Let's think one level of abstraction higher - do we want to marginalize voices in the first place, given that some of them are fighting oppression and injustice

Let's think one level of abstraction higher - do we want to marginalize voices in the first place, given that some of them are fighting oppression and injustice, and others are advocating for bigotry and hate? That is, if we cannot effectively discriminate between those two types of voices, whether because "moderation doesn't scale" or any other valid technical reason, or because it's not always possible to disentangle "legitimate" from "illegitimate" marginalized voices. It's not a binary category, because people who advocate for stuff are not one-dimensional caricatures. Do we accept the bad in the name of $good_stuff_that_comes_with_making_marginalized_voices_heard, or do we marginalize it all straight into oblivion by going after the forums of spreading ideas? The answer to this question seems almost implicit for you, but it's far from obvious to me.