* Michael Rogers <m.rogers at cs.ucl.ac.uk> [2007-11-14 11:35:31]: > Matthew Toseland wrote: > >> That means the ref can be around 38 bytes (IP address + port + hash) - > >> small enough to exchange over the phone, IRC, etc. > > > > Not if we want to remain undetectable. We need a key for the outer > > encryption > > wrapper. > > Sorry, I don't follow - do you mean JFK takes place (or should take > place) inside another layer of encryption? How is the key for the outer > layer exchanged? To avoid CPU DoS it has to be a symmetric key, and to > avoid an infinite regress of key agreement protocols it has to be > derived non-interactively from the contents of the refs. Therefore it > can't be secure against an attacker who's seen both the refs, regardless > of how much information they contain - so what's the advantage of long refs?
Who said that people should exchange their references over an insecure medium ? ATM the key is a XOR in between both node's identity iirc We are not debating the strength of the key here... The only purpose of the outter layer encrytion is to avoid cheap pattern matching. NextGen$ -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20071114/e47e76b1/attachment.pgp>
