Re: Jonathan Jensen - What’s your number

=barney.craggs — Thu, 01 May 2008 18:27:55 -0000

The premise behind this isn't actually sound. Phone numbers and domain names are what's known as a controlled namespace; in other words
they are controlled by centralised and delegated authorities. So for phone numbers they are actually assigned by the ITU (?)
at a country level and then local telco's receive delegated authority to assign the numbers therein.

Domain names are EXACTLY the same. The top level domains are controlled, as are the region specific ones.

In both cases you actually LEASE the number or domain from the delegated authority and therefore as with any controlled namespace there is
little or no possibility or probability of ever actually "owning' the number. This "centralised" control is what ensures uniqueness.

/* Gets a little geeky beyond here so bail out now if you want */

The true problem with telephone numbers is not is the ownership rather the fact that traditionally the number has been tied to the service
provider directly. So if you obtained a telephone number from one Telco, as it had delegated authority for that chunk of numbers you couldn't
move it to another Telco. This was in part I suspect because telephone numbers are regionalised such that the average human is capable of
remembering the numbers.

Number portability seeks to resolve the changing of service providers, however is only possible within geographical boundaries say the same
exchange - remember the heritage of circuit switching here.

What is more useful in the modern and connected world is actually a controlled namespace that is NOT geographically constrained. One where
you can be assigned an non-reassignable number (one that never changes and is always yours) which you can attach to whatever service you
desire. Now IPv6 gives us a namespac big enough to handle this scenario but have you ever tried to remember an IPv6 number? Most people
couldn't. This is where an abstracted identifier technology comes into play.

With an abstracted identifier, such as iNames (based on the XRI standard from Oasis) you are able to create and link a human readable
identifier (mine's =barney.craggs) to an underlying non-reassignable iNumber (directly mappable to an IPv6 address). This has a number of very
cool advantages not least being the ability to have a single number if desired but also the ability to assign different iNames to the same
iNumber for different purposes. So for example I could give business one iName and my friends a different iName both of which can resolve to the
same iNumber (if I want) or be routed to multiple destinations.

The possibilities are immense and I have over-simplified the technology for this comment, but in essence;
- controlled namespaces are good as long as you acknowledge their limitations,
- geographically constrained identifiers are limiting especially when the geography is actually directly tied to a service provider,
- the use of combined re-assignable and non-assignable identifiers opens up a world of possibilities and a powerful driver for consumer
choice.

Disqus - Latest Comments for =barney.craggs

Re: Jonathan Jensen - What’s your number