This month’s New Yorker contains and article by Malcolm Gladwell on a power law approach to social and environmental policy. Power Law holds that in many cases, social problems are primarily caused by a small minority of cases that are statistical outliers.  For example, he argues that up to half of automobile pollution in Denver is caused by a small percentage of the most inefficient and ill-kept automobiles.
Gladwell thinks that too much inefficiency takes place in current social policy – such inefficiency comes from an assumption that people comply with laws, and that current approaches have a possibility of alleviating social policies.

He uses the example of homelessness:

The cost of services comes to about ten thousand dollars per homeless client per year. An efficiency apartment in Denver averages $376 a month, or just over forty-five hundred a year, which means that you can house and care for a chronically homeless person for at most fifteen thousand dollars, or about a third of what he or she would cost on the street.

Efficiency tells us that it is more effective to simply pay for apartments and services for the homeless. This is a small fraction of the hospital and policing costs of the most hard to do cases. And while this may not provide the transition to wellness and respectability that is expected by most social programs, it is far more reasonable to accept that some of the most difficult cases may never be able to live a life without this support.
The problem, he says, is that while efficiency tells us to simply deal with the problem by providing apartments and direct care for the hardest cases, our sense of justice and equality tells us otherwise – that we should offer similar services to all people; why should the working poor not get a free apartment? The solutions offered by power-law are palatable neither by the left of the right.

However, Gladwell runs into a problem that he seems ill-prepared to solve. If it is more efficient to simply treat the worst cases, then why is it not more efficient to treat the problem as a whole? And what is the line that divides the problem cases from those that are solvable? The problem is the dividing line between the hard and the very hard problems, and deciding at what point power law should rule.

LINK