Blog/2025-08-14/Crime Rates Are Incomplete

A strongly opinionated visualization proves anything

A common argument about crime rates is that big cities in the US are actually much safer than one might expect. New York's is 4.1/100k while Denver's is almost three times that at 11.9/100k. People often consider big cities less safe, though, and the stats wonks always find this puzzling. But the answer seems obvious to me: people consider themselves to be secondary victims of crime if they have witnessed sufficiently unpleasant things. If a man is killed in front of you, it doesn't matter that you yourself weren't killed. It only matters that a man was killed in front of you.

To a person witnessing a homicide, I don't think it's particularly convincing to tell them that the chance that they'll be the next person killed is 4.1/100k or whatever. You saw a person die, which is unpleasant, and you sensibly decide that this is a place where people are killed. This isn't outrageous. In fact, with guns collateral damage seems inevitable since the people wielding them are rarely good at using them. However, from experience I know that walking by a crime scene after it has been handled rarely feels like anything. You would never know. I bicycled by Twin Peaks at six in the morning, some five hours after a man was murdered there and it didn't feel like anything.

One thing I thought would be fun was to just take a toy model of this and see what it produces. This is fairly simplistic, but what if we said:

• There are people walking around a city
• They can see things within some range of them
• There's a guy who will kill one person
• Let's count how many see it happen
• Let's count how many see the dead body
• Let's assume the dead body disappears after a while

And then one huge one which is that we guarantee a homicide, i.e. no matter how small the population we make sure someone's gonna die. This will force a sparse area to have a high homicide rate.

Visualization for ▼ city-homicide

Fairly simplistic, and if you're lucky the criminal will kill someone in a hidden alley, but if you run the visual a few times, varying the population you'll see that it kinda has the results we'd expect:

• The higher the population, the more witnesses there are
• The higher the population, the more likely someone sees the dead body
• The lower the population, the higher the homicide 'rate' (because we made it so)

So this rather simplistic model explains people's belief that big cities are unsafe. They are more likely to see violence and its results there than in rural areas. And that's what informs someone's opinion as to what's happening. So people are always going to claim denser areas are more unsafe than rural areas unless there is a massive difference. It just comes down to whether they saw something bad done or not.