| How Ants route Information | 2003-12-23 09:15 0 comments |
 by Flemming FunchHere's another overview of how ants find food, with the thought towards applying that to program algorithms. It is from MUTE, a new file sharing program. They use a similar, but somewhat improved, approach to decide how best to find a certain user in the network. I fail to see exactly why it would be useful or efficient to do it that way, but there's something there, of course. Generally speaking the thought is that we have a network, similar to a neural network, and it learns how best to find stuff based on the success of past attempts to find it. Which is how it should be, of course. But ants only need to signal where there's food and where the home nest is. It is, despite the simple 'random' approach to mapping it, a rather centralized scheme of organization. The ants are not centrally controlled, but it all revolves around bringing food back to the nest and similar activities. We have potentially billions of pieces of information and billions of places to find it. I'm not sure how we can manage that really well without each node keeping a whole record of everything it has been asked for and where it sent it and where it came back from. Kind of like a google database in each node. But maybe I just don't get it yet. |