Edit: to be very clear, it’s an early draft

It’s still an early draft.

You can view it as the proposed successor to IPv6, but really it’s more of an extension to IPv4.

Put simply, IPv6 is beginning to be seen as a failure, so IPv8 looks at why IPv6 adoption is still so poor and proposes something else.

IPv8 is backwards compatible with IPv4 (IPv4 becomes a subset of IPv8) which should help adoption since there’s no grand switch-over day like IPv6 requires.

But it also includes a bunch of other stuff, including the idea that every single network element has an identity. So, like, your router can identify itself in a verifiable manner to your ISP by using a JWT.

This allows some good things, for example right now it’s a flaw in the internet that you often have to simply trust an IP address for important data. Like a router can advertise routes and currently your home network trusts your ISP for routes because somewhere in your router there is a static IP assigned and your home router will just trust that IP…. Stuff like this which leaves a lot of infra open to attack such as spoofing or man in the middle stuff.

In IPv8 your router can now verify the identity of who is actually talking to it. This is good, but the downside is that it makes it much more difficult to be anonymous since traffic is now deeply identifiable.

It also allows for every single network element to be uniquely addressable. Under IPv4, typically there is port mapping so the public internet sees “you” as your ISP, but under IPv8 it’s proposed to make every single network element directly addressable by IP, identity, and DNS8. This also enables cool things like potentially you could have asynchronous communication without needing to keep a socket open and dealing with port exhaustion but it also means you’re much less anonymous than under IPv4.

It’s probably moot anyway since these days your ISP and every server / router in the middle is maintaining detailed access logs anyway so your anonymity is already gone, but it makes it much simpler to deanonymize and potentially even undermines stuff like using a VPN by creating many more vectors by which a VPN can leak identity.

You should look at the reply by Moidialectica@hexbear.net below. It is an unvetted proposal standard to the IETF that should not be taken seriously.