[OpenID] Facebook support for OpenID. Where?

Wed May 20 15:46:04 UTC 2009

Thanks for the constructive feedback, Andre.

This is the first iteration of our OpenID relying party. I agree that some of the interface has rough edges, but imho it works. We are working on testing different approaches to help set expectations and make an even smoother flow.

I would consider most of the implementation to be "work in progress". At Facebook, we work very iteratively- push something out, see how it's used, tweak, push, etc. Changes to the main home page are especially important to test extensively.

Some earlier posters asked about why we only use GMail for registration so far. We chose to start with GMail bc Google has made available both contacts and a verified email, both of which help improve conversion. We will work on testing other providers, and if we see a positive impact on growth, then we will launch them as well. It's very much a data-driven business decision.

Ultimately I would love to get OpenID to the point where it makes business sense to accept signups from all providers on the front page, but frankly, for our business, we're not quite there yet.

We're keeping track of open issues in the forums and on the developer wiki. I'm open to any constructive feedback and ideas.

----- Original Message -----
From: general-bounces at openid.net <general-bounces at openid.net>
To: general at openid.net <general at openid.net>
Sent: Wed May 20 08:29:58 2009
Subject: Re: [OpenID] Facebook support for OpenID. Where?

Just want to throw these 2 cents into the discussion..

The fact that, if I'm logged into my provider, I'm simply swept off my
feet and dragged into my facebook account feels very unnatural. In an
exclusively usability perspective, the action should *always* start by
user initiative. At least there should be a prompt suggesting "We've
detected you're logged into ___insert__provider___. <button>Come on
in</button>".

Have they addressed this? Is this behavior only temporary?

Thanks,
--
André Luís

On Wed, May 20, 2009 at 4:10 PM, Peter Williams <pwilliams at rapattoni.com> wrote:
> "Most notably, you can now register for a Facebook account with your Gmail account, or can link an existing Facebook account with Gmail or other OpenID-participating services if they support automatic log-in."[http://www.silicon.com/retailandleisure/0,3800011842,39432536,00.htm]
>
>
> Im hearing that the RP performed a profiling act (the selection of those features of the protocol that suits the RP). That is ... the protocol can only now be used according to the trust/business/value/legal principles that drive the adopting party.
>
>
>
> We might ask, politically: is the implementation consistent with UCI - where users are in control?
>
>
>
> now since SSO tends to be all about the politics of cooperation, one might analyze what is it in the interworking environment that induced Facebook to make the choice they did, in profiling the standard?
>
> We COULD assume malice or greed, but that's really last year's (US) political scene. We might speculate that its something about what Facebook considers itself to be doing in the world that has driven the choice.
>
>
>
> Only if they preserve X, when putting profiling limits on openid choices, can they adopt the protocol. Now what is X?
>
>
>
> My guess is the X is openid was adoptable only when its deployed with "rp-centric" trust models.
>
>
>
> It seemed critical to Facebook that user was not being spoofed/phished (for presevation of the FaceBook brand). It seemed critical that a FaceBook introduction to a (user's choice of) OP also not be facilitating spoofing/phishing on those OPs (for the protection of FaceBook's indirect reputation = brand). Hence, impose rp-centric trust models in which any reliance on an OP is contingent on the user having PREVIOUSLY logged into the OP (and therefore having already taken decision prior to a Facebook introduction on whether one has or has not been phished etc).
>
>
>
> I find this quite UCI, is a perverse sort of way. It would be also a legally-driven, reputation-driven profiling choice - in the absence of a trust model. It could even be seen as quite clever. Rather than extending openid with extensions (e.g the PKI-style extensions of some proposal), they went the other direction, and imposed profiling limits.
>
>
> ________________________________________
> From: general-bounces at openid.net [general-bounces at openid.net] On Behalf Of SitG Admin [sysadmin at shadowsinthegarden.com]
> Sent: Wednesday, May 20, 2009 12:21 AM
> To: Santosh Rajan
> Cc: general at openid.net
> Subject: Re: [OpenID] Facebook support for OpenID. Where?
>
>>You may not agree with my views, or I may not succeed to
>>convince you, but I thing it is unfair for you to suggest that I should not
>>express my views here.
>
> That wasn't his suggestion. What grows tiresome about your expression
> of those views is how you try to establish links between your beliefs
> and whatever may be happening at the moment, apparently without
> regard for showing any actual connection. Over time, this creates the
> appearance that you are either so frantic to make us listen, you care
> less about being relevant or valid than repeating the same points
> over and over, or so fanatical about your beliefs that you can't
> understand the different perspectives people here might have - much
> less, adapt your approach to fit into what *they* want, instead of
> just what *you* want (and think others, if intelligent, *should*
> want).
>
> In your original post on this thread you suggested that because this
> is "Facebook", we say good things no matter what they do. If the
> implementation Facebook had taken to OpenID followed your advice,
> would you still be criticizing them? Still be placing (some of) the
> blame for their (alleged) Fail on not doing so? If they had, would
> you be praising them, instead as an example of what everyone else
> *ought* to do?
>
>>Also sites that use user names as logins can easily integrate OpenID.
>
> Usernames usually have a character limit (say, 16 or so), but most
> OpenID's are (much) longer than that; already, this makes OpenID
> integration *very* difficult. How much room will you allocate for
> primary fields in your backend database? If that room has already
> been allocated, and the structure decided on, how drastically will
> the entire datacenter have to be overhauled?
>
> I spent a while worrying over maximum length of URI's (theoretical
> maximum limit of URL's, and this is limited by a combination of
> server at OP, server at RP, and user's browser; probably 4,000+
> characters!), but eventually decided to store primary keys of a
> hash's length at most; now, hashes *can* collide, but if you get more
> than one result you just retrieve them all and do more exact
> comparisons on the fuller string!
>
>>The problem is for sites that "use" email addresses as Identities, or
>>require verified email addresses. Here implementing OpenID in the current
>>form is not practical without including email addresses as identifiers.
>
> Poorly-hidden secret of database efficiency: you get *lots faster*
> lookups if you organize by numeric primary keys. Sure, the topology
> on paper will have username or some other field shown as the main
> index, but your *actual* topology doesn't have to match that
> perfectly.
>
>>And to make OpenID truly "universal", we need to somehow include email
>>addresses into the scheme of things.
>
> To make OpenID *truly* universal, it would have to be compatible with
> irc:// and all the rest. That might be a good place to focus your
> efforts (and, of course, I have to mention XRI).
>
> -Shade has been advocating for privacy in OpenID for over a year, but
> was never told to get off the privacy soapbox
> _______________________________________________
> general mailing list
> general at openid.net
> http://openid.net/mailman/listinfo/general
> _______________________________________________
> general mailing list
> general at openid.net
> http://openid.net/mailman/listinfo/general
>
_______________________________________________
general mailing list
general at openid.net
http://openid.net/mailman/listinfo/general
-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 7461 bytes
Desc: not available
URL: <http://lists.openid.net/pipermail/openid-general/attachments/20090520/41d16460/attachment-0002.bin>