Error is architecture

Error is architecture. Imagine five hundred ants are exploring a space. Number four hundred and eleven finds the sugar completely by accident. Except it’s not an accident because the ant has left a scent train.There are five other hundred ants, so surely one other will come along and make the same mistake. Suddenly you’ve got two chemical trails on top of each other, doubly hot, attracting a third ant, which makes it triply hot, attracting five more, which makes it octuply hot. Suddenly what you’ve got is a blazing hot pheromone highway of bumper-to-bumper ants.

Radiolab’s episode on the science of emergence is, as usual, amazing. This particular phrase stuck me, though.

Error is architecture. In some cases, error isn’t an accident. In many cases, error is part of the process of discovery.

What I find extraordinary about emergence is that the same ants that wander haphazardly are the ants that guide other ants to a food source once they find it, though they don’t do so deliberately.

The same genes that randomly mutate are those that can mutate into a useful trait and be passed down and amplified by natural selection.

What also occurs to me is that several of the topics I’ve been interested in lately—social proof, filter bubbles and spirals of silence—are all emergent phenomenon.

There is certainly an interesting parallels.

On Radiolab, E.O. Wilson described the way he demonstrated the ant pheromone organ shortly after discovering it.

For demonstrations, I would write my name [in pheromone], and a column of two hundred, three hundred, four hundred ants would come pouring out. And they would write my name in ant.

In many respects, this is an emergent behavior gone wrong. An ant colony has no interest in spelling out E.O. Wilson’s name, as cool as that might be (I’d love to see a video of this).

Another example of an emergent behavior being subverted is Robert Cialdini’s description of buffalo herding as a part of his discussion of social proof in Influence.

This aspect of the social proof phenomenon always reminds me of the way certain Indian tribes—the Blackfeet, Cree, Snake, and Crow—used to hunt the North American buffalo. There are two features of buffalo that make them especially susceptible to erroneous social evidence. First, their eyes are set in their heads so that it is easier for them to see to the side than to the front. Second, when they run, as in a stampede, it is with their heads down low so they cannot see above the herd. As a result, the Indians realized, it was possible to kill tremendous numbers of buffalo by starting a herd running toward a cliff. The animals, responding to the thundering social proof around them—and never looking up to see what lay ahead—did the rest. One astonished observer to such a hunt described the deadly outcome of the buffalo’s obsessive trust in collective knowledge.