What do we do? We treat platforms with algorithmic news feeds as publishers not platforms in the Section 230 sense.

Think about it this way: imagine if you took a million random posts or videos. You would find a wide range of political views, conspiracy theories and so on. Whatever your position on any of those issues, you could find content pushing those views.

So if your algorithm selects and distributes content that fits your desired views and suppresses content that opposes your views, how are you different from a random publisher who posts content with those exact same views?

This is kind of like the "secret third thing" of Section 230 where you get all the protections of being a platform and all the flexibility of being a publisher and we need to close that loophole. Let platforms choose which one they are.

Another example: if I create a blog and write a post that accuses my local mayor of being a drug addict and a pedophile, I can be sued for defamation. You can try the journalism defense but it won't shield you from defamation. Traditoinal media outlets are normally very careful about what they publish for this reason.

But what if I run Facebook or Twitter and one of my users says the exact same thing? Well I'm just a platform. I have a libel shield. But again, my algorithm can promote or suppress that claim. Even if I have processes to moderate that content, either by responding to a court order to take it down and/or allowing users to flag it and then take it down myself with human or AI moderation, the damage can't really be rolled back.

We've let tech companies get away with "the algorithm" being some kind of mysterious and neutral black box that just does stuff and we have no idea what. It's complete bullshit. Every behavior of such an algorithm reflects a choice made by people, period. And we need to start treating this as publishing.