Unfortunately, some apps don't support text selection and on some websites the text selection is unpredictable.

I'd actually compare screen OCR to screenshots. Instead of every app and every website implementing their own screenshot functionality, the system provides one for you.

Same goes for text selection. Instead of every context having to agree on tagging the text and directions, your phone has a quick way of letting you scan the screen for text.

To be fair, I still use the "hold the text to select it" approach when I want to continue with the "select all" action and have some confidence that is going to do what I want.

> some apps don't support text selection and on some websites the text selection is unpredictable.

That correctly identifies the problem. Now why is that, and how can we fix it?

It seems fixable; native GUI apps have COM bindings that can fairly reliably produce the text present in certain controls in the vast majority of cases. Web apps (and "desktop" apps that are actually web apps) have accessibility attributes and at least nominally the notion of separating document data from presentation. Now why do so few applications support text extraction via those channels? If the answer is "it's hard/easier not to", how can we make the right way easier than the wrong way?