High-speed anonymising network proposed. Research paper PDF download: arxiv.org/pdf/1507.05724v1.pdf (466.2 KB). -- Tom
It sounds amazing. But I haven't read the paper yet. If it is truly amazing, I wonder what chance it has, and how many years it might take. For better or worse, Tor has a fairly solid lock on the anonymity userbase. Except for the haters, of course
They're both anonymity networks. But Tor works at application layer, whereas I gather that Hornet would work at network layer. Connections in Tor are roughly analogous to HTTPS in web browsing, plus SOCKS5 proxy capability. I gather that connections in Hornet would be roughly analogous to UDP-based VPN connections.
It's network layer onion routing. But that means that it can operate stateless, that is to say, it doesn't need to maintain node-node TCP connections (which limit scalability for example). Tor is effectively an overlay on the IP internet. The other factor being that the latency is reduced. My personal feeling is that low-latency anonymity is very hard to do, and in particular, the paper excludes highly resourced adversaries who can do traffic correlation across the network. The only defense against that is a highly used network, with traffic padding.