[OpenID] [SPAM]Re: [LIKELY_SPAM]Re: [LIKELY_SPAM]Re: Combining Google & Yahoo user experience research

Mon Oct 27 18:38:05 UTC 2008

Sticking the string verified in front of a attrvalue, vs sticking some xml nested "confirmation" tagging in front...of the same attr value. What's the difference (excepting type theory and security engineering methods)?

If the conforming processor treats one prefix as another, semantically, there is none. All one needs the unambiguous syntax.

Simplistic prefixes feel more like openid culture. Don't want to make openid turn into saml, though semantic convergence would be nice, particularly for gatewaying.

Not sure how much convergence there really is yet within the saml community itself, tho, the more I look into it, even over basics like metadata and key management. jar hell seems to have evolved into profile hell....

________________________________
From: Paul Madsen <paulmadsen at rogers.com>
Sent: Monday, October 27, 2008 2:10 PM
To: George Fletcher <gffletch at aol.com>
Cc: general at openid.net <general at openid.net>
Subject: [SPAM]Re: [OpenID] [LIKELY_SPAM]Re: [LIKELY_SPAM]Re: Combining Google & Yahoo user experience research

George, seems to me that the PAPE model for expressing that an email address be/is 'verified' is very different (and preferred) to explicitly labeling the attribute name.

Stick the string 'verified' in front of an attribute name, and the implication is that it is somehow qualitatively different than a 'non-verified' attribute. And of course it (unless supplemented with something equivalent to PAPE) hides the detail as to how the verification occurred.

FWIW, I don't think PAPE is actually the place for this sort of attribute assurance information ... but the model could be reused.

paul

George Fletcher wrote:

I think the difference here is that the process that the RP desires is
something like...

1. have the user enter their email address
2. determine that the domain owner of the email address supports "email
verification"
3. use a pop-up window to direct the user to the domain owner's "email
verification" endpoint
4. have the user prove "ownership" of the entered email address
5. return verified state to the RP

I would also like to see either a PAPE policy or an AX attribute that
signifies "verified email" where if the RP trusts the OP the user
doesn't have to do any additional authentications.

Thanks,
George

Nat wrote:

Perhaps you can construct a PAPE policy that signifies the verified
email and send the email by SREG.

=nat at TOKYO via iPhone

On 2008/10/27, at 21:21, George Fletcher <gffletch at aol.com><mailto:gffletch at aol.com> wrote:

In the discussions I've had, there was one other use case. That is a
site that isn't ready yet to support the full OpenID cross-domain SSO
concept, yet wants to streamline their registration process such that
they don't have to use the out-of-band email verification mechanism.  In
this case, a small extension to the OpenID protocol (similar in concept
to AX) could be constructed that would allow a user to verify their
ownership over the email address using a "synchronous" process vs the
current async one.  So, if the RP's only concern is to verify that the
user "owns" the email address they've specified, then the RP doesn't
want the email address mapped to an OpenID, they want to know that the
email address is valid and the user knows the password to it.

This use case isn't really related to OpenID other than it's possible to
use the current flow and protocol (with a small extension) to implement
it. If AX is about exchanging attributes, this would be an extension to
"verify" attributes:)

Thanks,
George

Chris Messina wrote:

On Sat, Oct 25, 2008 at 3:03 AM, George Fletcher <gffletch at aol.com><mailto:gffletch at aol.com>
wrote:

I think there are at least two use cases involving email addresses
that
can be easily confused...

1. Use the email address as an indicator or pointer to a valid
OpenID as
the email address is an identifier that the user currently remembers.
 - this is the use case that EAUT is targeting and, if I understood
correctly, what Chris is discussing as well

Yes. The point is, until MySpace users become familiar with using
their "MySpace OpenID" or "OpenID" (depending on how we recommend
MySpace market this behavior), we have ample reason to make it
possible for people to use identifiers with which they're already
familiar and use in common practice to sign in to web services:
namely, email addresses.

It also would provide a means to "upgrade" legacy accounts keyed to
email addresses to use remote authentication via OpenID... reducing
the need to remember discreet passwords (or sharing a unique password
between sites).

Although there will be increasing numbers of URL-formatted/based
identifiers out there for people to use for OpenID authentication, it
seems that a great way to simplify OpenID's offering and to make it
more palatable to those who argue that they identify themselves by an
email address is indeed to figure out the best way to enable that
possibility, and to disarm that argument.

2. Verify an email address for those RP's that want/need/require a
"verified email address"
 - this is more about the RP getting a verified identity attribute
 - the expectation is that an OpenID based flow would allow a user who
has to verify their email address to do it in "real time" rather than
the async email method used today

A common complaint that I hear from people using OpenID to sign up for
new services comes down to an "OpenID tax": once they've successfully
authenticated with OpenID (which definitely isn't quite as fool proof
as I would hope), they're immediately asked to provide an email
address and then to validate it, receiving an out-of-band token. The
feedback I hear is that most people would rather just sign up with
their email address in the first place than have to deal with this
silly process.

This will continue to be a valid criticism unless or until we are able
to make URL-based verified identifiers more useful than email
addresses to RPs.

I believe we need to keep these two use cases separate because the
intentions/outcome is really quite different.

For clarify of conversation, +1.

Though solving both issues with one protocol/approach would be
ultimately ideal.

Chris

SitG Admin wrote:

I'd guess that a contributing factor here is that most OPs don't
support
passing the email address via SREG.

Since the discussion here seems to be about not only verifying E-mail
addresses, but using them in place of a URI, does it matter whether a
RP supports *receiving* an E-mail address via SREG?

I don't want users' E-mail. I don't *need* users' E-mail. I don't
care. Is the requirement (that a user be able to receive E-mail at
their address) going to require me to be able to send them E-mail so
I can confirm their OpenID?

-Shade
_______________________________________________
general mailing list
general at openid.net<mailto:general at openid.net>
http://openid.net/mailman/listinfo/general

_______________________________________________
general mailing list
general at openid.net<mailto:general at openid.net>
http://openid.net/mailman/listinfo/general

--
Chief Architect                   AIM:  gffletch
Identity Services                 Work: george.fletcher at corp.aol.com<mailto:george.fletcher at corp.aol.com>
AOL LLC                           Home: gffletch at aol.com<mailto:gffletch at aol.com>
Mobile: +1-703-462-3494
Office: +1-703-265-2544           Blog: http://practicalid.blogspot.com

_______________________________________________
general mailing list
general at openid.net<mailto:general at openid.net>
http://openid.net/mailman/listinfo/general

--
[cid:part1.01040502.00040806 at rogers.com]<http://feeds.feedburner.com/%7Er/blogspot/gMwy/%7E6/1>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gMwy.1.gif
Type: image/gif
Size: 11322 bytes
Desc: gMwy.1.gif
URL: <http://lists.openid.net/pipermail/openid-general/attachments/20081027/e6b83b60/attachment-0002.gif>