I highly doubt it.
I’m pretty sure typical harassment comes in the form of many similar messages by many different users joining a bandwagon. Moderation wouldn’t really be fast enough to stop that; indeed, Twitter’s current moderation scheme isn’t fast enough to stop it. But the current scheme is capable of penalizing people after the fact, particularly the organizer(s) of the bandwagon, and that creates some level of deterrence. An opt-out moderation scheme would be less effective as a deterrent, since the type of political influencers that tend to be involved in these things could likely easily convince their followers to opt out.
That may be a cost worth paying for the sake of free speech. But don’t expect it to make the anti-harassment side happy.
That said, it’s not like that side can only tolerate (what this post terms as) censorship. On the contrary, they seem to like Mastodon and its federated model. I do suspect that approach would not work as well at higher scale - not in a technical sense, but in terms of the ability to set and enforce norms across servers. But that’s total speculation, and I haven’t even used Mastodon myself…
This is also the approach celebrities in general need to take as they get drowned in messages (elon musk could spend 24/7 reading messages sent to him and read only a tiny fraction), so harassment can probably be solved by whatever solution we come up with for celebs.
Can make some fairly elaborate "allow" rules depending on why you might want to read messages from non-contacts like "this person is a contact of a contact" or "this person has "IT" in their Twitter bio or "this person is on the 'good person' white list that Mr. Whitelist maintains for the community".