There is a video that is currently being shared on social media by a number of people I know. It is about the reintroduction of wolves into Yellowstone and the wide-ranging effects this “keystone species” had on the environment. A keystone species is a top-of-food-chain creature whose presence or absence has wide-ranging impacts. In the case of wolves, which had been hunted and eradicated, their reintroduction managed to stabilize elk population and a number of other species as well. Even more, vegetation was affected through secondary means. Indeed, the narration in the video says even “the river adapted” to the reintroduction (fewer grazing animals => less deforestation near rivers => strengthened river banks).

All this is terrific news. But there is a way that this gets reported that gives me pause. The description of the causal loops are presented as if all this was preordained. As if it was obvious that the rivers would strengthen their banks if only there were wolves.

I wonder if the planners really knew that would happen, or whether it is an after-the-fact understanding. Large system responses can be like that: in hindsight it seems obvious. But that is can give the dangerous illusion that we can predict the future if only we are smart enough.

So we get system “flow diagrams” like the one below, which seem at once reasonable and like a Rube Goldberg device:

Click for full size

Click for full size!

The point about some of these large-system effects (eg, the “butterfly effect,” an idea popularized by Edward Lorenz, where the beatings of some butterfly’s wings in one place causes a windstorm elsewhere) is that they are unpredictable ahead of time. We know there are likely to be effects, but cannot say for certain what they will be.

It is tempting for hindsight to make us feel smug when looking at our forebears. They should have known, it is easy to think to ourselves. But the nature of such large, open systems as that this is not always the case. It is impossible to know and account for all initial conditions.