Ubiquitous Data Flow
The best solution for surfing in privacy, Tor, or The Onion Router, suffers from some weaknesses that allow attackers that monitor the boundaries, entry and exit nodes, to track users by observing incoming and outgoing traffic. Time, sequence and size of network packets are simple means to uniquely identify network transfers and thus users, and can quickly obliterate their privacy. Unfortunately this problem won’t be solved as long as every click translates to a network request.
I propose a Tor-like but radically new system, using a Ubiquoitous Data Flow model. End users connect to a random and changing set of nodes and permanently stream data to all of them, sending payload and (lots of) padding noise, both of which appears the same random stream of data to eavesdroppers. With this approach the most vulnerable point on privacy, the ISP, won’t have any more identifying metadata, for they won’t be able to identify the the exact time of user interactions, the network requests and responses.
Payload then travels between relay nodes in a similarly permanent wall-of-noise manner until reaching an exit node. As with Tor, there’s no way around having individuals and organisations operating these, with the huge difference though, that once the exit node feeds the network, the data goes up in the all-permeating constant stream of data, or, for the observer, noise. Having no way to link the metadata of the exit node with the metadata of entry nodes, there would be sense in operating exit nodes for tracking.
Yes, this would mean a lot of waste in internet traffic but if we can burn Gigawatts to mine Bitcoin, we surely should be able to invest something to prevent 1984 from happening. And yes, personally identifying information at the visited website and all the ad- and track- and spyware installed on them would still have to be thought of and mitigated. However, at least the network, the easiest centralised access to our entire online communications, to all our data, would be safe with this approach.
Disclaimer: of course this will only work until Cryptogeddon, that is until quantum computers crack all current means of encryption, but that’s a tabula rasa kind of event anyways.
Note: this approach takes a page from Steganography, the principle of hiding information in a massive haystack of irrelevant data, because that is something we’ll have more and more and ever more, so we should be able to hide our payloads in that overwhelming wall of noise.