Ipv6 defined "privacy extensions" to combat this very thing - and most operating systems use it by default. The idea is that devices generate random addresses instead of just using the device Mac address. I can't say how well this defeats tracking, however, this article (https://www.internetsociety.org/blog/2014/12/ipv6-privacy-ad...) concludes

> ... the ability of security services to track you on IPv4 versus IPv6 is pretty much about the same.

> It would be trivial in that circumstance to blow away any kind of NAT and the pseudo-anonymity/plausible deniabililty it provides and make client devices performing illegitimate activity directly identifiable.

Identifying the device and identifying the person using it are two entirely different things. Multiple people use the same device all the time.

Big ISPs don't want to become copyright enforcers; it costs them a lot of money already and enabling IPv6 would cost them even more. (ISPs that could cheaply deploy IPv6 have already done it, so the ones that haven't generally can't afford to.) They're greedy but they're not proactively evil.

I don't agree. Keep in mind who owns the big ISPs and what corporate empires they're apart of. Several of the biggest ISPs (especially the cable ones) are part of giant media conglomerates - their corporate parents are also the parents of some of the biggest copyright owners out there.

Ex: https://en.wikipedia.org/wiki/List_of_assets_owned_by_NBCUni... (Comcast)

https://en.wikipedia.org/wiki/AT%26T#Corporate_structure (AT&T)

How exactly would this work? It's my understanding the IP addresses have to be allocated in a somewhat uniform manner to make routing feasible so it looks like it would require you to give a block of IPv6 addresses to the customer's router for it to allocate to the attached PCs, phones and what not, which seems like it would take us back to chewing through the address space pretty quick and also not necessarily achieve the desired effect unless the router keeps logs of the assignments it makes.

But if every device has a fixed or randomized identifier then surely the identifier contains absolutely no routing information whatsoever (the machine might have an address starting ffff:ffff:... but might be behind routers eeee:eeee:... or dddd:dddd:... in completely different parts of the world).

If ISPs gave every customer as many IPv6 addresses as are in the entire IPv4 space to distribute among their personal devices, and "every customer" amounted to the entire population of the world, that would use roughly 1 / (10^19) of the total space.

https://en.wikipedia.org/wiki/IPv6

That puts it into perspective, thanks.

> It would be trivial in that circumstance to blow away any kind of NAT

Why do you think so? NAT can be used for IPv6 is exactly the same way it's used for IPv4.

AFAIK no home router you can buy does NAT66, so in practice it's not used.

IPv6 will _not_ have NAT be the default case... IPv4 NAT is the default for the vast majority of [home] routers, and offers a bit of privacy protection built-in.

This is true, but if the op meant what they wrote, then the default is irrelevant. If you keep a NAT, moving to IPv6 doesn't blow anything away.

I would expect the home router to provide some level of device obfuscation in this case, specifically to help protect user privacy. I don’t its a fight that can be outright “won” but certainly you don’t just hand over unique device identifiers if at all possible.

I truly think that we are losing privacy when we give up NAT.

I really don't think it matters too much for client devices anyways, we've found plenty of workarounds by now where it matters. Maybe the best case is NAT-optional.