6 ms·

Also, now that many people's connections are IPv6, the server can identify individual computers behind a router. Though your laptop's IPv6 may not be a permanen

Also, now that many people's connections are IPv6, the server can identify individual computers behind a router. Though your laptop's IPv6 may not be a permanent address, it lasts long enough to track you. If you're logged in to any web services on your computer, other web requests can then be correlated to non-authenticated requests from the same IP. I feel like IPv4 with NAT was an accidental privacy win. I think the solution for this is for operating systems to allow applications to request a fresh IP address, or something like that. Obviously that would be a highly difficult and disruptive change at multiple levels...

theonewolf 6y ago
One could run every tab / sub-process within a container with its own networking stack. It is complex, but doable.
bradfitz 6y ago
Speaking of which, you might be interested in this thread: https://twitter.com/bradfitz/status/1263643749592166400 https://twitter.com/bradfitz/status/1263643749592166400 Which resulted in this Chrome bug: https://bugs.chromium.org/p/chromium/issues/detail?id=1085662 https://bugs.chromium.org/p/chromium/issues/detail?id=108566...
CarelessExpert 6y ago
NAT only anonymizes folks behind that one router, and in a typical household that's down to just a couple of individuals. Combined with other hints (including browser fingerprinting), NAT provides very little privacy protection, especially given how long-lived the average IPv4 DHCP lease lasts. > I think the solution for this is for operating systems to allow applications to request a fresh IP address, or something like that. Obviously that would be a highly difficult and disruptive change at multiple levels... IPv6 already has this capability: https://www.internetsociety.org/blog/2014/12/ipv6-privacy-addresses-provide-protection-against-surveillance-and-tracking/ https://www.internetsociety.org/blog/2014/12/ipv6-privacy-ad... That randomization occurs on a higher frequency than a typical DHCP lease expiration. For a home that has a /48, this is no better or worse than NAT, since you can still only track devices to the home. If your individual devices have their own IPs within the ISP address space, then the best an outsider could do is track devices to a given ISP subnet. Of course, again, the real threat is browser fingerprinting and similar technologies, which are entirely independent of the underlying network.
- etaioinshrdlu 6y ago
  No, I don’t think privacy addresses as currently implemented addresses the problem. All browsers on my machine still share the same ipv6, whereas I think they should get unique ipv6 addresses if they want them. Fingerprinting is more tractable to defeat by disabling all JS, for example. Also, you seem to be saying that because NAT doesn’t provide much privacy, we shouldn’t care if we get even less privacy. I disagree with that.
  - CarelessExpert 6y ago
    > All browsers on my machine still share the same ipv6, whereas I think they should get unique ipv6 addresses if they want them. Yup, sorry, I overlooked that part of your comment. That said, I'm still not convinced it buys you much... maybe a bit of an improvement, but... > Fingerprinting is more tractable to defeat by disabling all JS, for example. Sorry, that's just completely unreasonable. JS is basically table stakes at this point for virtually any website to reasonably function. Worse, it feels like an arms race between browsers and the ad tech giants when it comes to try and defeat fingerprinting. Now, imagine you could break the 1:1 connection between browser engine and user. For example, imagine a bank of cloud hosted browser engines out in the cloud. You have a thin "client" that uses this headless cloud browser using a remote display technique of some kind (just batch up, serialize, transport, and render DOM updates, while passing back user interaction events?). Between sessions the state is wiped, and they're randomly shared among clients of the system... I'm sure there's a zillion legal and technical hurdles, but unless you break the 1:1 relationship between browser and user, fingerprinting feels inevitable. > Also, you seem to be saying that because NAT doesn’t provide much privacy No, I'm just saying IPv6 provides no less. And either way, while it's not nothing it ain't that much.
Spivak 6y ago
Is anything stopping you from using NAT with IPv6? I really see no reason to assign every device on a network their own public address when I can just keep the pool of addresses on my router to be assigned for whatever and be able to assign devices huge swaths of private IPv6 space if I want. Does not using NAT really help anything when the firewall still wont let you in without an existing outbound connection?
fulafel 6y ago
This holds only when with IPv4 NATs that are fronting masses of people, such as in ISP or bigcorp NATs - otherwise the privacy effect is weak enough to be negligible and it can be weaker than IPv6 use of rotating privacy addresses.

