[12531] in Commercialization & Privatization of the Internet
Re: About the creation of new 3-letter domain(s)
daemon@ATHENA.MIT.EDU (Dave Crocker)
Fri May 20 11:19:09 1994
Date: Thu, 19 May 1994 21:20:17 -0700
To: Stef=com@nma.com
From: dcrocker@mordor.stanford.edu (Dave Crocker)
Cc: com-priv@psi.com
Stef,
At 9:08 PM 5/18/94, Einar Stefferud wrote:
> nma.17301.drey.huntingtonbeach.ca.92647-5615.us
(not that it helps much, but I only intended the 92647 portion of US postal
code.)
> nma.17301.92647-5615.us is unique, so why all that other stuff.
The fact that the particular string you cite is unique isn't all that
useful, unless a) you can define the algorithm that produces it, and b)
assert that it is globally applicable.
THAT is the requirement I'm trying to satisfy: A scheme for easily
(automatically) assigning domain names for the global mass-market.
>And what if I move and someone else now owns my name because they now
>live where I lived.
I am not nearly as enamored of the permanence as are others. Permanent
assignments of unique identifiers are something with which we have no
meaningful experience. I'm trying to solve one problem at a time, in this
case, scaling of DNS usage for several billion people. Puting 'permanence'
into the equation strikes me as an appealing distraction.
>We are screwing around here with people's identities, and acting as
>though there is no reason for anyone to become emotionally attached to
>their identities.
Like, their postal address is their identity?
> people very strongly bind
>themselves to their identifiers, precisely because their name is how
That's why I'm suggesting re-using strings that are already in their repertoire.
>The current .us scheme for locality based addressing is also not going
>to work, mostly because people balk at being forced to use names they
>don't like, so they go off to .com where they are not so constrained.
Yes, but .com is going to fall apart. It doesn't scale either
administratively or in terms of storage retrieval. The fact that it is
surviving does not mean it isn't in trouble. The best suggestion that came
out of a working group task on this was to subdivide by virtue of initial
letters, so that I guess NMA.COM would go into NMA.N.COM, but eventually
might go into NMA.N.M.COM. This does not strike me as a viable solution.
>For example: Hunting Beach will shorten to HB, but there is at least
>one other HB (Hermosa Beach) that "needs to be distinguished"
>according to the .us scheme. So what do I have to do for a locality
>name in ca.us? Frankly, I see no need to distinguish one HB from the
(Hmm. One would think that you and I hadn't hashed this out before...)
Acronyms create collisions. On the other hand, use of postal code brings
down the scope to a level that makes most city acronyms viable, I suspect.
>Who cares, if as Jon Postel so many times said: "A Name is just Name".
A name may well be a name, but administrative overhead is a pain.
>Look at dbc.mtview.ca.us, just because there are two MT's in .ca.us.
>Can you imagine or appreciate how long it takes to tell someone that
>your address is "dbc.mtview.ca.us" over the phone?
The problem is with the short-form for Mountain View. It isn't a problem
if you use postal address. In this case, there is nothing to prevent
having the string MountainView registered as an alias to mtview.
>What we need are to let people register under .ca.us if they wish, and
doesn't scale.
>can claim unicity. Same rules as in .com. If I could have
doesn't scale.
>Or how about deriving 2 character county codes? Most people would not
The idea of county reference is theoretically reasonable. I didn't include
it because postal code is sufficient for reducing granularity and is more
familiar to people.
> Cheers...
Indeed.
Dave
+1 408 246 8253 (fax: +1 408 249 6205)
