> > From: "Scott G. Miller" <[EMAIL PROTECTED]> Date: Mon, 22 Jul 2002 20:37:29 -0700
>No, I don't think its necessary except in very small networks (which of course >might characterize the current network). > Why do you think that it is important to have multiple specialties in a small (or immature) network? I ask to seek a better understanding of network development for myself. I personally think that the network could be successful either way. However, I think that key specialization is a primary factor in so many aspects of how Freenet operates so it should be well understood and perhaps used to mold network traffic into a more free-flowing stream then it is. *If* it does not matter if nodes have single specialties or multiple ones, wouldn't it make sense to chose a simpler, more easily understandable model to build around? >Theres nothing that should prevent this currently, in fact, my node >has two distinct peaks. > I am not sure what you mean by "this". I know that Freenet does little to manipulate key specialization and that nodes exist now with any number of specializations. What I was getting at was that I don't see multiple specialties being a requirement for anonymity. I invite someone more knowledgeable to refute this. So if multiple specialties are not needed for proper routing or anonymity and complicate the network and make it harder to understand, why not reduce the number of key space references per node to one? >Separating these is very very bad. If the network thinks your good >for something and you think its something else, the node will constantly >be fighting the network. > Sorry ... by separation I do not mean decoupling. What I mean is the concepts can be thought of as separate entities for the purposes of understanding their utility. Looking at third party (incoming) specialization: Currently, if a node wants to, it can keep track of what keys others are asking it for. With such info it can do a few things: It can selectively reply to requests closer to the keyspace that it wants to occupy and therefore, manipulate it's own keyspace. It is pretty useless to do so due to the nature of Freenet keys (i.e. like keys do not imply like content). But it is possible. It can see how mature its reputation is with the rest of the neighbouring nodes. If all the requests that it receives stay in the same specialties over time then it has a good idea that the Freenet around it has stabilized. If they are all over the map then it implies that either the node is young or it is surrounded by young nodes. How close an incoming request is to a nodes data store specialization gives a means (independent of HTL) of how close the request is to being fulfilled. If the incoming request is in a less dense region of the nodes key space histogram, then it is likely that the request was recently started ... regardless of what the HTL is. It could also imply a request from a young node. I'll leave discussion of what one can do with first party (outgoing) key specialization for another time since this message is getting far, far too long and nobody is likely still reading at this point ;] Michael _______________________________________________ devl mailing list [EMAIL PROTECTED] http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl
