> My opinion is that this should work on a per-content basis, where the platform only becomes legally responsible for a unit of content once it associates itsel

> My opinion is that this should work on a per-content basis, where the platform only becomes legally responsible for a unit of content once it associates itself with it by promoting, modifying, etc. this content. Such ideas are certainly worth discussing, but I don't see how it would work. If the platform is suppressing some items then it is in effect promoting all the items it doesn't suppress, so would be responsible for them. The only way to avoid responsibility would be to present everything to the user, including the most blatant spam, in no particular order. Or you could allow certain kinds of "curation", perhaps, but if the curation is anything other than filtering by application of a published algorithm, then probably the platform would have to take responsibility for anything that doesn't get filtered out. But filtering with a published algorithm wouldn't stop professional spammers. Here's a random idea: delegate the filtering to lots of separate smaller companies which would be a less juicy target for lawyers and it wouldn't matter if they went bankrupt every now and then.