edit: grammar
The only practical way to stop this is to employ an army of censors to read private conversations and delete the ones deemed bad. But of course this is also horrible and dystopian. So, as always, it’s trivial to write a scathing critique of Facebook no matter what they do.
The problem is that hate doesn’t need an algorithm to amplify it; it can just amplify itself, as it has throughout all of human history. The problem isn’t the algorithm, it’s the people. And you can’t just “fix” people.
Then typical user Facebook feed will be 99% pictures of funny cats from one of their loved friends, which are too good to unfriend (if that friend is, for example, mother).
(Not to mention, it's practically impossible to configure filtering for 99% users).
And then these users will just leave the service because it's boring, and that's it, end of Facebook.
We already had a lot of services with no feed ranking. All of them died or started doing the ranking.
> Facebook does exercise editorial control
As any other mainstream service with feed and user-generated content. For example, Twitter or TikTok.
> It's fair to ask that this curation consider other objectives as well
The big neural network which defines feed ranking is developed by hundreds of engineers, and already optimized for (I guess) hundreds of parameters, including time spend of website, ads revenue, fraud protection, lower risk of suicide and so on.
But Facebook does exercise editorial control, in the service of engagement. It's fair to ask that this curation consider other objectives as well, or at least counterbalance the side-effects it's known to have (divisive content is more engaging and so is amplified; at least correct it back down to neutral).