On Sun, 27 Oct 2013 23:37:39 -0700 "Christian Huitema" <[email protected]> wrote:
> Abstract > > Traffic analysis is used by various entities to derive "meta data" > about Internet communications, such as who communicates with whom or > what, and when. We analyze how meta-data can be extracted by > monitoring IP headers, DNS traffic, and clear-text headers of > commonly used protocols. We then propose a series of actions that > would make traffic analysis more difficult. > > Available for now at: > http://huitema.net/papers/draft-huitema-perpass-analthreat-00.txt This is the best summary of the metadata problem I have seen. I agree with this recommendation: "Use encryption. In particular, never send a user identity in clear text." After that...problems. Obfuscating sources may provide less metadata to spies, but it is also data that system administrators use every single day to protect their systems from viruses, spammers, and hackers. Remove too much of that and you have simply exchanged one problem for another, larger one. There needs to be a section for all these proposals that deals with the practical real world barriers to implementing the proposed solutions. That way the proposals that might actually be possible to implement should be more identifiable. The problem with the proposals I see for fixing the metadata problem through technical means is that in the end, to actually be effective, they all seem to boil down to 'We had to destroy the internet, in order to save it'. I think there are small technical changes around the edges that can help, but I really see the solutions for the metadata problem as more political and social than technical. Concentrating on making encryption really, really easy to use would go a lot further at this time than messing with deep changes, because people are not even using what is already available. -MD _______________________________________________ perpass mailing list [email protected] https://www.ietf.org/mailman/listinfo/perpass
