I think the interesting thing about this article is how quaint it feels five years on. In terms of methods, sure, but also the goal. There's this implicit assu

I think the interesting thing about this article is how quaint it feels five years on. In terms of methods, sure, but also the goal. There's this implicit assumption that a) ML could effectively spot bad actors, and b) this is a fight someone, somewhere is able to win. I feel the information environment has degraded so badly over the last decade that this feels naive.