Do you have actual evidence of this? What ASN operates this way?

Why would any CG-NAT split their volume that way?

IPv4 addresses are not that scarce yet, and realistically any CG-NAT will have several IPv4 addresses per metro area, if only to allow for reasonable levels of geolocation (e.g. to not break the "pizza near me" search use case).

Sounds awful, though. Maybe we should get more widespread usage for IPv6 instead.

Yes. I’ll never forgive IETF for standardizing CGNAT back in 2013. They should have just said “no, deploy IPv6 with a transition technology”.

If that had happened, IPv4 would likely already could be regarded as a relic of the past.

The ietf standardization was irrelevant so I would give them some slack. ISPs were using CGNAT already in a widespread fashion. The ietf just said, “if we’re gonna do this shit, at least stay out of the blocks used by private networks”.

Surely IPv6 makes location spoofing harder, you're not identified by just location anymore but uniquely identified down to the device?

This was solved in 2007 with Privacy Extensions.

It has been a non-existent problem for roughly 20 years now. Why do people still keep pulling out "uniquely identified down to the device" as an argument?

Windows, macOS and most Linux distros by default rotate SLAAC addresses every 24 hours.

That is really interesting. I wonder if we have any internal data on this. I will check.

We are trying to work with ISPs everywhere, so if port level geolocation of the IP address is common, we surely need to account for that. I will flag this to the data team. To get the ball rolling, I would love to talk to an ISP operator who operates like this. If you know someone please kindly introduce me to them.