[CCWG-ACCT] Resolution of Mission Language related to regulation and contract

[Andrew Sullivan]

On Tue, Nov 24, 2015 at 04:39:56PM -0500, George Sadowsky wrote:
> So let's probe this a bit.
> 
> Does that mean that future guidebooks can't disallow any future strings from consideration, unlike round 1?  I think you'll find a lot of people who would disagree that
> 
> 	.horribly_insulting_string

This is the reason I included the "use in natual language" bit.  The
reason that string is "horribly insulting" is because of the
linguistic context, not the DNS.  The DNS doesn't carry the meaning,
but there's nothing wrong with ICANN making policies about strings it
will delegate.  It does that all the time.  It also has a policy, for
instance, that it won't delegate ASCII strings outside the LDH range,
and that strings that have "-" in the second and third positions MUST
be A-labels, and so on.  ICANN also won't delegate underscore labels,
at least not yet.  None of these restrictions come from the DNS
either.

> Much as I'd like to treat domain names as arbitrary strings that
> have no meaning for the computers that parse them, they do have
> semantic content, sometimes very strong and offensive semantic
> content, and they will evoke strong reactions

I think you're conflating "how things get used in natural languages"
and "how things get used globally".  There's a whole literature (much
of it full of urban legends) about brands that didn't translate well
across cultural borders because of supposed meaning in the new
language.  (My favourite urban legend one is the Chevy Nova, which
supposedly meant "doesn't go" in Spanish.  From what I saw of my
neighbour's car, that's what it meant in English too.)

ICANN is no more in the business of deciding what strings are
offensive than it is in the business of deciding what's a country.
Instead, it can appeal to other sources for decisions -- ISO or
community consensus or whatever -- for that reasoning, and merely
needs to be interested in how domain names (especially at or near the
root) are used by people.  That _certainly_ includes thinking about
their natural languages (and then appealing to those other sources for
decisions).

>BTW, if they didn't have semantic content, they really would be
> useless to us, wouldn't they?  If that were the case, we might as
> well go back to using numbers instead.  POTS, anyone?

About 40% of the people I interact with over the Internet realise that
"anvilwalrusden" is an anagram of "andrewsullivan".  Does
"anvilwalrusden" have semantic content?  Well, it does to me (and
probably to all of you now, too -- quick, don't think of a pink
elephant!), but I'll wager a pretty good lunch that it has negligible
semantic content to anyone I've never met in (say) West Vancouver, to
say nothing of South Korea.  Canadians always know exactly what I mean
when I give them my email address, <ajs at crankycanuck.ca>; USians can't
even pronounce it unless they live near the border.  And lots of
labels in the DNS have at best ambiguous semantics anyway -- this is
why certain country codes are able to sell expensive domain names to
television celebrities or those wanting a domain name that looks like
an adverb in English.  And what is the meaning of ns1?  How about
_dkim?  In Korean?  In Urdu?  ICANN gets into very deep water when it
starts trying to be a meaning-maker.  I don't think you want that.

> I think we had better get this right at the beginning rather than
> having it bedevil us in future new gTLD rounds.

If we don't talk about "meaning" and instead talk about the way
people use them, we'll be just fine.

Best regards,

A

-- 
Andrew Sullivan
ajs at anvilwalrusden.com