I think the fundamental tension here is that platforms want the promote material to optimize engagement metrics, and people want to benefit from that boosting, but platforms don't want to boost anything that's too bad. They don't just host content, they act as promoters and editors.

Nobody gives a shit that racists have email newsletters (I assume they do?). I mean, people might, in the abstract, rather they didn't, but nobody's really trying to stop them or say they can't do that. Google adds a "what's trending in email" feed to gmail and starts promoting those on there, however, and a whole lot of people (and advertisers) might ask Google to knock that shit off. And... like, rightly so?

> Nobody gives a shit that racists have email newsletters (I assume they do?).

Maybe not racists but a lot of people definitely do care if folks trying to manufacture WMDs or refine uranium/plutnoium have email newsletters.

This particular concern applies to every other medium, online and off, since the 1940s, and doesn't seem to have been a major problem in practice. More broadly, law enforcement opposing criminal conspiracies has kinda always been a thing. It doesn't seem, in practice, to have caused any major problems for people who want to exchange messages or publish newsletters (physical ones, or email) about how taxes suck, or even about how some ethnicity is inherently bad.

>... and doesn't seem to have been a major problem in practice.

Based on what reasoning?

> And where do you draw the line between merely an app or website and a digital "public square"? Certainly Facebook or Twitter seem like the latter by virtue of their size and reach, but what about Hacker News? Mastodon instances? The Something Awful forums?

I favor the framing "the internet as a whole is the public square" over framings such as "this social media company is or is not a public square". A particular social media company has no inherent social obligation to host a diverse range of opinions. Governments cannot tell social media companies to moderate in a certain way, but social media companies should be able to moderate however they want to based on just about any motivations.