user@domain URNs, independent from OpenID spec

Mario Salzer mario at erphesfurt.de
Tue Jul 5 04:54:50 PDT 2005


Xageroth Sekarius wrote:
> Interesting points but ultimately it's still an e-mail address. If I
> share it with someone it won't be my OpenID, it won't be my webpage,
> it won't be my phone number or my zip code, it will forever be my
> e-mail address even... when you make it so that it isn't. Stepping on
> top of (arguably) the most shared personal identifier with a
> completely different usage and no actual correlation between identity
> layers is not a weak point.
> 
> Explaining the difference to nerds? Easy as hell. Explaining why that
> e-mail link is not broken even tho it took your Mom to a webpage
> instead of opening Outlook? Much harder.
> 
> It's also confusing from a robot perspective if you think about it.
> 

You mean something like email harvesters, don't you?
I wouldn't be worried about that one at all. I first would be worried
if user at domain shortcuts for OpenID would be used by webmailers and
a OpenID user at domain identifier became a real email address (- turns
out to be quite the opposite concern).

And then I still don't understand why "@" shall ultimatively and
only ever mean E-Mail. There are lots of other schemes which use this
delimiter, think of Jabber. Many people will complain over there too
that this isn't a universal address and user identifier and deosn't
work as real email address automatically - but how did this hurt the
protocol per se?
And I also dispute the long-livingness argument ("forever be my e-mail
address even..."). Phone numbers often only change if you move houses,
but Email addresses get abandoned pretty regularily these days.

The point of resolvability (also because this idea is as new as DNS
was 20 years ago, doesn't make it unusable) seems also rather negligible
to me - right now we could have OpenID URLs like
   "http://provider.com/share/openid/572509"
which don't bear ANY meaning to human users (and resolve to empty Web
pages on top - someone alert TBL please!). How much more convenient
is "572509 at provider.com" in comparison, even if it doesn't work as
email address right from the start?
   (I made up this example, because I already forgot the ugly URL for
one of the first OpenID demo servers I signed up. No lie, I really
forgot it.)

But btw, I almost fell for the too-emailish argumentation myself, and
wanted to propose something like "user#domain" and "user*domain" as
alternative to the sementically overcharged @ syntax. But then such
shortcuts would instantly get as ugly as URLs again.
So what I'm trying to say is, this whole discussion circles around
the question if an "at @" denotes the SMTP protocol or not.
I disclaim this.



More information about the yadis mailing list