“The conventional logic of the entertainment industry was let's make stuff that people like and want to see. And that didn't even have to be good. … Algorithmically oriented entertainment is less concerned about whether someone likes something than whether they will watch them at all. Pleasure is as useful a reaction as disgust, confusion, yet bafflement. … As long as it's provoking a reaction of some kind and compelling someone to keep watching, one strong reaction is as good as another.”

…often outrank like Netflix in the App Store, you know, on like phones. So there clearly is an audience of some kind. But, you know, that audience is, I think, in large part acquired just by spraying social media with advertisements. And that's a model that I think many are saying is unsustainable.

…errors racked up millions and millions of views. Does that mean that people liked it more than this really quality long read on a literary publication like The Baffler or something? Does that mean people liked it more? And I'm curious if you get the sense that this is like passive or active.

like I'm a really big animal person, and on certain meta algorithms, Meta has discovered that when they show me a picture of a suffering animal, I will stop and look at it. So they're like, oh, Emily likes images of suffering animals. So we're going to show her all this terrible, like, graphic…

Yeah. I mean, my feed is filled with stuff that I find repellent and hate, but am fascinated by. And so I just see more and more of it. And, you know, again, yeah, it is just strong reaction as opposed to like delight, which I think was the prevailing sort of idea. It's optimized for capture. Yeah.…

That's funny. Because another, yeah, another thing I was going to flag