I agree with the ironic sentiment. We've pushed up the complexity of a lot of features because we can, not because the value is there. Unicode itself is a great example of such: it's a dumping ground for "everything that resembles text" - and while it's easy to consume and there is some benefit to be had, it's also hard to comprehend on any level beyond the very basics. It's quite a ways away from the early telegraphy encoding systems.
With respect to selection, selection is key to all forms of interactive editing because without it you don't have interactivity, you have a linear workflow. As such it does deserve some respect as a fundamental feature, even if it's not critical to fix every last bug. Even if you are working via screen reading and voice to text, you can still benefit from having selection tools available to you.
If your goal is, for example, to edit a book, you often need much more than a typewriter can give you. You need figures, equations, centralized text, footnotes, rich text styles, tables.
If your goal is even more ambitious, say, publishing the same book online, such that others can use the contents of the book and select it, look it up, share it, edit it etc., you need many of the above to be understood by the text engine, so you end up with shaping, support for all human scripts with all their accumulated idiosyncrasies etc.
Without any of this, you can barely have an academic life for example on the internet,or engineering disciplines and so on.
Also, original point about typesetting stands, even in Thai case.
Also i find it ironic that we care a lot about let's say caret movement, when whole classes of devices are unavailable for the blind.