@whispersystems re: contact discovery, would hashing work if preimage space were larger? Change phone #'s to qr codes or bitcoin address?
-
-
-
@Wolverineks The idea is to use an identifier already in a contact card. Otherwise discovery would have to be manual anyway. - 1 more reply
New conversation -
-
-
@whispersystems@moxie crypto-capability-based discovery? (i.e. Alice gives Bob an *encrypted* capability to access Carol) -
@bascule We'd like discovery to be instantaneous. Install an app and immediately know who you can communicate with. -
@whispersystems aah, I see
End of conversation
New conversation -
-
-
Use an exorbitantly-expensive hashing method. Require rehashing with increased complexity as per Moore’s law.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Hash pairs of contact’s identifiers + one’s own identifiers. Matches are required to have uploaded hash of same pair.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Client sends systematically-redacted contact identifiers, server returns bloom filter subset with only potential matches.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@whispersystems Could a bloom filter be used but updated infrequently & distributed via bittorrent (or similar) method under the hood?Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@whispersystems@moxie don't know if this fits your adversarial model but we have a PSI proposal for large sets: http://research.microsoft.com/en-us/um/people/senyk/pubs/sapsi.pdf … -
@senykam Thanks, but I don't think so. In our case P1 and S are both honest but curious and willing to collude. -
@moxie that's a tough one...I haven't read it yet & you might have seen it but this newish paper claims to scale PSI: http://eprint.iacr.org/2013/515.pdf -
@senykam An interesting idea, but the network overhead of OTing the intersection of two bloom filters is 2λ * m. Just m is too much for us. -
@moxie too bad. Anyway thx, it's very useful to get details about adv model and params of interest for diff deployments. Will think about it
End of conversation
New conversation -
-
-
[good piece] RT
@whispersystems: The difficulty of private contact discovery. An unsolved problem:https://whispersystems.org/blog/contact-discovery/ …Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@whispersystems Nestor Makhno on redphone contacts, nice touch#tickThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@whispersystems@moxie Why not anonymize contact queries by sending them through Tor? -
@whispersystems@moxie Of course, it would also be nice to send TextSecure messages through Tor by default.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.