[Openid-specs-ab] Next steps: Extension ideas

Mon May 13 02:13:15 UTC 2013

> I am not seeing a wave of adoption at anything like the scale the
Internet needs.

I gather that you are talking about the wave of adoption of OpenID Connect.
One of the thing that Eric has been pointing out was that for
authentication, just being able to cope with HTTP is not enough since
password remains in other protocols like IMAP and so on. In this respect, I
think the bindings of OpenID Connect for other protocols are quite
important.
Not doing it does not help the adoption of OpenID Connect.

Also, it is important to note that even if we do not do it here, it will
happen somewhere else as long as there is a demand. It will be resulting in
fragmented / ad hoc implementations. This is not desirable.

FYI, the AB/Connect work actually is rooted in Contract Exchange (CX)
Working group, which gathered requirements from bunch of companies. It is
essentially a contract / payment protocol. In the due course, the CX WG
figured out that factoring out a portion of it would have a more general
applicability, and started Artifact Binding working group, which is this
working group. CX WG actually is waiting the current AB/C work to be
complete so that they can start off from there.

When we analyse the cause of slow adoption etc., we should not base it on
the conjecture but on the fact / figures.
If we do not have a we should deduce from the adjacent cases.
In our case, it is OAuth. OAuth is still building other profiles and
extensions, but it is not slowing the adoption. So, as I said before, the
argument that extensions will slow the adoption has little ground.
We would have to find other causes.

Possible causes are:

1. Low awareness figure: We could do some survey on the awareness of OpenID
Connect.
    Outside of the identity community, it probably is virtually unknown.
    Then, how could applications adopt it? We have to go to their community
and tell about it.
2. Bad developer documentation: Spec is needed for the
real interchangeability, but the developers
    would not read it. There has to be an easy to read introductory
document with sample codes.
    We do not have it now.
3. Chicken and Egg problem: if there is not enough RP that supports it, OP
would not support it,
     and if there is not enough users covered by OpenID Connect, there is
no reason for RPs to adopt.
     We need a strategic expansion plan in this front. We do not have it
either.

These are the things that needs to be addressed in terms of adoption.
Stopping the development of the extensions and bindings does not help.

Nat

2013/5/13 Tim Bray <tbray at textuality.com>

> I think Mike's argument from marketing reasons is pretty strong. I am not
> seeing a wave of adoption at anything like the scale the Internet needs.
>
> There's also an argument from humility. It is obvious to me that we need
> an interoperable basic authentication protocol. Once we start getting
> deployment on that, we'll be in a position to learn from observation what
> the next most important unmet need is; my confidence that we actually know
> know, right now, what's most important, is not high.
>
> -T
> On May 10, 2013 10:44 PM, "Mike Jones" <Michael.Jones at microsoft.com>
> wrote:
>
>>  I’ve thought about this today and while my reaction may surprise you, I
>> feel pretty strongly about it.  I think that we **should not** jump
>> right into defining new Connect extensions because it would send the wrong
>> message to the marketplace.  It would be easy for us to stall adoption by
>> having people think “Connect is fine but I’ll wait until extension X is
>> done before deploying”.  Rather, we should be clearly communicating that
>> “OpenID Connect is done – build it, deploy it, and it will solve problems
>> for you now.”****
>>
>> ** **
>>
>> If we want to move on to new work, I’d suggest that many of us focus our
>> energies on **finishing** something else important that we’ve already
>> started – Account Chooser.  In particular, while there is a site and a
>> JavaScript file, there isn’t a standard.  That needs to happen.  Let’s do
>> that before any Connect extensions.****
>>
>> ** **
>>
>> We need to establish a reputation for finishing what we start.  That’s
>> far more important than starting more things.****
>>
>> ** **
>>
>>                                                             -- Mike****
>>
>> ** **
>>
>> *From:* openid-specs-ab-bounces at lists.openid.net [mailto:
>> openid-specs-ab-bounces at lists.openid.net] *On Behalf Of *Nat Sakimura
>> *Sent:* Friday, May 10, 2013 2:59 AM
>> *To:* openid-specs-ab at lists.openid.net
>> *Subject:* [Openid-specs-ab] Next steps: Extension ideas****
>>
>> ** **
>>
>> Now that the core connect is largely done, we may want to start
>> discussing a little bit about what we may want to do as the next steps. *
>> ***
>>
>> ** **
>>
>> I have three things in my mind. ****
>>
>> ** **
>>
>> 1. granular purpose statement per claims****
>>
>> 2. privacy level certified request object ****
>>
>> 3. link/rel metadata for the responses****
>>
>> ** **
>>
>> 1. granular purpose statement per claims****
>>
>> As of now, OpenID Connect has a facility to indicate the purpose of the
>> use for the entire request object. It should cover 80% of the cases, but
>> sometimes, some of the individual attribute request is not obvious why that
>> is needed. It will be beneficial to be able to show the user how the
>> individual claims are being used. It was discussed in the METI report that
>> was published today. (See
>> http://nat.sakimura.org/2013/05/10/info-label-win/ for more details). It
>> is possible that it becomes a part of new guideline in Japan. ****
>>
>> ** **
>>
>> The implementation of it is simple. We just need to define the per claim
>> usage. It could go into individual claims as the "purpose" member. ****
>>
>> ** **
>>
>> 2. privacy level certified request object****
>>
>> ** **
>>
>> The idea is simple. The privacy commissioner or privacy trust framework
>> assessor signs the request object after determining that it is following
>> the privacy principles such as data minimization. Then, we may be able to
>> skip the consent dialogue. (Sending the notification should be coupled with
>> it.) ****
>>
>> ** **
>>
>> 3.  link/rel metadata for the responses****
>>
>> ** **
>>
>> Basically, something like
>> http://tools.ietf.org/html/draft-sakimura-oauth-meta-02****
>>
>> ** **
>>
>> Any additional ideas welcome. ****
>>
>> ** **
>>
>> --
>> Nat Sakimura (=nat)****
>>
>> Chairman, OpenID Foundation
>> http://nat.sakimura.org/
>> @_nat_en****
>>
>> _______________________________________________
>> Openid-specs-ab mailing list
>> Openid-specs-ab at lists.openid.net
>> http://lists.openid.net/mailman/listinfo/openid-specs-ab
>>
>>

-- 
Nat Sakimura (=nat)
Chairman, OpenID Foundation
http://nat.sakimura.org/
@_nat_en
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-ab/attachments/20130513/dbb346dc/attachment.html>