[ppml] PI and potential aggregation

>    Now to the minor change... Allocate out the whole /44, as opposed to
> a /48 out of a reserved /44.  The advantages are:
> - ARIN stays out of the business of allocating "site-sized" addresses
> for IPv6.
> - This relieves qualifying organizations from developing justification
> for "the remainder of their /44".
> - ARIN doesn't need to deal with "return to the trough" process.
> - This results in no higher burn rate of IPv6 addresses, if reserve
> really means reserve.
> - We maintain 4 bits worth of potential aggregation we would have lost.
> - This gives us an opportunity to keep site-sized advertisements out of
> the IPv6 routing table.
> - this should increase the probability that these /44 blocks will
> continue to be advertised and routable in the steady state IPv6
> Internet.

Very good analysis and I agree with this. 
The idea of reserve addresses here, reverses the position 
of end user members and ISP members. In IPv4, ISP members
continually come back to the trough and therefore expend 
some effort to be sure that ARIN's policies and processes
are reasonable. In IPv6, ISPs don't come back to the trough
and this reserve concept would lead to end users doing that.
The end result would be that end user organizations will be
more motivated to see that ARIN is in line with their 

--Michael Dillon