I remember a fair while back there being a distinct mismatch between the theory and the reality on the ground. Hardly anyone was actually talking to their intended audience. There was a lot of box ticking.

But unfortunately it’s not worth pointing this out, or daring to think of better ways. I’d note your comment is greyed out, and mine suggestion (https://news.ycombinator.com/item?id=16035056) is way into negative territory.

Better to just check boxes I guess.

Perhaps putting the HTML standard on a diet would be a more efficient approach?

I'm a perfectly able person, yet I've had to roll a number of bookmarklets just to make most sites halfway readable. And Jesus help you if you have to rely on a screen reader for the crazy sites we have today.

I think this lady was spot-on in her article, but everyone wants so much control that every website ends up reinventing the wheel, badly.

Screenshot the webpage -> Neural Network to segment the page and decide what is the navigation -> text extraction from image, with corrections from (rendered) HTML source.

All the parts are there and work better than existing screen readers.

You seem to be overestimating how close to practicality neural-network approaches are for most problems. Very much of the potential application space is still firmly within the domain of research. That doesn't mean what you described is completely impossible, just that you could get a PhD for pulling it off.

[1] https://arxiv.org/abs/1609.04938v1 (incidentally, I noticed that the version of the paper that was accepted for ICML 2017 no longer has the HTML part)

The exciting thing is that in this case it doesn't need to be. Segmentation (which is solved) is sufficient to identify semantically separate parts of the page, and then OCR can be used to place the text from the html on the representation of the page.

I expect that to be significantly different from the task of segmenting objects in a photo, because most of the boundaries are not revealed by texture differences, but implicit in the arrangement.

https://arxiv.org/abs/1704.01474

https://arxiv.org/abs/1711.07695

https://arxiv.org/abs/1710.07363

I'm pretty sure I've seen some webpage work too, but I don't have the reference.

Finding the text in a screenshot of a webpage should be easier than doing it for handwritten documents; it's the "decide what is the navigation" part that I'm doubtful about.

But that's beside the point. Your comment is like saying "They have powered exosuits now, so why do we still need to put ramps on buildings for people in wheelchairs?"

I can only hope that your level of apathy and contempt is not shared by more web designers.

Identifying a better solution to a problem is a long way from apathy and contempt. In fact it is how things improve.

Far from your exosuits vs wheelchair comment, this is more like saying that wheelchairs are good enough.

Legs are billions of years old and wheels where invented thousands of years ago. Yet we still use them were they remain most effective. No doubt there is room to improve mobility options, but cost and battery capacity are very real constraints. And sometimes mimicking nature too closely is less effective; i.e. planes without flapping wings.

All that to say the web was born more accessible than magazines. It's probably more effective to maintain that semantic heritage than make machines that can translate flattened images like an interpreter. Still, experimentation is great. Nothing wrong with trying to make the machine. I just wouldn't advise that the whole web forget accessibility because the tools _should_ already exist.

EDIT: fixed typo

That reaction is pretty disappointing really. It's the "crutches are fine, why should they want artificial limbs" that is really kind of insulting.

Words like "apathy and contempt" do not seem to be "just trying to acknowledge that artificial limbs". Instead they are directly saying that people who hope for artificial limbs are apathetic and have contempt for the problem.

In-fact the complete opposite is true.

When the tech is both widespread and affordable, you can start designing for the tech.

Don't even understand what bothers you so much about making pages more screen-reader-accessible in the first place. Great, I'm glad that you have experience in this area and can attest to the fact that screen readers don't always work even when HTML is semantic. But why are you actively resisting semantic HTML?

It sounds to me like you're saying, "It doesn't work well, so we shouldn't usw it at all." What would you prefer, div soup?

It doesn't even need to be the case that semantic HTML is a killer solution for screen reader users. Just needs to be better than nothing once you net costs and benefits.

So I'm genuinely curious: when is it better to not use semantic markup? Or when does the cost of using it become prohibitive?

So I'm genuinely curious: when is it better to not use semantic markup?

It never is - but nevertheless it happens all the time. Perhaps you hadn't noticed it was exactly what this article was about?