On 4/15/06, Gregory Maxwell <gmaxwell at gmail.com> wrote: > Lets say I run a node, I don't care much about my own privacy... Thus > I peer fairly promiscuously, connecting with people who I hardly know > as long as they don't look too much like an EvilForce(tm). > > I do, however, have some friends who care more about anonymity than > me. They trust me not to operate a subverted node.. and perhaps they > have a few other friends like me... well connected because we're peer > sluts but personally trustable. > > My friends might have some more friends who care even more about > anonymity. They want to be an extra hop away from the wildly connected > portion of the graph. > > As it stands right now, my friends (and their friends) can't use > freenet.... They can't use it because they can't stay connected. As > my client uses it's capacity for it's zillion and one promiscuous > peerings, the chances that at any given time my friends will be backed > off is really good. > > Since my friend has a higher standard for who he'll peer with, he > doesn't have many peers. Thus the chances that he'll spend most of his > time partitioned from the network are pretty good. > > This could be drastically improved if I could set a preference on > peers which I wish to reserve capacity to keep my connections up > with... such information would be kept private to me but it would > enable my less connected friend nodes to stay strongly connected, > while I'm able to have widespread connectivity which makes my node > valuable to him.
That sounds like a quite bad idea if the small world theory and our routing algorithm works. At least according to my understanding of it.
