In Lunney (1999), the court made clear that even when a company chooses to exercise editorial control over some of the content produced by its users, it continues to have no liability with regard to any of the other content that it helps to publish. Although algorithmic recommendations were not at issue in Lunney, it’s hard to imagine that such would matter, given the court’s hostility not only to holding service providers liable for harm they cannot specifically predict, but also to imposing a duty to investigate the conduct of customers.

Even if a plaintiff could adequately allege recommendation conduct by a social media outlet that gives rise to liability (still have to ignore section 230), it’s unnerving to see this opinion give the impression that, under New York law, publishing services are subject to the same product liability standards as any other service.

Even the most damning allegations about YouTube’s algorithm, drawn from cherry-picked studies and expert interviews, would clearly establish that YouTube viewers train their own instance of that recommendation algorithm. That is, even as alleged, the algorithm only continues to present extremist content to users who actually view it when offered.

I could spend days picking apart the tortured logic of the complaint, but it all boils down to the same thing: the plaintiffs concede there is nothing unlawful about the content or its hosting per se, but insist that making it easy to find is a dangerous activity that must be undertaken with care. They pretend the freedom to publish somehow does not include helping people find what was published.