Matthew Toseland wrote: > We should in the near future implement reliably exposing our peers' locations > to our peers, and use it to implement using our peers' peers' locations for > better routing. According to Oskar and Vive, who refer to published papers, > this should significantly improve performance relative to simple greedy > routing. I believe the correct algorithm is to simply route to the node which > either has the closest location to the target, or whose best peer is closest > to the target.
One issue to consider: what happens when nodes go offline? There's no point routing to my peer based on his peer's location if his peer is offline, so how often do we need to update this information and what are the traffic and anonymity implications of doing so? Just a thought, Michael
