That's a good point Dave, but I like their approach better than their conclusions. They're attacking the problem by looking only at the network protocol rather than the users on the network. A trust system can block spam, leeches and spoofers, but it operates on a plane above the basic message routing system of the network which can still be misused. Remember that users and nodes are not the same thing; it's easy for users to spot malicious users but it's hard for nodes to recognize malicious nodes. These guys are working below the surface and not many others are.
I would add a usage restriction to your voting system: if a user has a bad reputation then they should not be allowed to run a supernode. That might bridge the social and technological gap in network security allowing the client to focus on networking rather than bot-sniffing.
|