Ants and Non-Selfish Routing
Faithful readers of Traffic will know of my fascination with the traffic organization of ant colonies. I’ve just been reading a new paper, “Priority rules govern the organization of traffic on foraging trails under crowding conditions in the leaf-cutting ant Atta colombica,” published in a recent issue of The Journal of Experimental Biology, by Audrey Dussutour and colleagues from France’s Université Paul Sabatier.
Dussutour, working with a colony at the University of Illinois, manipulated a bridge on an ant trail so that it was too narrow for two opposing streams of ants to pass abreast. A clear pattern emerged: Ants heading out to the food source always gave way to returning ants that were laden with food (some of which were followed by ants without food). A set of “clusters” emerged, which had its own interesting pattern in same-direction traffic: Even though the ants returning without food were in theory held up by the slower, leaf-carrying ants, those ants still refused to “jostle” past. The results of this strategy were worth noting:
As unladen ants move on average faster than laden ants, these ants were thus forced to decrease their speed. By contrast, this decrease was counterbalanced by the fact that, by staying in a cluster instead of moving in isolation, inbound unladen ants limit the number of head-on encounters with outbound ants. Our analysis shows that the delay induced by these head-on encounters would actually be twice as high as the delay induced by the forced decrease in speed incurred by ants staying in a cluster.
A strategy that appeared to be slower for some individual ants actually benefited the colony as a whole; this is a pattern that often does not hold in human traffic — when, for example, individuals change lanes in unstable traffic, perhaps temporarily improving their own position but having what Benjamin Coifman terms a “butterfly effect” on the lane they have moved into, as well as the one they left.
The French team’s experiment reminded me of a passage from Robert Frank’s book The Economic Naturalist. Frank, based at Cornell, writes about the quaint old one-lane bridges around Ithaca, New York. He notes that a “first come, first served” social norm has emerged at the bridges, so that a stream of steady traffic from one direction wouldn’t hold up cars from the other direction for an undue amount of time. Typically, self-restraint, as in the case of the ants, can help improve overall efficiency.
But when traffic is heavy from both directions, he notes, this norm actually penalizes drivers. As he writes:
“Suppose a ten-car caravan arrived from each direction, with ten seconds separating the cars in each caravan, and with the first driver in the northbound caravan reaching the bridge a split second before his counterpart in the southbound caravan. If no one followed the first-come, first-served norm, all northbound cars would cross the bridge, after which the ten southbound cars would cross. Northbound cars would experience no wait at all, and as readers with a pencil, paper, and a little patience can easily verify, the southbound drivers would experience total combined waiting time of twelve minutes and thirty seconds… In contrast, if all followed the first-come, first-served norm, the first northbound car would cross, followed by the first southbound car, then the second northbound car, followed by the second southbound car, and so on. If you are patient enough to add up the relevant waiting times, you will see the total waiting time would be 80 minutes—37.5 minutes for northbound cars and 42.5 minutes for southbound cars—more than six times as long as when there was no norm.”
Of course, at construction sites and the like, where a flagman is present to wave clusters of vehicles through, this problem does not exist or is mitigated.
I am not sure what the implication of this is. Perhaps we humans simply prize courtesy over rote efficiency (though overall the logic of traffic seems to be that everyone pursues his or her individual efficiency, beyond any impulse towards altruistic politeness). Perhaps it is because we have not evolved to act in concert, as ant colonies have (as Dussutour, et al. note, “ants from the same colony presumably act with a unity of purpose very different to the multiplicity of individual interests pursued by pedestrians or drivers moving in a traffic stream”). Perhaps the Ithaca bridges are simply outmoded in an era of heavy traffic. And on those Ithaca bridges there’s no clear hierarchy of commuters, as in the ant example. But it’s not a stretch to say that a bridge metering system, perhaps inspired by some ant-traffic-derived algorithm, would get people home faster than the traditional way of doing things.
This entry was posted on Wednesday, April 1st, 2009 at 7:58 am and is filed under Etc., Traffic Wonkery, Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.