Moving Right Along on the Inclusions Table...

[John C Klensin]

--On Wednesday, 03 January, 2007 15:30 -0800 Kenneth Whistler
<kenw at sybase.com> wrote:

> Martin asked:
> 
>> At 04:27 06/12/22, Kenneth Whistler wrote:
>> 
>> > Internet identifiers don't *require* word separators, and
>> > if anything the predominant use of the existing "-" has
>> > been to cause trouble and spoofing, rather than to
>> > "do the job".
>> 
>> Can you explain? Do you mean e.g. spoofing www.microsoft.com
>> with www.micro-soft.com? Or something else?
> 
> Yes, I meant stuff like that.
> 
> In other words, we can argue that hyphen is supposed to
> function as a word separator, and expect people would use
> hyphens to "improve readability" in domain names. I.e., for
> English at least, people "should" register:
> 
> night-at-the-museum.com
> 
> But of course, they don't. They register:
> 
> nightatthemuseum.com
> 
> instead, not because it is easier to read, but because it is
> shorter and *way* easier to type correctly. In fact, if you
> actually type night-at-the-museum.com, you get redirected by
> a cybersquatter to (today) an ad for laser hair removal.
> 
> The trend in real use, as I see it, is for people to *expect*
> that they can type in multiple-word domain names they "hear"
> as phrases, without hyphens (or spaces, of course) and get
> to the right location. I suspect practice in domain name
> registry is tending to follow that expectation, and in such a
> context, use of hyphens is more likely to come from a spoofer
> than from the intended registrant.

This is certainly consistent with my observations.  There is an
added tool that is sometimes used, although it doesn't work with
IDNs, and that is to register, and actually place in the DNS,
NightAtTheMuseum.com.   If the user types
"nightatthemuseum.com", that is always matched, but
NightAtTheMuseum comes back from queries and, depending on UI
design, may appear to the user in a number of contexts.   Again,
that isn't available to us with IDNs (since case-mapping occurs
exclusively in the client and only lower-case strings are
registered) but it is at least a small part of the current
practice.

> I just don't see any *requirement* here to start down the
> road of looking for word separators on a script-by-script
> basis, for Ethiopic or anything else, for domain names.
> This is a specialized string identifier usage that requires
> some degree of attention to completeness of repertoire
> on a per-script basis (i.e., not leaving out letters and
> combining marks willy-nilly), but which does *not* need to
> obey the same rules of punctuation for legibility (or word
> separation) that normal text does.

Agreed.

   john