Isn't there an argument to be made that YouTube has simply improved (perfected?) the algorithm news agencies have been trying to perfect for decades? We all know that sensationalism sells in media, so is the difference between sensational headlines and 'radicalising' YouTube recommendations really a qualitative difference (the latter of which is decidedly sinister and perverse), or just a case of machines outperforming humans?

> difference between sensational headlines and 'radicalising' YouTube recommendations really a qualitative difference (the latter of which is decidedly sinister and perverse), or just a case of machines outperforming humans?

No, the difference is that YouTube doesn't get a free pass; you don't hear “so and so was radicalized on Fox News (or OANN, or wherever)”, but that's not because it doesn't happen, instead it’ll get blamed, if anything, on groups and alternative media that the radicalized individual fell into after being radicalized by something shaped like traditional media.

I've seen it do both. I remember the days when it shoved you towards literal Nazis, and then years later I got in a friendship-ending fight with somebody who was far more pilled than I had imagined. Among the wild stuff he was trying to tell me, was the idea that YouTube pushed people towards moderates, and he had a study that was fairly recent.

If this study was anything like accurate, then AT THE TIME it was pushing people very gently towards more moderate YouTubers, in a subtle way. My nazi-adjacent former friend was really upset about this.

Thing is, both he and I remembered and acknowledged the time when it was feeding people extremist views really noticeably. His problem was essentially that it had stopped. I guess my problem is: who stopped it, why, and when are they going to switch back to doing that again?

If you have the study, I'd consider it, but second-hand anecdotes from the "nazi-adjacent" aren't really compelling evidence. Google did take some minor steps toward curbing things a year or two ago, but I don't know that anyone has done a comparative analysis of the current and former state of YouTube's recommendation system.