On Sun, Jan 15, 2012 at 10:39 PM, Michael Grube <michael.gr...@gmail.com> wrote: > On Sun, Jan 15, 2012 at 10:09 PM, Evan Daniel <eva...@gmail.com> wrote: >> >> Submitting a response to the Pitch Black paper seems a bit premature, >> given that in the real world we probably have network distribution >> problems even without an active adversary. > > > Could you be more specific? Are you talking about the clustering that occurs > on its own? > > The paper I am writing simulates proposed solutions to the pitch black > attack and measures their effectiveness. Assuming I can get that done in > short order, I will begin looking for better approaches if they can be > improved upon.
It's been too long since I looked at this to give as complete an answer as I'd like. I can dive back into it at some point if there's interest. One big thing: there's anecdotal evidence that (opennet) link length distributions show very different patterns for nodes doing a lot of downloading than for nodes that aren't. Specifically, nodes doing downloads have a more uniform distribution, not a 1/d distribution. I don't think there's been any systematic investigation of whether this occurs, or how big a problem with routing the resultant networks have in simulation. If someone is seriously interested in this, I think my periodic network stats scripts probably have enough information to tackle it. Send me an email and I'll get you raw data to play with. (I don't realistically have time anytime soon.) (Yes, I realize that's an opennet problem and Pitch Black is a darknet problem. My response is that the opennet problems should be a lot easier to investigate, given that we have a large live network, and we haven't even bothered with that much.) Evan _______________________________________________ Devl mailing list Devl@freenetproject.org http://freenetproject.org/cgi-bin/mailman/listinfo/devl