[bestbits] The decentralization of IP addresses

[Cedric Knight]

Hi - my first post to bestbits list, cutting down on the cross-posting a
bit.

On 28/11/15 06:18, David Cake wrote:
>> On 27 Nov 2015, at 4:49 AM, willi uebelherr <willi.uebelherr at riseup.net> wrote:
>>
>> The decentralization of IP addresses.
>>
>> We need a completely self-organizing Internet. And this is possible
>> only through massive decentralization. We can look at the
>> difficulties at the beginning of the Internet with tolerance. They
>> were mostly of a technical nature. But today we have other
>> conditions. And from these other conditions arise other
>> possibilities.
[snip]
>> The decisive factor for this solution that we need in the future no Internet Governance.
> 
> Systems that are geographically based, like the postal and phone
> systems, involve a great deal of governance. Just ask the ITU. The
> main difference is that it is largely done by states. Why would your
> proposal be different?

Good question, but given the possibilities opened up by ICTs, would it
not be premature to assume that other models *cannot* exist, and that
managing an address space (or certification such as PKI) always has to
require any central or hierarchical co-ordination?  There is often an
assumption that fully decentralised addressing does not scale as it may
require each node to store a full set of identifiers - however, that's
not so different from internet routing tables that consist of over half
a million IP prefixes.  It may be possible to route globally unique
identifiers and protect them against forgery in a mesh-like system: see
for example .onion addresses.

Much of the way the net has evolved (for example,
manufacturer-programmed MAC addresses) is a historical accident,
including use of global registries.  See Radia Perlman's talk on
"Folklore of Network Protocols" at
https://www.youtube.com/watch?v=LbZ5ruco0jM

>> No organizations that compete for the award of rights of IP
>> addresses. No organizations whose livelihood is based on the sale
>> of global IP addresses.
> 
> Why would states not compete for the award of rights of IP addresses
> if they were organised globally? Why do you wish to get rid of the
> RIRs? How do you think this relates to names, do you think that
> global generic (as opposed to country) names are a bad idea or is
> your proposal only for numbers?

Is the (IAB/IETF >) IANA > RIR > LIR > user allocation pattern always
going to exist?  Fortunately I don't think states have yet tried to
interfere with RIRs in the way they do with DNS, but that may have been
simply that they are less visible to policy-makers, and RIRs may need
protecting.  There is no necessary relationship between ccTLDs and
states, but states may feel a sense of "ownership" of related TLDs.  The
public [I]nternet is classically a set of federated services; but
corporate, governmental and economic forces have in the 21st century
tended to favour centralisation, with fully decentralised and P2P
functions partly a reaction in the opposite direction.

I'd suggest the theory and details are more worth raising with IETF/IRTF
participants like those formalising .onion addresses, TRILL and
decentralised protocols.  You would first need interoperable
specifications (or building on top of what we have) and then an economic
or social mechanism to adopt the new system.  Look at how long IPv6 is
taking... in the meantime though, IMHO questions about ICANN and IANA
need to remain high on civil society's agenda.

-- 
All best wishes,

Cedric Knight
(GreenNet)