Cell phones didn't cause decline in crime, plus a rant on pundits

Steve Sailer floats the idea that the greater presence of cell phones could have been a key factor in the '90s crime decline. Here is a graph that shows the number of cell phones per capita (data here), along with the homicide and violent crime rates (data here). Click to view full size.



The crime rate isn't measured on the same scale as cell phones per capita, so I divided the crime rate data by its maximum. The curves look the same as the original ones, but now they fit onto the same graph as the cell phone curve. Here is a table of the cell phone data, to better see the small numbers early on:

year	cells/cap
1984	0.0004
1985	0.0014
1986	0.0028
1987	0.0051
1988	0.0084
1989	0.0141
1990	0.0212
1991	0.0300
1992	0.0433
1993	0.0621
1994	0.0927
1995	0.1286
1996	0.1660
1997	0.2067
1998	0.2561
1999	0.3156
2000	0.3890
2001	0.4507
2002	0.4888
2003	0.5458
2004	0.6203
2005	0.7011
2006	0.7783

The crime rate slows throughout the late '80s, peaks in '91 - '92, and declines afterward. Cell phone prevalence increases steadily from 1984, but not to a high enough level to affect crime. Between '92 and '93, only about 5% of Americans had cell phones, and these were obviously the richest 5%, not people who would frequent the bad sides of town to report crime in progress. Most victims wouldn't have had one.

Remember the 1995 movie Clueless? A running joke is that in Beverly Hills, people are so filthy rich that everyone has a cell phone. And in the 1996 movie Scream, the police make a big deal out of a suspect who has a cell phone -- it being very unusual. In fact, a contemporary review of the movie describes the character as "Sidney's cellular-phone-toting boyfriend Billy," as though you're as likely to carry a cell phone as a gun. None of the victims have cell phones to call for help. If they were still considered queer through 1996, they couldn't have been common enough to affect crime much.

What percent of people need cell phones before they could make a dent in crime rates? I don't know, a little less than a majority? Between 2000 and 2001, 40% of people had cell phones. In the 2000 movie Urban Legend 2: Final Cut, many of the characters have cell phones, and they're just normal college students this time. They all try to call for help using them when the killer is approaching. But by this point in time, the bulk of the decline in crime had already happened.

In general, the problem with every guess about why crime declined in the '90s is the it can't account for why crime declined in the '30s through the '50s, why it rose during the 1910s and '20s, as well as during the '60s - '80s. See this graph of the homicide rate during the 20th C. Crime rates rise and fall periodically, so we don't need an ad hoc explanation for why crime decline or rose in a certain period -- it's just going to. It's like asking why a pendulum went left, stopped, then went right -- there's no outside intervention, that's just what pendulums do. The interesting question is why do crime rates rise and fall?

For that, we build a system of differential equations that models how criminals, police, and victims interact with each other. It's pretty easy for such models to show oscillations like we see in the crime rate. They don't have to, though. It may be that only if the parameters satsify some inequality, there will be oscillations -- the parameters being things like "the average cost criminals incur when caught," or "the average benefit criminals reap by getting away with a crime," or "the size of the police population," etc. I'm sure there are people doing this, because physics and biology are too overcrowded, and mathematically trained people are starting to spread into uncolonized lands -- uncolonized by anyone who knows how to keep track of flow rates, that is.

Some criminologists are just dumb. Others are smart and know a lot more about the literature than I do. It must be that this kind of thinking -- look right before X happened to find its cause, and find a separate cause for each effect -- is hardwired into us. It probably didn't pay much in Darwinian terms to build models of crime rates, or other things that are complicated and need a fancy model to understand. Our brains are designed to answer simple cause-and-effect questions like, "Why did Bungo die? He was hit with an arrow." Or "I touched that plant, and then got a rash."

For some reason, anything that oscillates, we can't understand intuitively, even when it's really simple. Take fashion -- fashions change just to be different from what was popular before. Nothing more. And yet, most people ignore that truism and seek some deeper immediate cause -- women are wearing low-rise jeans as a reflection of their sluttier behavior (the purported cause didn't happen), or sideburns on men are fading away because they want to look cleaner. No, they don't want anything -- except to look different from how they looked before.

But why don't most people who get paid full-time to understand crime not get it? The reason is that a clever theory is sexier than figuring out how something works, and People who Publish are 99% attention whores who just want a big audience. So, sexy trumps correct.

It's the difference between what a priest does and what an auto mechanic does. Who gets more props? And who actually knows what he's doing, and can prove it? We don't want our social world to be dissected like a car engine, so we ignore people who can figure stuff out. We look at social life as a precious living thing, something that has a deep secret essence that only a select few can peer into. We call them priests, pundits, or whatever, and listen in rapt attention to their charismatic tea-leave readings.

Luckily, modern states have neutered the religious priesthood from fucking stuff up, were their visions taken seriously. Unfortunately, our punditocracy get to hobnob with those in power, whisper their visions into their ear, and get us into a big mess. Once I'm elected dictator, I'll choose policy advisers like I would engineers for a nuclear power plant: if you can prove you know what you're talking about, and you're better than chance in predicting things, you're hired. Otherwise, Smithers, release the hounds.