Those Pesky Slashes

[Xageroth Sekarius]

It's for some of those same usability issues that Martin is raising
that I planned on using an alternative syntax to URL's completely in
the project I'm involved with. Aside from just usability, there's my
concern for privacy in OpenID. Some people are going to want to log in
to a site but not necessarily share with them all their daily
happenings or deepest feelings. It seems odd to have to register a
seperate subdomain for the sake of privacy and then carry two "global
identifiers", one for some privacy and the other for sites you trust
or don't care.

Before OpenID came along, I was planning on using <a
href="http://xageroth.blogspot.com/2005/05/whats-your-sein.html">sein</a>
which was meant to apply to all identifiers. Using a sein, levels of
anonymity could be given depending on the site being visited while
retaining a single identifier for yourself. The premise is simple,
since e-mail addresses are tied to a particular technology, using a
real e-mail address would be intrusive to privacy, and using a fake
e-mail address would step on top of real ones, just change the syntax
from "@" to "~". It wouldn't be difficult to expect users to remember
"jane.doe~somedomain.com" as their identifier since it feels familiar.
To parse you would detatch "somedomain.com", get
"http://somedomain.com/", follow a LINK redirect to identity services,
then query that service about "jane.doe".

This would also shorten the query URL since LiveJournal users could be
expressed as
"script.cgi?openid.is_identity=jane.doe" instead of
"script.cgi?openid.is_identity=is_identity=http://jane.doe.livejournal.com/"
using the identifier "jane.doe~livejournal.com" and having a link tag
at "http://livejournal.com/" which redirects to LJ's identity
services.

Consistency, as Ben Hyde pointed out, is definitely important but at
least having an "official" alternate syntax to alleviate some of these
usability/privacy concerns would diminish the need for unnecessary
identifier creativity.