits.

And a consequentialist British person, when considering how to convince a foreigner to stop using pictures of salmon, would try to phrase eir request in a way that minimizes the chances that the foreigner gets upset and confrontational, and maximizes the chances that they actually stop with the salmon.

If the foreigner refused to stop with the salmon pictures, the British person would try to shame and discredit the foreigner into doing so only if ey thought it would work better than any less confrontational method, and only if the chance of it successfully stopping the offending behavior was great enough that it outweighted the amount of bad feelings and confrontation it would cause.

This is a healthier and potentially more successful method of resolving offensive actions.


I post on a forum where a bunch of regulars recently denounced the culture of verbal abuse. The abusers, for their part, said that the victims were making mountains out of molehills: exaggerating some good-natured teasing in order to look holier-than-thou.

I was friends with some of victims and with some abusers; neither side were majority bad people, and it surprised me that people would view requests to stop verbal abuse as a Machiavellian ploy.

Not to say that asking for verbal abuse to stop can't be a Machiavellian ploy. In fact, as far as Machiavellian ploys go, it's a pretty good one - take something your political enemies do, pretend to be deeply offended by it, and then act upset until your enemies are forced to stop, inconveniencing them and gaining you sympathy. A conspiracy such is this is not impossible, but why is it so often the first possibility people jump to?

I think it has to do with something I heard one of the abusers say: "I would never get upset over something little like that."

I know him and he is telling the truth. When someone is verbally confrontational with him, he takes it in stride or laughs it off, because that's the kind of guy he is.

I am of Jewish background. I've had someone use an anti-Semitic slur on me exactly once. My reaction was the same mix of confusion and amusement I'd feel if someone tried a vintage Shakespearean insult. And yet I also know of Jews who have been devastated by anti-Semitic slurs, to the point where they've stopped going to school because someone in school taunted them. These people may differ from me in terms of Jewish identity, extraversion, demographics, social status, anxiety, neurogenetics, and some hard-to-define factor we might as well just call "thin skin".

The point is, if I use my own reactions to model theirs, I will fail, miserably. I will try to connect their reaction to the most plausible situation in which my mind would generate the same reaction in the same situation - in which I am not really upset but am pretending to be so for Machiavellian motives.

In the case of anti-Semitism, it's easy to see factors - like a history of suffering from past prejudice - that make other people's responses differ from mine. It's less obvious why someone else might differ in their response to being called ugly, or stupid, or just being told to fuck off - but if these differences really exist, they might explain why people just can't agree about offensive actions.

A thick-skinned person just can't model a person with thinner skin all that well. And so when the latter gets upset over some insult, the thick-skinned person calls them "unreasonable", or assumes that they're making it up in order to gain sympathy. My friends in the online forum couldn't believe anyone could really be so sensitive as to find their comments abusive, and so they ended up doing some serious mental damage.


Consequentialism suggests a specific course of action for both victims of offense and people performing potentially offensive actions. The victim should judge whether ey believes the offense causes more pain to em than it does benefit to the offender; if so, ey should nonjudgmentally request the offender stop while applying the Principle of Charity to the offender, and if ey wants the maximum chance of the offense stopping, ey should resist the urge to demand an apology or do anything else that could potentially turn it into a status game.

The offender, for eir part, should stop offending as soon as ey realizes that the amount of pain eir actions cause is greater than the amount of annoyance it would take to avoid the offending action, even if ey can't understand why it would cause any pain at all. If ey wishes, ey may choose to apologize even though no apology was demanded.

If the offender refuses, the victim should only then consider "punishment" by trying to shame the offender and make em appear low status, and only if ey thinks this has a real chance of stopping the offending behavior either in this case or in the future. Like all attempts to deliberately harm another person, this course of action requires of the victim exceptional certainty that ey is in the right.

Although people pretending to be offended for personal gain is a real problem, it is less common in reality than it is in people's imaginations. If a person appears to suffer from an action of yours which you find completely innocuous, you should consider the possibility that eir mind is different from yours before rejecting eir suffering as feigned.



1) Thanks to Kaj Sotala, Vladimir Nesov, and kovacsa-whose-LW-name-I-don't-know for originally encouraging me to turn the original essay into an LW post.

2) The deontological notion of offense doesn't really supervene on an idea of pain to other people. If two white people, talking where no black people could possibly overhear them, make a racist joke about black people, that is still "offensive", because racism is wrong no matter what. A consequentialist notion of offense could better ground such examples by theorizing that whites telling racist jokes to other whites creates a climate in which racism is considered acceptable, which eventually will end up hurting someone directly. Or it could decide not to, if it decided the link was too tenuous and hokey - but now any disagreement on the matter is honest disagreement about empirical facts and not philosophical disagreement about who's a bad person.