I did the math. SF is 2.6x more dangerous per sq mi than any other US city | Hacker News Reader

I did the math. SF is 2.6x more dangerous per sq mi than any other US city

docs.google.com

71 pointsgsibble57 comments

Per square mile is not useful. Crime rates are related to people. People commit crimes, not land. This is a made up statistic whose only useful purpose is to make San Francisco go to the top of a list it otherwise would not. It's why crimes are reported per capita

It’s the only way to make sound like it’s safer to live in the country.

OP's document lists 11 cities for comparison and says nothing about living in the country.

Everybody knows it's safer to live in the country, but that's not the argument he's making. He's saying it's also safer to live in any other city. I don't know if it's a great methodology though, going per sq mile.

Dunno, traffic accident probability are usually per mile driven, and rural people drive more to get anywhere.

Some counties have a single ambulance, and are 1-2 hours from a hospital.

Where I live, there's only volunteer fire departments. That can't be great either.

People just love hating on SF and/or cities in general, and will dig up any statistic that supports their viewpoint.

I think it can get to perception of how safe or dangerous a place is. If there are lots of crimes happening all around you as walk around a city, that can obviously seem more unsafe than if you're driving about the countryside and only rarely pass criminal activity. Even if crimes per capita are equal the perception of danger is much greater in the former than the latter.

Exactly.

> This is a made up statistic whose only useful purpose is to make San Francisco go to the top of a list it otherwise would not.

It's high the list no matter how you look at it:

> San Francisco has one of the highest crime rates in America compared to all communities of all sizes - from the smallest towns to the very largest cities. One's chance of becoming a victim of either violent or property crime here is one in 18. Within California, more than 98% of the communities have a lower crime rate than San Francisco.

https://www.neighborhoodscout.com/ca/san-francisco/crime#des...

Funny you chose that source, which does not list SF as being in the top 100 most dangerous cities in the US:

https://www.neighborhoodscout.com/blog/top100dangerous

It's about half as dangerous as city ranked #73, Farmington, NM, and over 6x less dangerous than #1 Bessemer, AL.

Low danger (low violent crime rate) does not equal low crime rate. It wouldn't surprise me if San Francisco has more tax evaders per capita. They are even probably concentrated in the financial district, which has lots of high rises, so more crime per square mile as well. I call for a crackdown.

The stat says that crime is much geographically denser on average than other US cities, even after controlling for population density.

This implies that crime levels are high throughout much of the city, or extraordinarily high in pockets.

And if you live here, you know this is true. In contrast to many other cities, where crime is often concentrated in pockets, SF has high crime almost everywhere with pockets of safety in expensive neighborhoods like Pac Heights, Russian Hill, Forrest Knolls, and the outskirts.

The stat means it’s harder to escape crime in SF, unless you’re one of the privileged few.

No, per square mile is meaningless.

Crime per square mile will be higher anywhere where people live in high rise buildings instead of suburbs.

So what? Is it worse that a robbery took place 15 floors below you rather than five blocks away? Because I don't see the difference at all.

If you're worried about personal safety, literally the only important denominator is per capita. That's the likelihood of something happening to you.

Per square mile has no relevance to personal safety. And if high population density is a factor in crime levels rising, that's already reflected in per-capita stats.

Absolutely it is worse, as the criminal has access to your building and you may meet them more easily in elevators and common areas.

No, because the crime rate is the same so it doesn't matter by definition.

But also, think about -- in the suburbs the criminal can just walk into your front yard. There's no difference between someone at the door to your apartment or the door to your house. (Heck, houses can be easier to break into because they have more doors to try out of view, windows that can be smashed, etc.)

And yes you can run into a criminal in an elevator. You can also run into them on a sidewalk, or get carjacked while stopped at a traffic light or to park.

But at the end of the day there's no difference. Crime per capita is crime per capita.

Well that likelihood is measured as crime per capita, of course.

Conditional on someone breaking in, you are much safer in an apartment building. If you live in a single family house and someone breaks in, he or she will probably rob you. If you live in a 100 unit apartment building and someone breaks in, he or she will rob either your unit, or or one of the 99 others. For it to be yours, you need to be unlucky that day, even after the break-in.

