950328 - provide feature to detect screen reader users

Status: RESOLVED WONTFIX

(Core :: Disability Access APIs, defect)

Description

[steve faulkner]

In Bug 946370 discussion has turned to author methods to detect screen readers, and there appears to be some support for this and a belief that it is already achievable using JavaScript (albeit not reliably). I believe that Firefox could provide a robust and reliable method to expose this info via its accessibility APIs.

Comment 1

[steve faulkner]

Indie UI spec provides a potential framework for this https://dvcs.w3.org/hg/IndieUI/raw-file/default/src/indie-ui-context.html#userScreenReaderSettings

Comment 2

[alexander :surkov (:asurkov)]

Steve, what the proposal is? (how different from bug 946370?)

Comment 3

[steve faulkner]

(In reply to alexander :surkov from comment #2)
> Steve, what the proposal is? (how different from bug 946370?)

Hi alex, the implication from the discussions on other bug is that javascript can already be used to detect screen readers(and that it is a good thing) so users can already be fingerprinted as disabled and therefore any expectation of privacy, in this regard, is an illusion. If that is so, why not expose a method via firefox to signal the presence of a screen reader? It would provide a simple robust flag rather than relying on the vagaries of the promoted javascript methods.

Comment 4

[alexander :surkov (:asurkov)]

Ok, got it. Iirc something like that was wanted by Google. I'm curious if they did any proposal on that.

Comment 5

[Trevor Saunders (:tbsaunde)]

(In reply to steve faulkner from comment #3)
> (In reply to alexander :surkov from comment #2)
> > Steve, what the proposal is? (how different from bug 946370?)
> 
> Hi alex, the implication from the discussions on other bug is that
> javascript can already be used to detect screen readers(and that it is a
> good thing) so users can already be fingerprinted as disabled and therefore
> any expectation of privacy, in this regard, is an illusion. If that is so,

I wouldn't say its a good thing, but it seems kind of inevitable.

> why not expose a method via firefox to signal the presence of a screen
> reader? It would provide a simple robust flag rather than relying on the
> vagaries of the promoted javascript methods.

While I could probably live with this right now it seems like most use cases for this could be handled with css media stuff, and imho layout / styling is the layer at which most of these changes should take place.  Also it makes it ever so slightly harder to do really evil things like XHR the screen reader running state somewhere instead of just change if a element is visible.

Comment 6

[alexander :surkov (:asurkov)]

(In reply to Trevor Saunders (:tbsaunde) from comment #5)

> > why not expose a method via firefox to signal the presence of a screen
> > reader? It would provide a simple robust flag rather than relying on the
> > vagaries of the promoted javascript methods.
> 
> While I could probably live with this right now it seems like most use cases
> for this could be handled with css media stuff, and imho layout / styling is
> the layer at which most of these changes should take place. 

iirc the point was to skip unperfromant ARIA part if no consumer

Comment 7

[Dylan Barrell]

As a proponent of AT detection, here is what I consider to be important:

1) Detection that a screen reader is running is useful but of limited use, I would argue that this information alone is only useful for creating text ghettos (which we do not want)
2) In order to do useful things, we need to know which assistive technology software and version is running so that if we know about bugs/limitations in that version, we can work around them
3) Better than 2 would be to also have a way to detect specific features so that we can implement progressive enhancement. I am currently submitting some comments to the IndieUI group that elaborate on this

Comment 8

[James Teh [:Jamie]]

We have no intention to implement or drive this at this time.