1. 7

Interesting way to look at privacy online. It’s difficult to fight data collection, so making it less meaningful could be a good effort.

On the other hand we see lots of innovation in various ML techniques so I wonder if usage patterns could soon be retrieved from all digital noise?

  1.  

  2. 1

    For a slightly silly backend version, see Squawk, which a few of us dreamed up in a bar at LCA 2016 …

    1. 1

      What I do for this is running OONI which has a similar effect I think. They btw. also have some random stuff.

      https://ooni.torproject.org/

      1. 1

        It would cost them more, and you can also improve your noise techniques using an adversarial AI. Techniques that are used to trick AI with image detection should also apply generally in finding patterns in signal noise. People say obfuscation isn’t true security, and they’re right but if you do it very well it can still be valuable in situations where real security is infeasible. I wouldn’t treat this as perfect protection, but just adding noise might make it hard to find relationships in the data.

        1. 1

          See also: TrackMeNot, which does fake searches. I’ll be honest, I was hoping to play around with using the techniques of TrackMeNot for noise, too, but I’m not, at all sure, how to measure it’s effectiveness, unfortunately.