It would be good to have both options available. Perhaps the perfect system design needs them both eventually anyway? I imagine you'd need to be able to differentiate between harmful-extreme negative content and simple-overvalued "boo" posts.