> People die in huge numbers in very violent automobile accidents.

That's offset by the illusion that people believe they're in a very high degree of control over the situation. It enables the brain to create an illusion and say to itself: I won't make that mistake, my actions dictate the outcome, I'm a good driver (as the comical studies always demonstrate, people tend to dramatically over-estimate their driving skill).

> Most physical violence ... is between people who know each-other

You don't actually think that's going to happen, that's the issue. If you told someone their pal is about to rape or murder them, they would not believe it, they'd be far more afraid of the menacing looking stranger. That's where ignorance of risk - as with sharks - comes into action. I would guarantee a low percentage would believe they're more likely to be murdered by someone they're close to than a stranger. If a person doesn't have a meaningful comprehension of the risk (as with nuclear), the risk context inherently can't be judged properly. I suspect there's a big illusion of control that is involved between two known relations as well, you think you have some influence over whether a friend will murder you (surely they like you, surely you could talk them down from a violent confrontation, etc.); versus that almost entirely missing from the menacing stranger scenario.

>Most physical violence (assault, murder, rape, ..) is between people who know each-other, but typically people are more afraid of different-looking strangers.

That (the belief that this is an error) is itself a statistical fallacy, one that's isomorphic to "most traffic accidents happens near your home" (most of your driving is near your home). But the relevant danger metric is "risk per mile driven". That's lower near your home. You are right to use less caution there.

Likewise: most of your conflict is with people you know, but most of your interaction is likewise with acquaintances! The relevant danger metric is "crimes committed against you per unit of interaction time". On that measure, strangers are in fact more dangerous.

Raising your guard around acquaintances will lead to costly false negatives. Raising your guard around strangers will lead to a small number of false negatives and save you from costly false positives. Do you deliberately go by yourself down dark alleys filled with strangers, since "if I'm going to be a crime victim at all, it'll be from someone I know"? Then you understand the basic fallacy there.

This logic doesn't make sense to me. If walking down a dark alley filled with strangers is "interacting with strangers" then so is walking anywhere - the mall, the bus, the beach, every restaurant, etc. Our lives are full of strangers, the vast majority of which never even notice us.

Yes, they are all interacting with strangers! To make the metric more precise, you’d want to weight by something like the closeness of the interaction. But yes, similarly, you have a heightened guard around these people, for a good reason. How many of those strangers do you enter a room alone with or lend your things to?

(Before you make the point about the coat check, ask how much prefiltering you do before you give someone your coat.)

Your model, by contrast, doesn’t seem to represent why that dark alley is dangerous. Because you only care about the raw crime count rather than the crime count per time interacted with. It would equally imply that the stranger in the alley is no risk to you, because “if a crime happens at all, it will be from a family member”. That is not good statistical reasoning.

When we see that most crime comes from acquaintances, that’s after applying significantly more countermeasures to strangers and after getting into more conflicts with acquaintances after we’ve focused our interactions on them.

If strangers really are safer per unit of interaction, then start saving your seat by leaving your wallet.

That’s precisely the point. Having a fear of people who you haven’t heavily interacted with, when wandering about in public in highly trafficked spaces, is irrational, because the risks are vanishingly small. I’ve e.g. personally watched a middle aged lady cross to the other side of the street in the middle of the day to avoid walking closely past some normal looking brown-skinned high school kids wearing hoodies and baggy pants, and then cross back to the original side after they were past. On several occasions I’ve seen men switch sides with their female companions so that as they walk along the man is between the woman and some unassuming homeless person sitting on the sidewalk minding their own business. Again, on a well-trafficked street in the middle of the day. These reactions are not based on reasonable fears.

Lots of people won’t let their 10- or 12-year-old kids walk around town by themselves, because the parents are afraid they will be abducted or assaulted or something. But such tragedies are extremely rare. If anything, if they are really worried about child abuse the parents should be carefully examining their priest, sports coaches, teachers, family members, friends’ relatives, etc.

Obviously if you take the random strangers off the street and invite them to your home, have a deep personal conversation, lend them money, ask them to take care of your kids, start kissing them, or whatever your definition is for a “close interaction”, they sort of stop being strangers at that point, and you are more likely to get into some altercation.

>Obviously if you take the random strangers off the street and invite them to your home, have a deep personal conversation, lend them money, ask them to take care of your kids, start kissing them, or whatever your definition is for a “close interaction”, they sort of stop being strangers at that point, and you are more likely to get into some altercation

In other words, you agree that you shouldn't lower your defenses for a random stranger, you only disagree about where the stranger/acquaintance boundary exists.

