Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Web accessibility for 2019 (sourcerer.io)
64 points by daftpanda on Oct 22, 2018 | hide | past | favorite | 11 comments


The author describes what he purports to be a simple, universal language for literally all content. Everything, he says, can be described and explained in terms of "role, states, attributes, relations, and actions."

> Does your web app use chemistry ring formulas or math equations or anything else from any other subject you can think of? So, you can definitely describe all those symbols and formulas to your friend, so your friend can understand those, right? Thus, you should be able to express those in accessibility concepts as well. You may want to practice on your own to see how it goes.

But why those five concepts in particular? Philosophers have been attempting to identify a set of universal groups for human concepts since the ancient Greeks, (it's called "ontology") and I see no reason to think that the author happens to have solved this problem.

For example, what's the difference between a state and an attribute? Is an action a relation between the subject and its object? If an object can change roles, is that a state change? Or is it a change in the relation of the object to its role? Is a role change just an action?

All of the author's proposed accessibility concepts mush together when you think about them critically.

In contrast, WAI-ARIA, taking the problem of taxonomy seriously, did not attempt to define five core accessibility concepts and apply them to all of human experience. They have a zillion different roles, developed more or less organically, to address specific UI use cases.

ARIA's complexity isn't just an unfortunate fact of history; this is literally the best that humanity can do. We can't simplify the world into five kinds of concepts, and it's not worth trying.


Those five concepts are not random, they go from the assistive technologies world: most if not all desktop APIs use these terms.

Agreed, the state concept (or even role concept) can be technically mapped into attributes, but I keep using them just to stick closer to existing terminology.

When it comes to ARIA, all browsers map ARIA taxonomies into desktop APIs, and thus I think it's fair to say ARIA is mapped into those concepts as well.

But this post wasn't an attempt to define accessibility concepts as such. It was rather about answering a question, whether a large set of accessibility problems can be solved, if the web came with a technology that provides greater access to express semantics to web devs.


No, those problems can’t be solved this way. Your taxonomy may seem more “expressive” to you, but your five concepts are too vague to build a useful screen-reader around. Web developers won’t use the concepts consistently; screen readers won’t interpret them consistently.

None of the relevant vendors are on board, because they’ve all considered a simpler taxonomy and rejected the idea, in favor of a richer, more complex system that can actually be implemented.


Having these concepts exposed to the web doesn't repeal the necessary of standardization, i.e. both web developers and AT vendors has to stay in agreement on content semantics they operate with, but it will make the agreement process simpler than it's now, because it doesn't require implementation from browsers for each and every new feature. Also it will enable experimenting on the web, where web developers and AT vendors can try new things before stuffing them into standards, which should boost the accessible web.


I think people have been moving away from the semantic organization because trying to properly dimensionalize the web has been so fraught.


Not so sure ARIA is the most practical answer since it's so big and full of edge cases.

Still, the promise of a more text descriptive web is exciting.

IME making JS optional, putting text descriptions, and preserving keyboard access to all features is much lower hanging fruit.


Agreed, however I bet making a website JSless may be a challenge in some cases.

So I'd be cool to have something more powerful and flexible than ARIA is to not limit web devs in choosing the techs.


Taking an existing website and removing Javascript may well be a challenge.

But building a website and simply not adding unnecessary Javascript in the first place? Far less of a challenge.


I doubt browser vendors can "fix" things for the more complex media types, because they tend to have no straightforward answer.

What you tend to see with video games is that people appreciate games that have a lot of configuration and control options. That let's the user figure out something that works for them. Unfortunately, that's not going to be expressible with some metadata like HTML attributes.


I would love to have some examples. I believe there are cases which don't make a perfect fit for metadata mappings, but keeping it on radar could help to evolve techs right way and address such cases.


Here's a relatively recent thread with Ubisoft asking for help to improve accessibility: https://www.reddit.com/r/disabledgamers/comments/8kf9bx/ubis...

The consistent theme I see is being able to change the controls, which is often game specific, making it difficult for the browser to help out.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: