Although ARIN was only the 4th RIR to run out of v4, there are policy differences between the various registries such that ARIN is the first to actually hit free pool 0. APNIC was the first to stop making regular allocations in 2011, and is currently projected to hit total 0 around 2018:
AfriNIC is the only RIR still making ordinary v4 allocations, but it only has enough v4 for 0.1 addresses/per capita; whereas ARIN was around 5. Everyone other than AfriNIC is on some kind of exhaustion rationing policy. If you want new IPv4 in North America or the Caribbean, your new choices are go on the waiting list and pray that IANA reclaims some space from someone somewhere somewhen, or buy it on the transfer market.
Expect a spike in v4 transfer pricing; last year they were running in the $8-$12 dollar range, but that is going to go up rapidly. It's estimated that the customer support trauma and equipment expense of running NAT44 CGN at ISP's makes v4 uneconomic past a price point around $40/IP, so expect a complete collapse of IPv4 transfer market prices 4-6 years hence when v6 laggards belatedly complete their externally facing protocol migrations. This week, my best guess as to when the Tier-1 backbone providers stop routing IPv4 (making the remaining trickle of legacy v4 use tunnel over native v6) is 2022. I'm still projecting the last internal use of IPv4 not to die until 2036, giving v4 a remarkable 50+ year run.
ARIN isn't at total v4 zero yet; there is a reserved /10 for making tiny allocations to new internet exchanges, or NAT64 space to accompany new IPv6 blocks. But no one will ever see a new v4 /8, /10, or even /16 from ARIN henceforth. Even a /22 is becoming a stretch.
Just getting up to speed with IPv6 and I really hope you are right about us getting serious about it although I stiill have my doubts.
I have to say, much to my surprise, it is remarkably straightforward to understand compared to what I was expecting (or at least it has been up to now !).
I think the hardest thing is to get out of thinking in IPv4 terms where we are forever matching subnet size to number of clients with all the extra complexity that brings together with the scarcity of public address space.
Once you can adjust your thinking to accept the size of a /64 subnet compared with the number of clients in that IP subnet everything else becomes a lot easier :-)