I recently had the pleasure to spend a month doing research at the Weizmann Institute of Science in Israel. This Institute ranks within the top ten every year in terms of research quality (the proof is here) and one of the projects I was able to work on, Antlab, was published in the journal Nature Communications just after I got back. The researchers were looking at how groups of ants are able to collectively carry large chunks of food back to their nests efficiently, and how this might be useful in areas like computer networking or morphing algorithms.
They found that when carrying heavy items, a larger number of ants act as the load bearers and carry the food in a certain direction whilst a small number of ants act as “scouts” and occasionally join the load bearers to push and steer the food before stepping back to see how their correction affected the motion of the food.
The ants they were observing were called the Longhorn Crazy Ants, so called because of their perceived erratic behaviour when it comes to find their way on their own. But according to Dr. Ofer Feinerman of the Institute, this individualism mathematically matches the levels of conformity amongst the ants perfectly – in fact, about 90% of the time they will follow along, and 10% of the time they will behave erratically, as their name suggests. This allows the ants to properly carry heavy items together as a group, as only a few of the ants are acting as the “scouts” by not conforming to the group direction of motion, with the rest all just going with the flow. The erratic behaviour of the scout ants is apparently crucial as it allows any ant with new information to interject and change the direction of the group.
The Scientists went further to start adjusting the levels of conformism and non-conformism by changing the size of the items the ants had to carry. Their model predicted that with larger items, more ants would be needed to carry it and thus more would behave as conformists rather than as the individual scouts. Whilst this vastly improved the speed and ability of the group to move in straight lines, with very few, if any, scouts available, navigating around any obstacles became impossible.
There ended up being a certain size of objects for which this model works best. Medium sized items, such as a cheerio, are large enough to require conformist load bearers, but not so large as to mean there are no available scouts.
You may be thinking: “Great, so now I know how ants carry food. How does this affect me?” There are, in fact, some quite staggering implications. Most notably are perhaps morphing algorithms, which start off as fairly basic algorithms that aim to improve as more information becomes available. Think of every line of code as an ant, and see that as certain “ants” more accurately or efficiently code for the scenario they are working on, they become the load bearers of the algorithm. The lines that don’t work so well are changed in a similar way the scouts jump in to adjust the direction of the algorithm. Eventually, as the code morphs through various generations, you get an algorithm that works very efficiently, and can be used to protect secure data (or hack it), model complex scenarios (like wind tunnel measurements), or even work out exactly how much power your smartphone needs when running certain tasks. So we all owe Antlab one big thank you, cheers to your (hundreds of thousands of) ants!
Editor: Serena Liles