Do you think SF is safe? Do you live there? If so do you go downtown? Or to parks, or to soma or ride the bus? Have you ever walked on Polk street? Take a glance towards the tenderloin if you do. I’ve seen brazen daylight heroin injection on mission street. A dude ducked into the entrance of tech shop to smoke a little crack with people inside looking at him. I’ve seen more public shitting than I would have thought possible. I’ve personally witnessed the phenomenon or brazen shoplifting (where someone just openly loads up a bag or even rides a bike through stealing stuff) I’ve experienced the elevated property crime. A good friend was held up at gunpoint. All in the last year I was living there.

I was there in December around Union square. There were three cops on every corner due to the rash of violent robberies and smash and grabs. I spent the week constantly looking over my shoulder.

San Francisco is very unsafe. There’s a general feeling of lawlessness and that crime can be committed with impunity. I would literally feel safer in Somalia because at least there I would be armed and expected to defend myself.

> I would literally feel safer in Somalia because at least there I would be armed and expected to defend myself.

Is there a word for something even more exaggerated than hyperbole?

People's feelings are subjective. I have no reason to doubt that the poster would "feel" safer, which is why I trust data over feelings. Lots of people in the US "feel" unsafe, despite available data on crime rates.

They wouldn't feel safer in Somalia if they were actually in Somalia. In the US, US crime is real and Somali crime is abstract. In Somalia, though...

You're engaging with this as if it's good faith science. Stepping back, do you think that's a good use of your time?

Calling out someone for engaging in good faith - what a bizarre world we live in.

It's a simple probability calculation. SF has a smaller area which results in anyone having the possibility of getting targeted by dangerous people. Larger cities also have more gang violence because they are bigger and gangs try to control territory and cause a lot of crime in the process.

SF has made itself a target by thinking people don't have the capacity to do bad things on purpose... and bad people will absolutely take advantage of this (ironic?). There is more nuance here of course, but I'm trying to quickly explain why less than 10 years ago I was walking around at 2am and felt perfectly safe, but now I won't even go to SF during the day.

There is already a much better metric for calculating your probability of being a victim of a crime within a city -- it's the per-capita crime rate.

Per capita it's still number 7.

Yup. Saying that there aren't many homicides per square mile in St Louis or Chicago is equivalent to saying that Palestine OH is a safe place because the town is really sparse and so the cancer per sq. mile is acceptable.

SF's (6.35)¹ rate of homicides is below the national average (7.8)². St Louis (66.07)¹, which is listed as a city of comparison, has about 10x more homicides per 100k. This may be counter to naive expectations where one might say that proximity has a causal effect on homicides, since proximity means opportunity.

---

[1]: https://en.wikipedia.org/wiki/List_of_United_States_cities_b...

[2]: https://www.cdc.gov/nchs/fastats/homicide.htm

Holy crap, I was wondering what the heck is going on in St. Louis. I can't find any super clear authoritative answers with a quick Google search, but did find one comment suggesting it's at least partly a statistical artifact based on city borders:

> Some of it is a statistical anomaly. Most cities in this weight class annexed their counties, and thus averaged in a whole lot of white upper class good schools suburbs. If you look at st Louis city and county together, the crime stats drop to the middle of the pack with other metros this size. [1]

This has always been a problem in comparing statistics between urban areas, because the administrative boundaries are entirely arbitrary. I've long wished for some kind of algorithm that could define metropolitan areas in some kind of objective way for statistical comparison, based solely on population maps, but I'm not aware of anything that exists.

[1] https://www.reddit.com/r/StLouis/comments/9pfu4l/what_are_th...

Yeah administrative areas are very arbitrary, and based on historical factors. If you really wanted to compare, I think comparing MSA's might get you closer. But either way this is a terrible metric

Maybe one of the anti-gerrymandering projects would be useful?

For example, http://autoredistrict.org/

There's a few things about this data that indicate why this is not useful:

1. The New York City Sq Mileage number includes water.

2. Density for each of these cities varies widely by community. For example, 6% of Chicago's sq mileage is in the O'Hare community, which has a population density of 1000 residents per square mile. A better comparison would be Chicago vs. the Bay Area

You'd probably have to compare areas of similar population density to make a serious statement.

On the other hand, I'm sure it also has the highest amount of total stock grants per square mile. Which is a totally valid statistics and proves that SF is the city to live if you want to succeed as a software engineer.

