[jdev] confusion over user at domain JIDs and email addresses?

[Mario Salzer]

Peter Saint-Andre wrote:
> 
>> I recently tried to advocate a similar user at example.com syntax as
>> shortcut for Web URLs (as replacement for less user-friendly OpenID
>> ...
> Well, gosh, if you remove the URI scheme, then an HTTP URI looks like an 
> FTP URI looks like a Gopher URI, an email address looks like a Jabber 
> address looks like a SIP address, etc. So don't remove the URI scheme.

You got something into the wrong throat there, I fear (reads like
you thought more about corrupting web pages).
That "URL abbreviation" thing was of course not meant to be used
as URL itself, but only as user input in specifically prepared
applications. Like with FTP clients + Web browser, the protocol/
scheme gets rather implicit in such cases. You wouldn't likewise
expect a Jabber client to nitpick "You must prefix your account
name with jabber:// else I won't accept it in the input field for
the JID here!".

On the contrary, there is a URI scheme for the thing I was onto.
It's just that nobody is expected to use it, ever - because those
userid shortcuts were entered into a couple of special input fields
only.  <http://xprofile.berlios.de/USERID.txt>
The problem is that those user ids are exposed as plain text
occasionally then, and this can disturb users (not so much XML
toolkits, let alone hurt WWW clients).

-

But back on topic. I'm more interested to hear how many people
(especially AOL users) misinterpret JIDs as email addresses, even
when expicitely labeled that way.

Consider for example a web page with textual content like follows:

    user: name
    jabber: creeesh at jabber.org
    popoflux: quit at example.com
    email: text at example.net

Email harvesters would register all three. But which one would
the average user pick for sending emails? Because of the identical
syntax all three look like email addresses - but as the alerted
reader has noticed, only one of them would work.

Most AOL users won't know what jabber means, but then also overlook
that one of the three is explicitely marked as email address (may
be even a link on this hypothetical web page). Whereas "popoflux"
is of course useless (made up to exaggerate here).

The usability problem of jabber and popoflux here is, that less
technically inclined people consider every string with a single
"@" inside to be an email address. At least that is the general
assumption.
But are typical AOL users really that dumb? And must @-strings be
eschewed for anything but E-Mail addresses therefore?

Are there any findings on how often mislead emails (= non-spam)
really hit Jabber servers? Usability assumptions are nice, but
I'd of course prefer any real-world statistic on this topic.

How many Jabber addresses eventually get misinterpreted as email
addresses of course highly depends on WHERE and especially HOW they
got publicized. If the context doesn't reveal that strings like
"user at jabber.org" are IM handles, then the likelihood of mislead
emails raises.
But how does that problem estimate in numbers? How many confused
users have reported 'broken' user at jabber email addresses to date
for example? How many SMTP port connections on jabberd hosts have
been verified to not carry spam, e.g.?

The JID vs EMail topic has sure already been discussed to death on
this list, but I fruitlessly searched the archives for concrete
notes. If I wasted so much time subscribing to the wrong list, then
I'd like to know that too ;)

mario