> -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of John Nilsson > Sent: Thursday, November 01, 2007 11:13 AM > To: [email protected] > Subject: [p2p-hackers] The PIrate Bay, BitTorrent and p2p-hackers > > Hi, > > Are the dwellers of this list involved in (or otherwise have > opinions about) the Pirate Bays attempt to engineer a BitTorrent > replacement?[1]
Interesting. Some random points - * Using per-user public/private keys for signing published content sounds like a good idea. This should allow building and sharing lists of "trusted" publishers, similar to how OpenSSH maintains the list of "known_keys". * Adding encryption to obfuscate the traffic and protect it from in-flight analysis by ISPs will only result in ISPs throttling ALL encrypted streams. These are VERY easy to detect through some trivial statistical analysis. * Using encryption to filter out "unwanted" observing 3rd parties requires authentication. And authentication requires proper trust model, so in the end they will end up with a form of a private, invite-only p2p model. * OTOH if the goal is to impede the identification of the parties involved in a swarm, then the only option is a multi-step inter- overlay relaying. This comes at a price of a substantial bandwidth overhead and it won't probably be practical for BitTorrent-like usage. * What would be a very good feature (outside of the "security" area) is a custom transfer protocol which is a reliable, congestion-aware delivery mechanism. Sort of like SACK'd TCP w/o in-order delivery and with TCP/Vegas like congestion detection. Which is what FastTCP appears to be. To explain - if I am copying one gig of data, it doesn't really matter in which order its parts arrive for as long as I receive them all. * I really hope this design effort won't go the way of Okopipi, a "community-designed" Blue Frog replacement. Alex _______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
