That's unfortunate. I never like hearing about free market solutions to real-life problems that are blocked through misguided attempts to increase "fairness." This only benefits the large ISPs who already have both large blocks of IP addresses and a legal mechanism for leasing them (leasing is effectively a form of reselling).

Still, those ISPs can start offering cheaper plans to those willing to take a NAT'd IP address (read: charging more if you want a raw IP. This is already happening in the commercial space). The logic still works. Those who really need IP addresses will be able to pay to get them. And those who don't will work with improved NAT and related technologies.

In 734 days, you will be able to get an IPv4 address if you really want one. Still, as I said, I like IPv6. Who wants to pay a premium when the "scarcity" is artificially created by a limited number of bits?

I'm not sure what the rules are on reselling IP-addresses (is it up to the individual IPv4 registries?), but even finite resources never truly run out, they just get more expensive over time (see Hotelling's rule). With a liquid-enough market in IP addresses, it get's even better (Hotelling's rule assumes the resource is used up, like oil, not reusable, like IP addresses). As the price of IP addresses goes up, more and more work will be put into NAT or similar workarounds (like how HTTP 1.1 introduced the host header), as those efforts will suddenly become cost effective. People who really need raw IP addresses will always be able to get them, just for a price. It is kinda similar to oil in that plotting current trends is always going to be misleading, as that will overlook the effect of future innovations. I actually like IPv6. I just highly doubt the dire predictions about what will happen to IPv4 734 days fraom now.

I couldn't disagree more. Experts/educators in all disciplines are susceptible to falling into the trap of wanting to start with the fundamentals because that's what their (advanced) understanding is based on. It's that kind of thinking that led to the disastrous "new math" where set theory was taught before arithmetic. I even knew a physicist who wanted quantum mechanics taught in first year college courses as a build-up to Newtonian mechanics as a limiting case. This is wrong because we learn by doing, not by starting with theoretical fundamentals. We move on to that later to back-fill our knowledge. Starting with the theory sounds good, but name me one serious programmer (coder/developer/engineer/architect) who didn't first learn to program by tinkering, possibly using "Hello World" as a starting point.