@dannyman I mean, it does to an extent. CSS has the color-scheme and light-dark() colors that respond to the browser’s settings.

But the browser can only do so much. CSS doesn’t convey the *intent* of elements, only their *appearance*. It’s much worse with React-style pages, in which the CSS *and* html elements have basically zero semantic meaning.

I use Reader mode to get around dark mode, which works sometimes for sites with mostly static blog contents, but a wholesale rewriting of the page style basically requires reading the minds of the developers.

@drahardja so ... HTML is fundamentally semantic ... there are a bunch of words that you want to read and some markup explaining how the author wants it to render ... but what happens on the clients side is fundamentally up to the client ... to represent the words ... that's why blind folks can browse the web with screen readers -- it's just words -- and you ought to be able to access a web browser that can cater to your own abilities as well.

This is more reliable than asking millions of random web sites to somehow understand every nuance of crafting CSS for each audience.

I believe the mainstream operating systems and web browsers already have a variety of accessibility settings, each frustratingly limited in its own way. We all know how to crank the font size up and down!

(Personally, I tend to avoid web sites that DON'T support a dark mode, which I find easier to read, but not to a degree that I need an accommodation. Kagi slow web has a neat switch for rendering random sites in light/dark mode.)

@dannyman HTML *can* be semantic, but only if the web developer cared. Many websites now use an intermediary framework like React that renders everything as divs, without any further meaning.

Some websites are considerate enough to use semantic tags like <article> but this is by no means universal, or even widespread in my experience.