There is nothing wrong with a descriptive approach to spoken language, but what I see you arguing is that written language should be treated the same way. This increases complexity in written language for no reason other than to protect mistakes in literacy.

There's real value in preserving spelling (it often contains etymologically relevant information to the current or past meaning of the word) and also grammatical structure. If the sound of two samples is indistinguishable, why make it harder to teach or to infer meaning from by accepting spurious representations as correct?

When you write it down, you gotta follow the rules, yo.