Slime molds have evolved to produce some of the most efficient networks seen in nature, but just how good are they? As good as the notoriously complex Tokyo rail system.
The slime mold Physarum polycephalum forms networks between food sources that are as efficient and fault tolerant as anything we can design. In order to test quite how good the mold is, and if it could be used to help human planning, a group of Japanese and British scientists pitted it against the Tokyo rail system, and found a surprisingly efficient result.
The scientists created a simulacrum of the area surrounding Tokyo, and placed the mold directly on the location of the city itself. Neighboring cities were marked by food sources. In order to replicate the geographical limitations of the area (mountains and lakes), light was used to create regions the mold would avoid, as it doesn't like illumination.
The mock-map was 17cm wide, and over the course of 26 hours, the mold spread out evenly, colonized each of the food sources, and then refined itself down to a network of tubes.
This resulting map and the current Tokyo rail system were then compared to a minimal spanning tree (MST) analysis of the location, which is the shortest possible network connecting all the cities (and is decidedly lacking in intercity links). The analysis was run over three factors: cost, transport efficiency and robustness. The existing Tokyo rail had a cost of 1.8 times the MST, where the Phyrasum network was 1.75 ±0.3. For transportation efficiency, the two were almost identical, with Tokyo scoring 85% of the MST, and the mold 85% ±4%. The only area where humanity's version was particularly better was robustness. In Tokyo, only 4% of faults in the network would isolate any part, where with the grown network it was up to 14%±4%.
This isn't the first time slime molds have been used in this way, as a similar experiment was undertaken recently with motorways around the UK. However, the combined Tokyo/British research has gone a step further, and mathematically modeled the way the mold grows, and simulate it. With this, they were able to find a map that offered lower cost than the MST (0.7x), and the same efficiency as the existing system.
With this model, the ability to map potentially complex networks can be undertaken, not just for rail or car, but also potentially for self-organized networks without centralized control, like mobile ad hoc networks, wireless mesh networks, remote sensor arrays. Or, if you're planning on founding your own country, all you need is some agar and slime mold to create an efficient highway system.
[from Science]