Of fire ants and feedback loops

by Benjamin Kuipers

Many complex systems include multiple feedback loops, often both positive and negative ones. These can interact in ways that lead to counter-intuitive responses to what seem to be obviously correct actions. Let me give you an example.

Here in Texas, we have a nuisance called fire ants (solenopsis invicta). Nests of little biting ants that leave infected pustules from their bites and stings. They migrated in from South America somewhere, and for a while were confined to about a square mile near Mobile, Alabama. A few prescient scientists suggested spending the money necessary to saturate that area with the appropriate insecticide and kill them off. They weren't listened to.

Several decades later, people decided they had a serious problem, and went all out to kill them off....

This very expensive effort came to be known as "the Vietnam of entomology". The fire ants won.

They are now a permanent part of the American south, and we learn to deal with them. Here's the interesting thing about feedback systems.

Suppose you discover the first few fire ant mounds on your property, and you decide to move quickly, and eradicate them completely, just like people should have done in Mobile, many years ago. You buy a good strong ant poison and kill every last one of them.

The next year, or possibly the year after, instead of 10% fire ants on your property, you have 90% fire ants. Why?

It turns out that fire ants move quickly into unoccupied territory, but they are not all that great against other species of ants who are already established there and aren't bothering you at all. However, if you spread general-purpose insecticide around, you kill the fire ants, but you also kill all the other indigenous ants. The next year, there's this nice big empty territory, with no other ants to compete with, and the fire ants move in a lot quicker than the other species.

(Incidentally, if you really want to control your fire ants, instead of broadcasting a general-purpose ant poison, you use slow-poisoned bait, and sprinkle it by hand into the foraging area around each individual mound. They take it home, feast and die, leaving their territory to their neighbors. Fortunately, fire ants have conspicuous mounds.)

But notice the paradoxical dynamics. An "obviously correct" strategy ("kill all the fire ants") that successfully defeats fire ants on the time-scale of days, actually encourages the fire ant population on the time-scale of seasons. A slower strategy requiring more patience ("feed them tempting poison") actually works. (For people who don't even have the patience for the bad plan, a popular approach is pouring boiling water onto the mound. This is a lot of work and gives you the bodies of a bunch of dead worker ants, but it has virtually no effect at all on the mound.)

The analogy with terrorism shouldn't be pushed too far. But with a little practice at looking for these kinds of second-order effects, and particularly looking for paradoxical responses where the long-term effect is in the opposite direction to the short-term effect, it is amazing what you see out there.

People will persist in doing things that seem sensible, but then turn around and bite them on a longer time-scale, and they'll never understand that they are causing their own misery. Simple aphorisms like "Don't eat the seed corn" are compiled warnings about this phenomenon.

Written 1 May 2006. Minor errors corrected.