Organized Complexity - "Unaverage" Clues

[Unaverage clues] are often the only announcers of the way various large quantities are behaving, or failing to behave, in combination with each other. “The kind of problem a city is,” The Death and Life of Great American Cities (p443)

A few years ago, I was helping collect data for a citywide traffic count. Unlike many traffic counts, automating the count as cars travel over pneumatic tubes, this was a pedestrian and bike traffic count, requiring the coordination of manual counters at various points across the city. I was one of those counters at what turned out to be one of the most trafficked points.

The resources required for such a comprehensive statistical study aren’t insignificant. So what do you do if you don’t have these resources and no one’s collected data for your area?

This awareness of “unaverage” clues—or awareness of their lack—is, again, something any citizen can practice. “The kind of problem a city is,” The Death and Life of Great American Cities (p443)

I had four clickers: two for pedestrians and two for bicyclists. I reset one pair every 15 minutes while continuing to take the count on the other. I did that for two hours. Over that time, two unaverage clues sat down on either side of the bench I’d found with a great view of the count. They were panhandlers.

Statistically, these two men were inconsequential - two clicks out of hundreds. However, their presence is a clue to the behavior of many more people. They indicate that a large number of people walk by this location. A fraction give them money. They have every economic incentive to find a valuable location. As long as they’re quiet, not hassling anyone, and in the public right-of-way, no one can tell them to move.

These unaverage clues, in professional-speak, are proxies - variables with close correlation to another variable.