The Crime Rates thing

One of the problems reactionaries draw attention to, as an example of the ineffectiveness of the modern state, is the threat of crime.
To this, progressives respond with statistics showing that incidences of crime per capita have been on a general downward trend for the last few centuries. Stephen Pinker recently published a book on this point, and anti-reactionaries are, understandably, making a big deal of it.
There are various measurement difficulties with crime rates, but Pinker isn’t a climate scientist, and so it’s not likely that all the measurement errors are going in the same direction.
Are perceptions of increased danger just wrong, then? They could very plausibly be the product of media sensationalism and heightened expectations. But I have doubts that frequency of crime, particularly frequency of crime per thousand of population, are really the right measure.
After all, fear of crime isn’t a passive, background thing. You don’t sit and worry that you’re going to be mugged before your next birthday. You worry when you walk home from the railway station at night. Fundamentally, you worry when you go out of a highly protected area.
So the intensity of your worry depends on what the risk is of going out of a protected area for a time. If people are doing that much less often than they used to, then it might have got more dangerous to do so, yet incidence of crime will be down.
There are a few reasons why people might be doing dangerous things less often. Many people have cars, which are mobile fortresses with lockable doors. That’s our old friend, advances in technology masking other problems. It might be that some areas that were dangerous are now safe. That would be a genuine improvement, a real diminution of crime and the threat of crime. It might be that some unsafe areas are now so unsafe that people rarely go there at all, which would actually be an increase in the threat of crime, without showing in the statistics. Then there’s the scale question:
Fleet Street in London is still Fleet Street. It’s the same length it was when Pepys walked down it, from Ludgate Circus to Chancery Lane.
If crime rates are the same as they were a hundred years ago, does that mean there should be the same number of crimes on it, or a higher number in proportion to the change of population? Or in proportion to the change of population divided by the change in the total length of road? But should that be all road, or important town-centre type street? Society does not scale linearly.
It would be interesting to see a town that stayed approximately the same size over decades, and look at its crime statistics, to eliminate these difficult scaling factors. But a town that stayed the same size would almost certainly have changed its role, relative to other towns. Not so helpful.
One thing that occurred to me is a football match. A big match is like a temporary town — in population, it’s the size of a small town. Is a football “town” the same size as a hundred years ago, and is it more or less violent?
Oh. The answer is that it’s smaller, because the level of violence had grown so much by the 1980s that the government forced reductions in stadium capacity, along with other measures.
That’s one indication that, isolated from scale effects and technology, violence has become worse.
Of course, history goes on: football stadiums are now less violent than they were in the 1980s, and for all I know less violent than they were in the 1930s, because they have become highly-policed zones.
The simple summary: there is no simple summary. Waving the hand and talking about “more crime” is wrong, at least relative to population. Some areas of life have become less policed, while some have become extremely heavily policed, and much safer. The things we do that expose us to risk of crime have changed, partly due to technology, partly due to bigger towns and cities, and partly due to social changes. If you want to show that some particular activity has become more dangerous today than in the past, concentrate on that — specific evidence is somewhat clearer than aggregate statistics. Moldbug’s “Map of areas a person can walk alone in confidence” is the right general approach (thanks to @lexcorvus for the link).
Update. I still don’t feel that I’m hitting the nail on the head. I feel that crime is something that should be at the edge of society. We have expanded — scaled — society, but we are measuring crime rate relative to the volume, not the size of the edge. A sort of square/cube law error.