This means you witness and see much more crime as a resident and why living in SF is such a different experience than any other US city.

That assumes that both your presence and the crime are equally distributed over area. If SF annexed a thousand square miles of crimeless, uninhabited wilderness, that would vastly decrease its crime per square mile, but would it affect how much crime each resident sees?

The stat indicates there’s moderately high crime evenly distributed, or very high crime throughout a significant portion of the city, or extraordinarily high crime in smaller isolated areas. It’s the second of those (and OP is right, it’s hard to avoid crime if you live here and can’t confine yourself at all times to a few safe residential areas) but none of those outcomes are good.

Sure, but some of the comparison cities (e.g. Memphis and Albuquerque) have huge areas of undeveloped land within their city limits. There's no good reason to compute a per square mile statistic given the vastly different definitions of what constitutes each city.

Don't know why you're being downvoted -- there are parts of Chicago that are 2x as dense as San Francisco, and parts that are less than 1/10th as dense. The denser parts of Chicago hold about the same number of people as San Francisco.

If you wanted to use this metric, you'd need to compare the incidents per square mile of different neighborhoods or city centers. Just doing incidents/mi^2 for a whole city is nonsense

Crime isn’t equally distributed; criminality isn’t equally distributed; sensitivity to crime isn’t equally distributed; tolerance of criminality isn’t equally distributed.

A quick informal sketch of why the “SF is hell / no it isn’t, here’s the stats, there are other places with higher crime rates / well I know SF is hell, here’s a synthetic statistic that puts SF at the top and thus reflects my experience / that synthetic statistic is bullshit / well my experience is not bullshit, so here’s another synthetic statistic” loop happens so often:

Feeling safe is important to people. People have limited mobility and spend it to move themselves into areas (cities, districts, suburbs, streets, city blocks) where they feel safe. Due to a constellation of interacting factors, highly crime-sensitive individuals are exposed to high-criminality areas at a higher rate in SF than in most other places.

Quick/dirty/concrete example of what I mean: I’ve lived in Sydney, New York, and San Francisco. All three places have high crime areas. In Sydney, I never see them. In NY, I saw them through the window of a Metro-North train. In SF, I saw them outside my apartment window.

I believe you have put into words my sense of things over the last few days, and it could probably be put bluntly as something like "crime is happening to privileged people in San Francisco and they're shocked by it".

My initial thought that sparked this comment was indeed “it should be ‘crime per hacker news commenter’ not ‘crime per square mile’”, and then I tried to make it more useful.

I’d be more interested in square mile “zones” and seeing which zones are the most dangerous.

Separately you could see which cities contain the most “dangerous zones”.

You’d need each crime to be reported with it’s location. Import them as ‘points’ in a Geopandas dataframe, import LA statistical reporting zones as a separate dataframe of polygons, then for each crime point, get the zone it falls within, then group by the zones.

More crime incidents doesn’t mean more dangerous. For example a shooting is substantially worse than say petty theft without any victim physical damage.

General lawlessness leads to violent crime. If people can get away with anything, some of them will just do anything they want. It’s just a hop skip and a jump from stealing to robbery to murder.

That's ridiculous. The motivations are completely different.

Where do people’s motivations come from? You’re unenlightened if you think it’s ridiculous.

People do a lot of things just out of sheer mimicry. If you air suicidal programming on TV, suicides reliably go up in the broadcast area.

Anyway, this specific area was already researched, it is called the Broken Windows theory.

“According to a 2001 study of crime trends in New York City by Kelling and William Sousa, rates of both petty and serious crime fell significantly after the aforementioned policies were implemented.”

https://en.wikipedia.org/wiki/Broken_windows_theory#Criticis...

This is exact same attitude that has gotten SF into this mess.

Income inequality and crime correlate? Would love to see some studies.

https://www.brookings.edu/blog/future-development/2022/08/01...

https://journalofeconomicstructures.springeropen.com/article...

https://web.worldbank.org/archive/website01241/WEB/IMAGES/IN...

https://www.nature.com/articles/s41598-020-80897-8

https://www.economist.com/graphic-detail/2018/06/07/the-star...

Sources:

https://en.wikipedia.org/wiki/List_of_United_States_cities_b...

https://docs.google.com/spreadsheets/d/105IDODri5Qbi3Mj-SGHF...

What is the SI unit for danger?