[OpenID] OpenID Accessibility

[tom calthrop]

Hi Lachlan,

The main issue was with the complexity of a typical trust screen when 
extensions that require user intervention are introduced into the 
interface. Go through a typical trust process using Fangs (Firefox 
extension which mimics a screen reader):

http://www.standards-schmandards.com/projects/fangs/

If you look away from the form and only read the screen reader output 
you are "closer" to understanding what a blind person hears when they 
interact with an OpenID authentication process. After many headaches 
with trying to simplify this we concluded that the only really easy 
approach was to "cancel" or "proceed" at trust and remove all extensions 
that require interface usage. I think I touched on it here:

http://www.barnraiser.org/openid_usability_part_2:_the_authentication_workflow

My feeling after spending several weeks on this (with Micheal who is 
blind) was that authentication should be super simple and that once 
authenticated a second process can ensue to (for instance) collect 
profile data and pre-fill a form with something as simple as a "pre-fill 
this form" button (goes to RP, asks if Consumer can have a key to your 
profile data once or for always... pre-fills form and you are back 
hearing a standard form being read out to you with pre-filled data).

Hope that helps,

Tom


Lachlan Hardy wrote:
>
>     We did some pretty extensive testing with Swedish Institute of
>     Assistive
>     Technology with our free OpenID server. We removed all AX/SREG support
>     which was the main issue when testing with blind people (apart
>     from the
>     obvious gotchas like captcha).
>
>
> I'm interested to know what the issues you discovered with AX/SREG 
> were. Did you publish the results at all? Or can you give us a quick 
> run down?