This is a really good article in its summary of the problem. It's making the negative argument that fake news is very different:
> Fake news, on the other hand, is almost always the opposite. You want to read that stuff. For example, Casey Newton pointed to this study in his Interface newsletter that says some of the “fake news” is even more engaging than the real news.
This is it. The problem is Bob going around posting "Alice is the antichrist" everywhere on Facebook to one hundred million Charlies may eventually result in one of them murdering her. It will certainly result in Alice being harassed and limit her career. What protections does Alice have against this? Note that "just don't use Facebook" doesn't help Alice, because the problem is between Bob and Charlies.
There used to be people who argued that spamfighting, at all, in any form, was censorship. They have largely given up on that one, since nobody wanted the spam, but moved their dogmatism elsewhere.
Edit: there is a second-order problem in "recommendations". If Facebook (or for that matter Youtube) was just a chronological list of posts the only way to get more attention would be actual spamming - posting the same thing repeatedly to get into the user's very finite wall space. However, that's not how they work. Facebook (or a computer system wholly controlled by Facebook) selects certain posts you might want to see over others. If Facebook "recommends" the post saying "Alice is the antichrist", to what extent are they responsible for the abuse Alice receives? It's not obviously zero.
> Fake news, on the other hand, is almost always the opposite. You want to read that stuff. For example, Casey Newton pointed to this study in his Interface newsletter that says some of the “fake news” is even more engaging than the real news.
This is it. The problem is Bob going around posting "Alice is the antichrist" everywhere on Facebook to one hundred million Charlies may eventually result in one of them murdering her. It will certainly result in Alice being harassed and limit her career. What protections does Alice have against this? Note that "just don't use Facebook" doesn't help Alice, because the problem is between Bob and Charlies.
There used to be people who argued that spamfighting, at all, in any form, was censorship. They have largely given up on that one, since nobody wanted the spam, but moved their dogmatism elsewhere.
Edit: there is a second-order problem in "recommendations". If Facebook (or for that matter Youtube) was just a chronological list of posts the only way to get more attention would be actual spamming - posting the same thing repeatedly to get into the user's very finite wall space. However, that's not how they work. Facebook (or a computer system wholly controlled by Facebook) selects certain posts you might want to see over others. If Facebook "recommends" the post saying "Alice is the antichrist", to what extent are they responsible for the abuse Alice receives? It's not obviously zero.
(People considering liability for speech acts in murder cases may like to read https://en.wikipedia.org/wiki/Derek_Bentley_case )