What’s the Distribution

Many of my posts are in response to something I’ve read, and I can’t think of a writer who so easily inspires a response than Malcolm Gladwell. His writing is entertaining and insightful, ever probing an everyday issue from a counterintuitive perspective.

Last month, he wrote about the power law distribution of homelessness. That is, if you were to chart the users of homeless shelters by their number of stays, you wouldn’t get a normal distribution (from left to right: gradually sloping upwards, peaking at a certain number, and then gradually sloping downwards) but rather a power law distribution (high numbers at the far left, sloping drastically downward and leveling off around zero).

What this means is that the high majority of homeless shelter users are only there for a day or two: 80%, according to research Gladwell quotes. Another 10% are periodic users, dropping in for a week or so, getting back on their feet, only to return a few months later. The last 10% are the chronic users: often mentally ill, staying in shelters for months at a time.

This last group is the one people associate with when they think of the category, “the homeless.” They are also very costly to care for, in term of their constant use of both homeless shelters as well as emergency health services.

Gladwell argues that our response to homelessness has assumed a normal distribution, when perhaps a more cost-effective approach is to provide high-intensity services to the chronic users. I read this argument somewhere else, that if a chronic user that is being treated with our current homeless solutions is costing us fifty to sixty thousand dollars a year in services, it is actually more cost-effective (and ,you would assume, better for the person himself or herself) to provide a home and enough amenities to turn things around.

I’m not here to argue homelessness from a political or moral standpoint. (At least not today.) What Gladwell’s article sparks in me is this notion that understanding the distribution of the problem is key to adopting the appropriate solution. Too often, we assume normal distributions and gradual slopes, when perhaps in reality we are looking at tipping points and sharp slopes.

As I’ve said before in this space, if you’re willing to give up 10% in value, sometimes you don’t save 10% in cost, you can save 50%; and if you’re willing to pay 10% more in cost, sometimes you don’t get 10% more in value, you can get 50% more. And as I’ve also said before in this space, sometimes it’s better to pay a dollar now, no matter how hard up you are in being able to spend that dollar today, then to have to pay ten dollars later. So after reading Gladwell, I’ll be extra careful to keep my eyes open for what distribution we’re really looking at.