[arin-ppml] IPv6 /32 minimum for extra-small ISP

[Milton L Mueller]

> The people who have full tables are well aware of the
> costs associated with a routing slot.  And, it's not just churn, it's
> the actual routes in your RIB, then the FIB, plus the churn.  You have
> memory and processor requirements that increase with each new route
> added to the DFZ.

Sure. And everyone knew that apples fell down, not up, before Newton.
 
> You seem to have a hypothesis hiding somewhere behind the Socratic
> questioning.  

A hypothesis? No. What I do have is a hunch that the social and private costs of various approaches to routing policy/aggregation are much less well understood than we think they are - and therefore we have an inadequate basis for understanding the economic and efficiency implications of alternative policies.

Hmm, it might be possible to formulate that as a hypothesis, but I am doing it on the fly: 

H1: RIRs' discussions of address policy, esp concerning the tradeoffs concerning aggregation and accessibility of address blocks, and our ability to formulate appropriate policies, would improve if we had a more sophisticated model of how the size and distribution of the costs of route additions. 

--MM