Solving for Extremes

Yesterday I was having a conversation with a few friends, one of whom works for the city of New York. She was talking about the smoking and trans fat bans here in New York and how a big part of the reason they exist is to be preventative. Health care follows a power curve: A small portion of the population drains the vast majority of the resources.

The idea of New York City's bans, then, is to turn around that situation and forcibly remove the most unhealthy aspects from the lives of the perpetually sick. It's an interesting strategy and I think it's a fundamental shift in thinking. As a society we have a tendency to treat symptoms, not causes. Just think about it: We have gum for bad breath, tylenol for headaches and, one of my pet peeves, fad diets. In the case of the last one, I was always seriously bothered by Atkins, for instance, because it didn't seem to teach people about calories (I've never read it and may be wrong). So rather than understanding the single most important factor in weight loss, people were eating a lot of cheese and then surprised when they gained weight again. (Just for the record since most of you probably don't know this, about four years ago I lost eighty pounds. I think this gives me some authority to speak on weight loss.)

Anyway, the point of all this is I've been feeling like all this stuff relates back to networks and Paretian distributions. My basic hypothesis (and this is very rough at the moment) is that we've spent the vast majority of time treating symptoms because we understood the world to be a bell curve where there was some giant average that most people fell in to. In that world, treating symptoms makes more sense because most of the people are kind of sick on occasion and just need a little Tylenol to get them through today because they'll be fine tomorrow. The thing is, that's not the world we live in. Rather, the vast majority of us are basically never sick and there are a few people who are constantly sick and draining resources. Rather than trying to treat the middle, then, organizations need to solve for the extreme.

In the article "Million-Dollar Murray" (which I just rediscovered . . . I knew I had read this someplace), Gladwell writes about abusive cops in LA. When the department charted out the complaints against cops they didn't get a bell curve with some average number of complains in the middle, rather they got a power curve: Most cops had 1 or 2 complains and a few of them had the majority of the rest. In the article Gladwell writes:

The report gives the strong impression that if you fired those forty-four cops the L.A.P.D. would suddenly become a pretty well-functioning police department. But the report also suggests that the problem is tougher than it seems, because those forty-four bad cops were so bad that the institutional mechanisms in place to get rid of bad apples clearly weren't working. If you made the mistake of assuming that the department's troubles fell into a normal distribution, you'd propose solutions that would raise the performance of the middle—like better training or better hiring—when the middle didn't need help. For those hard-core few who did need help, meanwhile, the medicine that helped the middle wouldn't be nearly strong enough.

"The middle didn't need help," that's a fairly fundamental departure. Turns out homelessness works the same way, the vast majority of homeless people are only without a place to stay for a short period. It's only about 10 percent who are "the chronically homeless, who lived in the shelters, sometimes for years at a time. They were older. Many were mentally ill or physically disabled, and when we think about homelessness as a social problem—the people sleeping on the sidewalk, aggressively panhandling, lying drunk in doorways, huddled on subway grates and under bridges—it's this group that we have in mind." In addition these people are taking up a disproportionate amount of resources. It's estimated New York City spends 62 million on about 2,500 perpetually homeless people.

So what do we do? We don't throw them on the streets, but we can reassess how we spend money. Just thinking about the homeless case, when you understand that you're spending 62 million on 2,500 people you start to wonder if there might be better ways to spend the money. By focusing time and resources on edge cases and extreme conditions we may find ourselves far more successful in dealing with the problems we face in a connected world.

I don't think that's a particularly useful conclusion, but I just needed to get some of this down. I've been thinking about how to apply this stuff non-stop lately and I'm sure you'll see more. If you have any thoughts or additions please leave them in the comments as usual.