[12477] in Commercialization & Privatization of the Internet
About the creation of new 3-letter domain(s)
daemon@ATHENA.MIT.EDU (Barry Shein)
Wed May 18 21:11:17 1994
Date: Wed, 18 May 1994 20:45:42 -0400
From: bzs@world.std.com (Barry Shein)
To: dcrocker@mordor.stanford.edu
Cc: dfazio@mr.net, com-priv@psi.com
In-Reply-To: Dave Crocker's message of Wed, 18 May 94 16:29:37 -0700 <199405182329.QAA00877@Mordor.Stanford.EDU>
[BZS NOTE: Near the very bottom of this is my itemized summary of some
of the issues, people forced to live within 72 year long lives, on
average, may want to skip to that right now.]
>From: Dave Crocker <dcrocker@Mordor.Stanford.EDU>
>Perhaps I stepped into the twilight zone?
Careful, as the arguments fail the sneering starts?
Let's try to consider this with some objectivity for a while at least.
At least it's a constructive suggestion with some delineable merit (as
opposed to a mere objection.) If it's rejected, hey, I'll wait until
2004 and maybe then it'll be more obvious (one way or the other.)
>To repeat: If you change how the number is used (presented, etc.) then
>you lose some of the benefit of its familiarity.
I just don't consider using phone numbers in reverse order to be a big
problem, particularly when compared to the problems raised with other
schemes thus far suggested.
I do consider the internationalization problems with geographic naming
to be a bigger problem than merely reversing a familiar and
well-defined string of digits.
So what's the answer, is it .moscow., .moskva, .mockba., all of them,
one of them, what?
Do we insist on transliterating to Latin-1 (or even USASCII)?
By what standard do we transliterate?
Do we try to eventually encode all of UNICODE into geographic domain
names? Do we allow 8-bit characters? 16-bit? More? Do we just hope
this issue is never raised or just say "USASCII end of discussion"?
Can one even code something as common as an umlauted character into
this geographic scheme? What if two place names differ only by one of
these diacritical marks, within the same country? Is that a problem?
Do you believe no such cases exist (unlike phone number conventions
this is a very, very large field to try to exhaustively consider)?
How do we deal with geographic areas for which there isn't even common
agreement on the base name? Is it the Danube or the Danu? Is it Munich
or Munchen? Beyond convention there's an enormous font of
contentiousness in geographic naming (go say "macedonia" to a Greek
nationalist in the wrong context, and then duck fast.)
Presenting phone numbers in the familiar order in user interfaces is
just syntactic sugar, a red herring of a problem and only barely
within the purview of this discussion except it would be nice if any
standard adopted tried to make this lexically simple to do. Like the
"only the country code may be three-digits" suggestion.
>I take it that you are not familiar with the Rose/Malamud remote-printing
>experiment (which one otherwise might call Internet-to-fax)? It uses
>phone numbers. You put one number in each dns subdomain, in reverse
>order. The human factors are poor.
Why? Why are the human factors poor? Perhaps this is a job for human
interfaces. Was this actually measured? Was it written down? Or just a
subjective impression?
It just seems odd to me when someone concludes they have shown that
phone numbers are too hard for people to use.
And I don't particularly consider swdsrv.edvz.univie.ac.at,
rs3.hrz.th-darmstadt.de, granuaile.ieunet.ie or
kdsub1.kdenshi.tosho-u.ac.jp (all real examples) to be lilting and
mnemonic names either.
Maybe it's just me.
>Rules for number grouping are highly variable. That's why the remote-
>printing stuff doesn't try to guess or make the user guess.
There's a lot of hand-waving here, Dave.
Does anyone have one example of an existant phone system that uses
three digits in its least significant portion?
Although I can believe it exists the observation that it only lends
itself to 1,000 possible values in the least significant portion would
tend to make one want to see an example rather than take it on faith.
My guess is that if such an example exists it is because two groups
have been arbitrarily split and could be recombined for those odd
cases, or some such. But an example would be nice.
As I said before it's not a huge and immeasurable domain of activity.
Are there more than about 100 conventions used throughout the world?
Are there more than 50?
That grouping-size varies is not a problem by itself unless it makes
any sensible algorithm impossible. They're just domain names and can
be between whatever, 1 and 32 or 64 digits per portion (depending on
which de facto or codified standard you prefer, but the range is more
than sufficient even in the most conservative scheme.)
I think this time round is going the same way as in '84: Ugh, digits,
it'll never fly, people can't use phone numbers...let's start sneering
and be done with it.
I don't love the aesthetics, don't get me wrong.
But I think the appeal to an existant and absolutely global set of
standards that most of the world's population (likely to use this at
all) is completely familiar with a priori (even if we just say "but
they're backwards") has its merits.
[BZS: THAT SUMMARY I MENTIONED ABOVE]
Here are my plusses for using phone numbers:
1. There exist well-established standards for them (format etc.)
2. They are a priori familiar to most of the world's population who
are likely to ever use DNS.
3. There is virtually no internationalization issue. Digits
are a near-universal character set.
4. Telephone company directory services and every other form of
existent infrastructure for looking up telephone numbers (CD/ROM,
phone books, minitel, etc) can be leveraged with this scheme.
5. The telephone companies end up helping a lot as name space
managers, even if unwittingly (at first, perhaps that would change.)
6. It eases a lot of the potential trademark issues. Your phone number
is your phone number.
7. They can be geographic if you like, or that can be minimized (e.g.
use 1-700 or 1-800 numbers or equivalent.)
8. It leverages a standard that has developed over 100 years of
practice and that we know is in widespread, global use and appears to
solve more or less the same problem we are attempting to solve here.
Ok, the minuses:
1. To be consistent with current DNS practice they need to be used in
reverse order and some people might find that awkward (at first,
whether we'd all get used to them soon enough is an open question.)
2. Apparently a lot of people look at the idea and, at first
glance, have an emetic reaction to it.
3. There exists at least one claim that this scheme presents
a human factors problem. I claim that the current practice
of around 2 or 3 billion people raises some doubts about that.
-Barry Shein
Software Tool & Die | bzs@world.std.com | uunet!world!bzs
Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD
