Obviously if you generate semantic HTML there is no problem. But the article argues (in the end) that accessibility instead can be achieved by generating different "object-html" renderings on the server based on e.g a header that says "i am blind". That is totally unrealistic!
Semantic HTML on the other hand is accessible by default, you have to specifically go out of your way to make it inaccessible or platform-specific.
But you also have to go way out of your way to make it look good. And with pure semantic HTML there are layouts that are flat out impossible. The only way to do certain things with CSS is by cluttering one's clean HTML with random divs everywhere (killing the markup/style separation in the process)...
Where the author suggested outputting audio when confronted with a "blind" header, I thought more along the lines of returning pure semantic markup. And presumably the abstract code that compiles down to html is purely semantic so it would be trivial.
I think HTML is intermediate code, not object code, not source code.
Just like you could code assembly by hand (and plenty of people still do) you can code HTML by hand. But in reality most html is generated by machine and then 'rendered' into a visual representation, the final binary end-product.
Moreover, if you're going to argue the "nobody's actually going to do it" argument, I'd say that "nobody's actually using CSS to generate one HTML page that works in a screen reader, cell phone, rich browser, and REST API" either. For the same value of "nobody", which is to say, not literally nobody, you can find a handful of people, but the idea that there's some mass of developers "correctly" using CSS is pretty silly too.