If someone can figure out how to get the ISPs of the world to participate in a routing prefix market, then it might be worth revisiting this idea. Note that there is nothing stopping establishing a routing prefix market now, so it could be done prior to changing address allocation policies.
Selling IP addresses clearly is a bad thing. But selling the right to use a block of n IP addresses could be workable. All the right-to-use certificates would need to be registered with an RIR to validate them in the same way that land ownership is registered in a land registry office. When you connect to the Internet, you buy a right-to-use cert for whatever you need/want. Present this to your ISP and they assign you a block of the size that you bought. If they need more addresses, they simply present a list of customer certs to the RIR and their allocation is expanded. If a 3 month buffer isn't enough, then the ISP pays for a 6 month buffer. The advantage to all of this is that it no longer assumes that everyone is following more-or-less the same business model. In the early days of the net, most people *WERE* following roughly similar business models. But today there is a lot more variety and that variety appears to be increasing. The whole IP address allocation system is creaky and old-fashioned and out of touch with the times. It needs to change and become simpler, more flexible and easier for outsiders to understand. But, on the other hand, with IPv6 looming which changes the playing field perhaps there is room for some fresh approaches. I think that a workable idea would be for a consortium of tier one providers to set a price on global routing table slots. The price would be sufficient to fund an organization to collect route info (like routeviews) and to periodically adjust the prices (yearly) with money left over to fund some research (like Bellcore?). Members of the consortium would pay for the number of routes they announce to other consortium members. The members would also collect money for routes from any peers who are not in the consortium. When an organization connects to the Internet, they must decide between free fate sharing with their ISP or paying for a route in the global routing table. This consortium really should be under the auspices of IANA, ARIN and the other RIRs but it may be necessary to start it independently first. There really should be no antitrust issues here as long as all consortium activity is done publicly and the prices are set in such a way that the global routing table can continue to grow but does not grow faster that technical capabilities and deployment. If this could work today then it can continue to work with IPv6 and any conceivable future networking protocol. --Michael Dillon