by zygote1 » 29 Apr 2011 18:59
The problem with ISP NAT isn't just technical. It's also that it gives the ISPs just the kind of excuse they need to violate neutrality against most kinds of non-HTTP traffic and claim it's for legitimate network management reasons. Not to mention better enforce their intentional price discrimination between "consumer" and "business" connections, the main difference with the latter being that the terms of service allow operating your own servers. So the fancier NAT features you describe may well be enabled for a particular customer only if they pay he extra fees for a "business" account.
Of course, the real solution to all of this is better competition. As long as most customers get their internet through a monopoly or a small oligopoly of big telco/cable/cellco providers, they'll engage in customer-hostile behavior and enact customer-hostile policies which technological advance will only make it easier for them to enforce against even the tech-savvier consumers.
Alternatively, if a VPN service of some kind became big, popular, and cheap enough, it could force ISPs to accept their proper roles as dumb pipes by making it literally impossible for them to know what was in customers' traffic or even where it was going and coming from. (If it didn't become popular enough, they could block encrypted VPN traffic for non-business-account-holders and get away with it, of course, and if VPN services remained difficult to use and expensive the customer wouldn't save anything by using them instead of, say, swallowing their telco's long distance fees instead of using VPN-tunneled VOIP, or whatever.)
Achieving that requires two things: making a VPN service with such low costs that it's very cheap or preferably even free without being unprofitable, and is easy to set up and has no geographic or other use restrictions (regarding both user location and remote connection endpoints). Free is possible, if supported by e.g. NXDOMAIN redirection ads and other similar methods (and they'll have clickstream data too).
The second thing is getting lots of people to use it. A big privacy scare (somewhat ironically, since the VPN provider will have access to the clickstream endpoint knowledge that ISPs currently do -- but not the contents) could perhaps accomplish that; or a popular enough application being deliberately degraded by enough ISPs (VOIP being a leading candidate there).