As for the other point: do you actually know that the risk is lower conditional on you not using standard countermeasures (that probably exist because of the legit risk)? Have you stopped locking your car, and started leaving your wallet out where it's easy to get to? Walked home alone through dangerous neighborhoods? If not, why not?

Don't equate the post-countermeasure crime rate with safety.

> Your model, by contrast, doesn’t seem to represent why that dark alley is dangerous. Because you only care about the raw crime count rather than the crime count per time interacted with. It would equally imply that the stranger in the alley is no risk to you, because “if a crime happens at all, it will be from a family member

Actually that's not my model; I'm not the OP. I was just pointing out what I saw as a flaw in your presentation. Y

You keep talking about things like "unit of interaction" - which is a handy thing to talk about in the abstract, but gives the impression that you've actually quantified human interaction - a concept I am highly skeptical of.

I can see that there is a reasoning flaw in looking at crime that has happened and seeing that it mostly happens between people who know each other and concluding that the greatest risk to yourself is from people you know.

One possible explanation that has nothing to do with "interaction time" is simply the fact that you apply more countermeasures to strangers than you do to people you know. Perhaps the real cause is a matter of trust, not time.

>You keep talking about things like "unit of interaction" - which is a handy thing to talk about in the abstract, but gives the impression that you've actually quantified human interaction - a concept I am highly skeptical of.

A precise quantification is not necessary to recognize the core point, that however you define interaction, you are in fact interacting with acquaintances much more than strangers. You don't need to put numbers on eye contact or anything like that. It's enough to go by e.g. physical unbarriered proximity.

>One possible explanation that has nothing to do with "interaction time" is simply the fact that you apply more countermeasures to strangers than you do to people you know. Perhaps the real cause is a matter of trust, not time.

That's saying the same thing: people apply more countermeasures to strangers because they (correctly!) recognize them as being more dangerous in the practically-relevant sense. After that filter is applied, and after swamping your interactions with acquaintances, the end result is a relatively low proportion of crimes from stranger-in-the-bushes. That doesn't mean predators in bushes aren't dangerous!

And, once someone has become trusted, they're no longer a stranger and likewise are less dangerous because they have more to lose from hurting you. Either way, by the appropriate metric, acquaintances/friends/family are less dangerous. The greater incidence comes from being around them more and from not using the same countermeasures.

Either way, it simply doesn't follow (as the OP suggested) that the rational thing to do is to be less wary of strangers!

> People die in huge numbers in very violent automobile accidents. For young people in the US (15–35), the leading causes of death are car accidents and gunshots (homicide, suicide, accidents).

Right. But nobody is talking about reducing that number, or reducing our reliance on cars. Drivers kill over a million people every single year, but that is acceptable death.

There are people talking about it, but there’s enough profit and personal identity at stake that it’s hard to have real society-wide conversations about what we really want to value as a society.

There exist countries where automobile deaths are not considered acceptable, or where not all policy planning takes an ironclad need for automobiles as the exclusive mode of transportation as a basic premise.

I think it's very misleading to lump suicide in with forms of death that are not under one's control.

I should say that homicide and suicide by gun are causes of death #2 and #3 for that age group (the order of the two is different between 15–25 vs. 25–35 year olds). Successful suicide rate is notably higher in places chock full of guns.

Which is not even to mention the daily occurrence in the USA of young children accidentally shooting other children (typically siblings or friends) or themselves to death.

Having a gun in your house is one of the biggest possible risks you can bring to your own life and the lives of people in your family.

We're talking about forms of death over which the deceased have little control. Fear of airplanes, nuclear accidents, etc, etc.

For the purposes of comparing evaluating people's fear of things that could kill them that they don't have control over suicides (60% +/- some depending on which party's sand you bury your head in) and intentional non-criminal (defensive) gun deaths should not be counted because they are not something people do not have complete control over. Don't try to kill yourself and don't try and commit a crime against someone who's armed and you have basically nothing to worry about.

People are afraid of gun accidents in the same manner in which they are afraid of other forms accidents that they have some degree of control over (like car accidents). They take steps to reduce their likelihood and encourage (through various means) others to do the same (i.e. social disapproval of people who do unnecessarily dangerous things).

For the purposes of discussing public perception of the risks of nuclear power "gun deaths" as a lump sum is not useful and the effort to pick it apart and analyze the components is not worth the risk of derailing any useful discussion with politics.