[ppml] Policy Proposals 2007-18 and -23

[Edward Lewis]

At 3:36 +0100 10/18/07, <michael.dillon at bt.com> wrote:

>Currently we give out addresses based on demonstrated technical need for
>them. Why should we ever change this?

The principle of "least surprise."  It's good to know that the game 
is about to end.

It's true that we currently have a disparity in run-rates at the 
RIRs.  Do we know that the run-rate at each RIR is a constant, has 
the newest RIR been operating long enough to predict it's workload in 
a year or two?

All of "this" (whether to *do* -18, -23, nothing, etc.) is based on 
conjecture and opinion.  There are many scenarios that may play out. 
The best that we can hope for is to pick the approach that has the 
best "expected value" according to some metric of happiness.

For every reason that can be given to agree with "leaving well enough 
alone" there is a reason to say that we should treat the last /8's 
specially.  Playing a hunch from my experience, trying to find a 
happy medium amongst the extremes is the best strategy.

I like reserving just the last 1 slash-8 per RIR for an even 
distribution.  It's more signal than significant to me.  Perhaps the 
slowest burning RIR will "catch up" and we will wind up with this 
being the same as the current approach.  Maybe that "surplus" to the 
slowest RIR is more needed there for dual-stack transition.  It's all 
conjecture.
-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                                +1-571-434-5468
NeuStar

Think glocally.  Act confused.