Oddly enough, someone proposed something very much along these lines
at a couple of RIR meetings (see "IPv4 Soft Landing"), and in fact
used the 'driving into a brick wall' analogy. Many of the folks who
commented on that policy proposal felt it was inappropriate for RIRs
to dictate business models (that is, if an ISP doesn't want to move to
IPv6, it wouldn't be 'right' for an RIR to force them to). The
proposer eventually gave up as the impedance mismatch between reality
and the RIR policy making process became too great to observe without
breaking into uncontrollable giggles.
A more interesting experiment:
We want uptake of IPv6, right?
Allocating even fairly large swaths of IPv6 to those who didn't really
need it would be less harmful than hoarding IPv4, right?
How about actually providing an incentive to return IPv4 space? How
about actually providing an incentive to provide IPv6 services along
For example, here, we're not currently doing production IPv6, because
we're not likely to be able to justify the cost of acquiring space from
ARIN. Our legacy IPv4 resources cost us nothing, both what we advertise
and what we don't.
If there was a way for us to trade in some swamp for IPv6, we might
be tempted to do that, which would encourage IPv6 a little more. It
would have to be on the same or similar terms as what we currently
enjoy, otherwise, it makes more sense just to retain the IPv4.
Further, there may be organizations that could be tempted into
returning paid ARIN allocations, perhaps by offering them a guaranteed
low rate (free, ideally) for IPv6 space in exchange for significant
chunks of IPv4 returned.
Now, really, would this be successful? Who knows. But I do know that
it wouldn't be costly in any meaningful way. If the RIRs get any
returned IPv4 space and hand out some free IPv6 space, "we" (the whole
Internet) win on both fronts. Maybe the RIR isn't making oodles of
money from registration services for that space, but then again, I've
never been convinced that the pay-for-addresses model is a good idea
in the greater picture.
At some point, it would make sense to evaluate the question of how much
IPv4 space is being sat on because of the costs of registering IPv6,
etc. Of course, this is the opposite problem: we're now talking about
dictating RIR business models.