What is not rare is the insertion between strong and weak directions. What is rare is the insertion between two characters with opposite strong directions.
I think it's appropriate to maintain this distinction for the practical case, and the expected user-friendly behavior of editors when editing text notably at the weak position bordered by a weak SPACE or start/end of a paragraph, for which the character with the strongest direction should still win: It is arguably, by very far, the most frequent case, even if "rare" does not mean a precise statistic that you won't encounter this case everyday (and that most users won't care about when they rarely type text using alternanace of directions without at least a weak space separation). I also counted the case of strong Perso-Arabic digits, which behave a bit differently from Euro-Arabic digits which are weaker (so they loose if they are sticked with Arabic letters which should still dictate the expected direction if insertion/deletion without ambiguity). If there are Perso-Arabic digits sticked with Arabic letters, there's no ambiguity, the whole is strongly RTL and a single caret is also enough (even if you may still indicate the direction of this single caret with an arrow head) The appearance of dual position carets should be limited as most as possible to only the very ambiguous cases like "latinARABIC" or "ARABIClatin" where the insertion point is between the two strong scripts, and that should then be displayed (respectively) not as latin|CIBARA CIBARA|latin but as latin_^CIBARA ^CIBARAlatin_ (here again, "_" is the right-side insertion caret pointing to the right, and "^" is the left-side insertion caret pointing to the left)
