> It's impossible to fully implement privacy on the Internet due to the way devices communicate with each other.<p>If we were designing things from scratch, it would be simple to have much higher level of privacy than we have online now -- probably even higher levels than real-life.<p>On network level, this means dynamic MAC and IP addresses, and multiple NATs. Perhaps even require non-logging proxies on customer's routers? (this way one can examine source code of such proxies and ensure that nothing untoward is happening). For browsers, no user-agent/cookies/javascript/etc... without explicit permission.<p>The reason we cannot do that is all the bad actors -- the risk/reward ratio for illegal things on the internet is way too low. In the real life, a robber pulling on every doorhandle to see if the door is unlocked will be caught very quickly -- but in the internet it is normal and happens all the time. Open SMTP servers are used for spam, open proxies are used for DDOS / password guessing, cookie-less requests may mean robot or DOS agent.<p>(There is an additional issue with browsers sending too much information to website -- it is a mix of people designing features and not thinking of security implications, and inputs from big corporations who get their income from tracking users. But this can be fixed much easier that network-level stuff.)<p>So the loss of privacy is not caused by technical reasons, it is all caused by humans acting like humans do.