[discuss] Government Engagement model [1 of 2] (was: Re: IPv6 Deployment and IG)

[John Curran]

On Dec 27, 2013, at 1:49 PM, Brian E Carpenter <brian.e.carpenter at gmail.com> wrote:

> All of which ignores the fact that IP addresses are allocated, routed, and used
> topologically, not geographically. Which means that all discussion of them in
> terms of physical geography and national boundaries is completely pointless.

Apologies in advance for the two-part diatribe which follows, but I believe that 
it is quite informative regarding the situation which we are presently in with 
respect to the collaboration (or lack thereof) between the Internet technical 
community and governments.

Brian is certainly correct, IP addresses are routed topologically, and should 
assigned predominantly on a hierarchical basis in order to facilitate aggregation 
of routing (aggregation in routing is a desirable trait for keeping the Internet 
running, although estimates vary based on your particular technical viewpoint)

The point of maximal hierarchical assignment is when a single IP address block
is issued to service provider, and they in turn make sub-assignments to their 
individual customers.  In this manner, one address block (with a handful of 
routes in the global routing table) may serves thousands of customers (I will
inject a personal view and note that maintaining this very high ratio of new  
users per new address block route is rather important for Internet scaling.)

The reality is that there is no routing aggregation above the ISP level; i.e.
the fact that hundreds of ISPs are in a given RIR region does not mean that 
there is a single "RIR regional route"; in fact, each ISP has to route the IP 
address block they receive, regardless of whether its from the regional RIR or 
a National Internet Registry working in cooperation with an RIR. It is at this 
point that folks often ask why we have RIRs at all, if there is no aggregation 
going on...  The reason to have more than one central IR is to better serve 
each region with consideration of regional factors such as supported languages, 
customs, and regional economics.  The reason to have not too many regional 
registries is that there is some significant registry coordination effort that 
are required - between the RIRs, with related parties (such as the IETF and 
ICANN) as well as in the development of global address policy.

I participated in the ITU IPv6 Study group exploring the question of the need
for Country-based Internet Registries, and had the privilege of explaining 
some of these issues multiple times over the years.  In the most recent round,
we went out of our way to seek out a "problem statement" of how the present
system was not working for any country or ISP, but to no avail.  There were
always rumors of folks who "wanted change" but no one ever vocalizing what
they actually were trying to accomplish and why (to be fair, the entire
intergovernmental format is not exactly conducive to frank discussion...)

It is quite likely that there were governments with very real challenges that 
they hoped to address via country Internet registries for IPv6, but what exactly 
they were, and whether changing the registry model would solve them is unknown
and unexplored by the ITU IPv6 exercise.  There also was a session at one of 
the IGF's touching on this topic, but in typical IGF fashion, we had three or
four discussion themes, one or two speakers for each, 90 minutes to cover it
all, and therefore barely enough time to outline the issue...

What's lost in all of the above efforts is that there is still a problem out
there (or at least the perception of a problem), and despite everything that 
has occurred, we are not meaningfully closer to a definition of the problem
(let alone work towards a solution)... This is very unfortunate, since there
may actually be real issues, with potentially real solutions available, but
instead of asking everyone involved in these efforts "are you satisfied with 
this outcome", much of the Internet technical community has effectively been
satisfied with the status quo outcome without realization that any outcome
is a failure if we never actually discussed the underlying concerns of all
the parties (including governments) that have been involved.  

To some extent, this is inevitable as a result of the engagement model that 
has been developed over the years: 1) Government perceives problem, and 2) 
Calls for process to discuss and possibly regulate or otherwise get involved, 
3) Internet technical community rises up with chants of "Hands off the Net!"
thus giving pause to all, 4) Government then stands down, start the 5) Internet
victory lap...   While this is slightly tongue-in-cheek, it's not to far from
the history of several government Internet initiatives over the years, wherein
the specific initiative was "defeated", and thus meaningful and constructive 
discussion of the underlying public policy concern or mandate never occurred.

It's probably not possible to address the perceived public policy concerns
that leads governments to advocate for country-based Internet registries, if 
indeed the real concern (as alluded by Milton) is simply a path to "more power 
over Internet uses and users."  The real question is whether we can get have
a forum where a meaning discussion of the underlying concerns (and whether
governments would participate in same) in case there are actual problems that
working together we can address.

I'll continue on this point (on trying to improve engagement with governments 
towards constructive outcomes on Internet issues) in another message, by means 
of an example...

/John

Disclaimer:  My views alone.