By now I’m quite used to Taleb (Fooled by Randomness, The Black Swan) complaining that real-world phenomena often aren’t normally distributed. I was a bit surprised to find Rory Miller (Meditations on Violence) making the same point, that the bell curve isn’t universal:
According to the research done by Dennis Culhane, it turns out the most common length of time for a person to be homeless is one day. The second most common is two days. These short time, one-time homeless account for eighty percent of the homeless. People are people and they are adaptable. If they find themselves homeless and don’t like it they will overcome and get on with their lives.
There were about 10% who come in periodically for a couple of weeks, usually in winter. The last 10% were the chronic homeless. It was this group that make up the people that most of us think of as homeless, whether you think of them as pitiful and severely disabled or alcoholics and grifters.
This means many things. First and foremost, it means the problem is small enough to solve, not just treat.
Philip Mangano, mentioned in the article, has a solution to the problem of chronic homelessness: it would actually save money to give them a nice apartment and provide for all their needs with a dedicated staff of social workers. It would be cheaper than it is to pick up their bills for Emergency Room visits and jail time.
I had a solution, too, but society isn’t ready to let people die. I firmly believe that when a safety net begins to enable, it must be removed from that individual. If the person still continues to behave in a self-destructive way society should have no guilt when they suffer the consequences. But that’s me — I’m aware that I don’t exactly have a standard outlook on problems.
Side thought (and there were many side thoughts from this article) my instinct, when given a problem, is to solve the people (shut down the threat, train the rookie, counsel the errant) to change them in a crisis or help them change themselves… others, including Mangano, solve the environment.
More side thoughts — criminals also follow this distribution. Violent crime is committed by a relatively small percentage of criminals, and they do far more than we ever get them for. Solve the problem or solve the person?
The article applies this to police misconduct — the vast majority of officers do an excellent professional job, a small percentage are asses. The whole idea of the standard response to negative media attention (more sensitivity training) is based on the bell curve assumption. Mass training always is trying to shift the curve a little bit to the ‘saint’ side. The trouble is that when you have a distribution that runs closer to 30% saint; 25% hero; 20% good guy; 15% civil servant; 7% lazy bastard; and 3% asshole the training insults 75% of your people and the 10% you’re trying to reach either don’t care or won’t act. Again, when the real problem is this small, you can solve it. I prefer firing, but our agency has a tendancy to put the worst officers in positions away from the public, which sometimes involves a promotion. Solving the environment instead of the person.