theonewolf 6y ago

One could run every tab / sub-process within a container with its own networking stack. It is complex, but doable.

bradfitz 6y ago

Speaking of which, you might be interested in this thread: https://twitter.com/bradfitz/status/1263643749592166400 https://twitter.com/bradfitz/status/1263643749592166400 Which resulted in this Chrome bug: https://bugs.chromium.org/p/chromium/issues/detail?id=1085662 https://bugs.chromium.org/p/chromium/issues/detail?id=108566...

CarelessExpert 6y ago

NAT only anonymizes folks behind that one router, and in a typical household that's down to just a couple of individuals. Combined with other hints (including browser fingerprinting), NAT provides very little privacy protection, especially given how long-lived the average IPv4 DHCP lease lasts. > I think the solution for this is for operating systems to allow applications to request a fresh IP address, or something like that. Obviously that would be a highly difficult and disruptive change at multiple levels... IPv6 already has this capability: https://www.internetsociety.org/blog/2014/12/ipv6-privacy-addresses-provide-protection-against-surveillance-and-tracking/ https://www.internetsociety.org/blog/2014/12/ipv6-privacy-ad... That randomization occurs on a higher frequency than a typical DHCP lease expiration. For a home that has a /48, this is no better or worse than NAT, since you can still only track devices to the home. If your individual devices have their own IPs within the ISP address space, then the best an outsider could do is track devices to a given ISP subnet. Of course, again, the real threat is browser fingerprinting and similar technologies, which are entirely independent of the underlying network.

etaioinshrdlu 6y ago
No, I don’t think privacy addresses as currently implemented addresses the problem. All browsers on my machine still share the same ipv6, whereas I think they should get unique ipv6 addresses if they want them. Fingerprinting is more tractable to defeat by disabling all JS, for example. Also, you seem to be saying that because NAT doesn’t provide much privacy, we shouldn’t care if we get even less privacy. I disagree with that.
- CarelessExpert 6y ago
  > All browsers on my machine still share the same ipv6, whereas I think they should get unique ipv6 addresses if they want them. Yup, sorry, I overlooked that part of your comment. That said, I'm still not convinced it buys you much... maybe a bit of an improvement, but... > Fingerprinting is more tractable to defeat by disabling all JS, for example. Sorry, that's just completely unreasonable. JS is basically table stakes at this point for virtually any website to reasonably function. Worse, it feels like an arms race between browsers and the ad tech giants when it comes to try and defeat fingerprinting. Now, imagine you could break the 1:1 connection between browser engine and user. For example, imagine a bank of cloud hosted browser engines out in the cloud. You have a thin "client" that uses this headless cloud browser using a remote display technique of some kind (just batch up, serialize, transport, and render DOM updates, while passing back user interaction events?). Between sessions the state is wiped, and they're randomly shared among clients of the system... I'm sure there's a zillion legal and technical hurdles, but unless you break the 1:1 relationship between browser and user, fingerprinting feels inevitable. > Also, you seem to be saying that because NAT doesn’t provide much privacy No, I'm just saying IPv6 provides no less. And either way, while it's not nothing it ain't that much.

Spivak 6y ago

Is anything stopping you from using NAT with IPv6? I really see no reason to assign every device on a network their own public address when I can just keep the pool of addresses on my router to be assigned for whatever and be able to assign devices huge swaths of private IPv6 space if I want. Does not using NAT really help anything when the firewall still wont let you in without an existing outbound connection?

fulafel 6y ago

This holds only when with IPv4 NATs that are fronting masses of people, such as in ISP or bigcorp NATs - otherwise the privacy effect is weak enough to be negligible and it can be weaker than IPv6 use of rotating privacy addresses.