FB and others could also do as you said - make quarantine zones for specific prejudicial groups and their ideologies.
Except time and again these groups have proven they're not wanting to stay in their own special self-moderated zones but are actively looking to broadcast. There are businesses built upon these prejudicial world views and deplatforming them just shows how they heavily they rely on broadcasting:
https://mashable.com/article/milo-yiannopoulos-deplatforming...
https://www.wired.com/2010/11/facebook-link-blocking/
https://444.hu/2016/11/14/facebook-is-reading-your-private-m...
(The intended receiver can be frustrated or confused that they couldn't receive the message their friend wanted to send.)
And as you typically have to opt-in to messaging with someone new, they could wait for complaints rather than have content-based prior restraint. And, react to specific abuse with post-action blocking of the harasser, or quarantining-the-harasser as someone who'd need extra explicit acceptance before their messages are delivered.
And, rather than trying to square-the-circle of global speech standards enforced by crude algorithms and rushed, underinformed low-wage moderators, they could make the remedy for unwanted speech simply unsubscribing/unfriending the source. And, for users who feel disturbed by broadcasts, FB could offer opt-in moderation circles – leaving all willing-listeners, with no grievances, unmolested by 3rd-party censorship.
Then, willing-speakers sending legal speech to willing-listeners wouldn't be collateral damage of their clumsy improvised censorship regimes.