Hacker News Reader

Comment by idreyn | Hacker News Reader

Increasingly I think the screenreader/ARIA proposition of navigating a (linearized ordering of) elements on a 2D page is fundamentally limiting for blind users. The web needs a more robust model of application state from which text, audio, and GUI representations can be derived. Our applications clearly structure "objects" and their "methods" under the hood, and it should be possible to present these more directly to users in whatever format makes sense to them.

View on HN

It has to be linearized because hearing, and time itself, are one-dimensional.

Screen readers can represent nesting of content, going into one element and coming back out, Apple's VoiceOver in particular uses this concept. It may be conceptually helpful in some instances or to some people but often it just makes things more cumbersome to navigate.

Maybe I don't get what you want to express, but screen readers allow for alternative presentations of an HTML document, allowing for alternative modes of content discovery, e.g. hierarchical outlines of headlines and "landmarks".

Several ARIA attributes further allow for communicating the current application state, e.g. whether content has changed (aria-live), a button is in a pressed state (aria-pressed), or a menu is opened (aria-expanded).

That would require some regulation, since commercial web design deliberately obfuscates the underlying structure.

I agree. For visually impaired users, HTML/sceenreaders are simply wasteful layers between them and the core behaviour/data of an application.

In an ideal world, we could have entirely different clients that focus on audio-touch interfaces like a some sort of keyboard driven Alexa skill.

Is that not the DOM?

Reply on news.ycombinator.com