Require payment of $1 per year per public IP, and suddenly wastrels like the Apple class A will be available again...
Recovering an /8 would take staggeringly more effort than it is worth in terms of how little it fills existing demand.
There is no scheme that will make IPv4 work in the long-term. The mathematics are very simple. You cannot fill demand for exponentially more than 4 billion numbers with a pool of only 4 billion numbers.
There's a big, huge, insurmountable difference between home NAT traversal, with port-forwarding, and Carrier-grade NAT traversal, where fuck you and your ports.
But we're already leaving behind the case where a single NAT near the edge of the network graph (the "home") is sufficient and are starting to see larger deployments of "carrier-grade NAT". So far as I know there is no equivalent kludge to UPnP Port Forwarding that works reliably on that scale, and building such a kludge seems like a major question of what sort of internet topology we really want. For hyperbole's sake, "carrier-grade NAT" and other similar cases of NATs speaking to NATs that what we're talking about is ultimately a balkanization of the internet, breaking the internet back into sub-networks of various addressability.
Even with economic incentives, we'd still be running out of IPv4 addresses for all of the "physical places" in the world that have 65000 devices. More critically, with IPv4 we've already hit scaling problems with the size of routing tables for the physical places that connect to the internet, we can't easily do "per-IP routing" and there aren't economic incentives that will quickly change that.
Ultimately, I'm wondering if this is the confusion you are having: IP (v4 or v6) is about the physical routing to the places of the internet. IPv6 doesn't have answers to device connectivity or movement because that's not its job. Logical Device to Physical IP Address mapping is the job of internet layers like DNS and mDNS/Bonjour. NAT is a series of hacks designed to deal with the fact that IPv4 doesn't do as great of a job at physical routing as we would like, now that we have millions of devices connecting to it every day. IPv6 is an actual attempt to fix how the internet does physical routing. It does solve the real problem we are actually having: physical routing on the internet. It solves it much better than NATs wish they could solve it, regardless of the "device vendors and academics" that "ran the show".
There's only Port Control Protocol, but almost no sane ISP will ever enable it, because it can't really solve much and the security implications make people really uncomfortable.
The big problem with carrier grade NAT is when you stick N people behind 1 IP, they can only have at most 65,536/N ports each. If you're a big city in India, say, and N=1000, that's 65 ports per person.
Your ports-per-person is your ceiling on maximum simultaneous TCP sessions per person. 65 ports is barely enough for a few browser tabs to be open simultaneously, and god help you if you've got another layer of NAT at the household level (which you probably do) and you're splitting those 65 ports between 2 or 3 other people (which you probably are).
Consequently, browsing the internet is a fraught process in a lot of places with carrier grade NAT. You can absolutely forget about P2P or anything else complex and port-hungry.
Multiplayer gaming, peer-to-peer anything really, is an unmitigated shitshow under heavy NAT.
You definitely do not want carrier grade NAT to become entrenched in North America, like it already is in Asia.
And at some point, even if we did want it, we'd run out of numbers for people who want to serve content, and at that point NAT isn't going to do anything for you, and the barrier for entry for new web businesses becomes sky high.
The faster we get to single stack ipv6 only, the better.
If you're an ISP, you could probably put multiple nat64 boxes in the core so that all your customers benefit from it and can gradually switch ipv4 off (since they can transparently reach ipv4 endpoints through the nat64 gateway).
Then there are a few applications that don't handle ipv6 properly (i.e. skype), and Apple's decision of mandating ipv6 support is meant to fix just that.
Then there is 4XLAT, ds-lite and others which allow for a single stack ipv6 network and carry ipv4 on top of it to the endpoints which need it.
The US Government owns:
6.0.0.0/8
7.0.0.0/8
11.0.0.0/8
21.0.0.0/8
22.0.0.0/8
26.0.0.0/8
28.0.0.0/8
29.0.0.0/8
30.0.0.0/8
33.0.0.0/8
55.0.0.0/8
56.0.0.0/8
214.0.0.0/8
215.0.0.0/8
If you're large enough to get allocations that huge, you're large enough to afford it.
Why is Apple a "wastrel" when it's building gigantic datacenters all over the place, yet MIT, who has an equal sized allocation for a university, is somehow fine?
It's not in Apple's interest to yield the block unless there's industry consensus and other companies and organizations with similar allocations are doing the same.
The real problem is that this doesn't make sense: we don't WANT IPv4 to persist. The sooner we all go to IPv6, the better.
Ideally, usage will gradually wane until only obsolete items even use IPv4, then a cutoff date will come from major ISPs and such cutting off IPv4 service and we'll finally pass the last hurdle.
At least, one might hope so.
And the holders didn't get the /8 in modern times "because they are huge" they got them in ancient times because historical reasons.
The fact that we've treated those allocations as sacrosanct means that we're stupid political people, not engineers. (What else is new? :-)
They could enforce renting (vs owning) going forward with all new assignments, but that will probably just encourage even more hoarding of existing space.
More to the point, the IPs were assigned to Apple by some agreement more or less akin to a perpetual lease. Under what conditions can that lease be terminated? (Actually some of the earliest allocation were apparently because "Postel says so". But probably still close enough to a handshake deal to be valid.)
> Actually some of the earliest allocation were
> apparently because "Postel says so"
IANA and ARIN, historically. You don't ever own your IPs, no more than you own your phone number.
I own legacy IPs. And I pay nothing, because I did not sign the ARIN registration agreement. If they could force me to pay by threatening to revoke my IPs, they would have done it already.
You may also want to look into concepts such as water rights. Nothing stops you from draining all the water out of the Colorado river upstream. Except the things that do...
http://www.finnegan.com/resources/articles/articlesdetail.as...