[nailer]

Avoid aria tags. The spec is unworkable (see this document) the browsers made by the disability industry extract vast quantities of money from disabled people with little effectiveness because they try and boil the ocean which unsurprisingly is ineffective.

Support efforts for computer vision based browsers, MCP and APIs.

[jraph]

> MCP

Respectfully screw making users rely on AI for accessibility. Just make the damn page accessible already. Actually, more like make sure you don't break the accessibility that's there by default with correctly written plain HTML.

[nailer]

> Respectfully screw making users rely on AI for accessibility.

Why? It's the right tool for the job.

> Just make the damn page accessible already.

Oh so just modify every website and expect the disabled people to wait while this happens?

This disabled web browser industry doesn't care about disabled people. Their solutions don't work, disabled browsers are expensive because government grants are given to purchase them.

[jakelazaroff]

> Why? It's the right tool for the job.

No, it's not. Why should disabled users be forced to indirectly interact with a webpage via a non-deterministic agent, rather than directly interact with one that's specifically designed to accommodate them?

[nailer]

> rather than directly interact with one that's specifically designed to accommodate them?

Because a world where that happens consistently doesn't exist, it hasn't existed for the last 20 years we've been using ARIA tags, and won't ever exist.

[jakelazaroff]

Your advice to "avoid aria tags" would make that a self-fulfilling prophecy. The ways to make it happen:

1. A robust set of web primitives that are accessible by default, and

2. A government that will actually enforce laws (which already exist!) requiring websites to be accessible

[nailer]

> Your advice to "avoid aria tags" would make that a self-fulfilling prophecy.

As mentioned ARIA has had 20 years to succeed before my Hacker News post. ARIA will continue to fail with or without me.

[jraph]

For a user running into broken pages, sure, you have to compose with what you have.

As a developer, however, get your shit fixed! And that fixing doesn't involve any MCP. Don't expect visitors to run AI...

[nailer]

> Don't expect visitors to run AI...

Why not? That's a more reasonable expectation than asking every website to change.

[jraph]

Because ethical issues aside, that's expensive to run, unreliable, non-deterministic hot mess that doesn't have predictable behavior. And predictable behavior is somewhat key.

I won't blame a disabled user seeking AI-based tools to browse the web to survive with what they have, but I will totally blame the devs who created the situation.

[nailer]

A user interpreting visual meaning is vastly more reliable than expecting all the websites the user interacts with to be ARIA compliant.

[RobMurray]

To hell with using vision based AI for web accessibility. it really isn't that hard to get right. Semantic html is already accessible. ARIA can help when devs want to use the wrong elements for some reason or for custom controls.

[nailer]

> it really isn't that hard to get right.

Yes you just need every website to use it, rather than fixing the client. Which is the 'boil the ocean' strategy mentioned in the comment you're replying to.

> ARIA can help when devs want to use the wrong elements for some reason or for custom controls.

But it can't. See this article.

[rhdunn]

See https://www.w3.org/WAI/ARIA/apg/patterns/ for a guide on how to create accessible markup for custom controls and the associated examples.

See specifically https://www.w3.org/WAI/ARIA/apg/practices/names-and-descript... for details on naming. That has extensive notes and details for labeling elements correctly.

See https://getbootstrap.com/docs/5.0/components/ for bootstrap markup on creating accessible components.

There are plenty of other resources.

[nailer]

I didn't ask for resources on ARIA, are you replying to another comment?

[rhdunn]

You said "see this article" re: how aria-label is not applicable to div elements, hence the second link which is the WAI-ARIA guide on labelling elements.

You also said that ARIA can't help with custom controls in that post, which is where the other links are applicable as they describe doing just that. I.e. using ARIA tags to implement tabs, accordions, etc. either with or without a framework library.

[nailer]

> You also said that ARIA can't help with custom controls in that post

I didn't write the post. The author believes in ARIA, I believe ARIA is fundamentally broken.

[RobMurray]

I did read the article. Why do you need to label a div? It's just a container, not a semantic element. If you really want to use a div for something semantic you can set role and aria-label. That is done all the time and works with screen readers.

[ramblurr]

What document?

Do you have any sources to back these claims up?

[nailer]

> What document?

https://news.ycombinator.com/item?id=48237159

> Do you have any sources to back these claims up?

Yes, asides from the article, check the prices of browsers from the disability industry and consider for yourself whether it's logically easier to fix every website or make a client that can adapt existing webpages.

[RobMurray]

Clients can't automatically fix all existing web pages, because the semantic information just isn't there. AI doesn't excuse web developers. It wouldn't even be a fix. Who wants to wait for an AI agent for each interaction?

Not all accessibility tools are expensive:

   - NVDA is free and open source
   - Narrator is included with windows
   - Voiceover is included with macOS and iOS
   - Orca is free and open source.
   - Talkback comes with Android
   - Chromevox comes with Chrome OS

[nailer]

> Clients can't automatically fix all existing web pages, because the semantic information just isn't there

Yes it is. Webpages semantically provide information on the purpose of onscreen elements by their appearance. Vast quantities of humans ensure the semantic information is there by using the websites. Websites that do not convery semantic information through their appeareance will die, because nobody is using them.

> Not all accessibility tools are expensive

I want to acknowledge your point here - the situation may have improved the last time I looked at accessibility tools, which at the time was mainly overpriced badly maintained proprietary software. I still think the "boil the ocean" strategy is discredited and wrong.

[RobMurray]

I'm glad that a large proportion of web developers are happy to boil the ocean then. I use the web every day with a screen reader. It works. 99% of what I want to access is fine. Often not perfect, but a whole lot better than what I could get out of AI.

I do use ChatGPT to research things, but I don't usually see that as accessibility solution. I completely agree that screen readers and browsers would benefit from AI, as they already are; Chrome and Edge can generate missing image descriptions. AI can certainly enhance accessibility, but it can't replace the existing technology that already works quite well a lot of the time.

The other positive about AI for accessibility is that the frontier models have a good understanding of what works. Instead of learning all the guidelines, you can just ask an agent to review the page for accessibility and fix any problems.

I realise I am only looking at it from a screen reader point of view, and yes, we are quite a small minority. But good universal design helps everyone, whether they just need to zoom the page for comfort without it breaking, their eyes are not that of a young person, they have dexterity issues using a mouse, and many more. Accessibility in general isn't serving a tiny minority,. I imagine most of us will come to appreciate it in some way.

[nailer]

> I'm glad that a large proportion of web developers are happy to boil the ocean then.

That’s not how that expression works. The ocean is every website. An individual developer cannot modify every website.

More to the point, most websites are not accessible, will not validate, and this will never be the case.

[rhdunn]

NVDA is a free screen reader for Windows (written by blind devs) that works with Firefox and Chrome.

You don't need to pay for a specialist browser as all web browsers (Firefox, Chrome, Edge, Safari, etc.) will implement the native accessibility model of the operating system they are running on (IAccessible/MSAA for Windows, etc.).

In Firefox you can press the right mouse button and select "Inspect Accessibility Properties" or select the "Accessibility" tab from the developer window and it will show the accessibility tree (roles, states, properties, etc.) just like the DOM tree in the "Inspect" tab. That is what the browser is displaying to screen readers and other accessibility software and uses the behaviour of the HTML elements along with the ARIA roles/states/properties defined by the webpage to construct that tree. Thus, it will display an ol/ul as a `role=list` unless overridden to be e.g. a `tablist` by the website.

See https://www.w3.org/TR/wai-aria-implementation/ for a specification on how browsers should implement HTML and ARIA to different operating system accessibility APIs.