What exactly are the "engineering benefits" beyond a larger address space? Most companies rely on private address spaces to logically separate them, and will ultimately end up using NAT66, ending up exactly where they started, but with significantly more complexity.

Every time this comes up, people come out of the woodwork to say, "Well, akually we talked about this exact thing in 1995 and decided this was the right way!" Yet, it's never convincing.

If it were a simple address expansion, we would have completed this upgrade around 1998. Yet here we are, 30 years later!

> Most companies rely on private address spaces to logically separate them...

Which is a hack driven by address scarcity. You can, and should, separate address spaces just fine if they're publicly routable.

> ...and will ultimately end up using NAT66, ending up exactly where they started, but with significantly more complexity.

Then those companies are managing their network very poorly. Which is their choice, but not really an argument against IPv6.

Telling people doing things like SCADA, who absolutely don't want their hot forging press to be globally addressable, that they're "doing it wrong!" is not helpful. This attitude is why nothing short of coercion and blackmail is required to get a large chunk of users to switch.

Literally no one is suggesting this. Having globally unique IPs doesn’t even slightly infer that they should be globally reachable. IPv6 still uses firewalls. It does mean that making an internal resource reachable when appropriate for that specific resource is vastly easier: you open the right firewall port and it’s done.