Imagine that your personal (in this case, reproductive) success and your life depend on effective communication (in this case, to find sufficient food), and that you can choose between two methods to achieve this success: communicating solely with a varying set of friends (in this paper, a “newscast”), or indiscriminately listening and talking to whoever happens to be in the (geographical) vicinity (a “multicast”). What would you choose? If you are undecided, on which parameters should you base your decision?
This refreshingly vivid, easy to follow, and comprehensive paper discusses this issue for a set of agents (initially 1,000) living in the computer-based “Sugarscape,” a 50-by-50 grid with randomly distributed variable amounts of sugar, on which agents feed. Based on limited vision (a few squares) and overheard communication, the agents decide to move to another square; this cycle is repeated for several thousand iterations. When two agents of the opposite sex meet at the same place, they may mate; the offspring will inherit suitably mutated genes, determining its communication behavior.
After some simulations, the authors conclude that populations using newscasting invariably die out after approximately 100 generations, while populations using multicast communication typically survive. The authors trace this outcome to the fact that multicasting (in their design) permits the transmission of outdated information; namely, listeners will not find the indicated amount of sugar. Hence, the key question is one of establishing a suitable tradeoff between the velocity of information dissemination and the rate at which information loses its value.
I have to add a serious objection, though. From the statistics provided, another cause not addressed by the authors also seems to apply: it appears that only three percent of all newscast messages actually carry information about nonzero amounts of sugars (as opposed to 21 percent of multicast messages). It is therefore not astonishing that agents relying on their “friends” to tell them about new food sources eventually die out; when your friends don’t know, it is no use listening to them. I hope this issue is resolved in future work.