Meaningful names and IP addresses (Re: [wg-c] Re: nine principles for domain names)

[Harald Tveit Alvestrand <Harald@Alvestrand.no>]

At 11:28 21.02.00 +0100, Philip Sheppard wrote:

>1. We do believe that the assumption that all gTLDs will/should stand for
>something is valid. The alternative is to not bother with a gTLD and use
>only the IP address. The moment you adopt the idea of gTLDs the net user
>will believe they stand for something. To give anything a name is to endow
>identity. If anyone believes in a system that does not endow identity they
>should argue for IP address only.

unfortunately this argument, while attractive, is false.

The name/address separation is critical to the operation of the Internet; 
keeping routing operational demands that addresses be tied (albeit loosely) 
to the physical structure of the network, and can be changed (albeit at 
some cost) when the network changes; while the use of names allows us to 
guarantee globally unique identifiers that are tied to organizational 
hierarchy, and are guaranteed to be immune against most, if not all, 
technical demands for change.

Thus, even if domain names were all of the form "a2w8r", or similar 
meaningless strings, they would still be vastly preferable to IP addresses 
for many purposes where long-term stable references to network resources 
are needed.

I personally agree with the conclusion (that gTLDs should be expected to 
"stand for something"), but not with the premise (that one could use IP 
address directly if one did not need the "stand for something" aspect.

Regards,

                Harald Tveit Alvestrand


--
Harald Tveit Alvestrand, EDB Maxware, Norway
Harald.Alvestrand@edb.maxware.no