On Tuesday 29 January 2008 22:55, Michael Rogers wrote: > Matthew Toseland wrote: > > The node at the end should be close to optimal even if we've redirected near > > the beginning, right? > > I've probably misunderstood how destination sampling works - is it the > case, as in Oskar and Ian's paper, that every node along the path > creates a shortcut to the destination with some small probability, or is > it only the originator that does so?
Any node along the path that needs a node can connect to the destination, in which case the originator (or a node further along) might connect to the node in the middle. > > I was thinking that destination sampling might be harmed if the request > took a strange path, even if it ended up in the right place... but to be > honest I don't understand it well enough to speculate so I'll shut up. :-) Hmm, it's possible. But on a network with good topology, the path wouldn't be far off optimal - unless the data was somewhere completely wrong, which would only happen if we'd had a *lot* of requests for that key and even then, is it a good idea to do effectively a global search? > > Cheers, > Michael -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20080130/c4fc12cf/attachment.pgp>
