LogFAQs > #950559354

scar the 1 posted...

I actually don't think we have, you're probably confusing me with someone else.
I'm not convinced. I'm sure there would be some amplification, and the problem wouldn't just disappear, but I think the business model of social media is at the core of this issue.

Yeah, don't get me wrong, I think removing those influences will markedly reduce the problem, I just think what will be left will still be troublesome.

As long as you assume that the salience and novelty means greater adoption/share rate, and people who find similar things salient and novel are more likely to be connected to each other, you will get amplification.

I suppose the primary disagreement here is simply how bad the remaining amplification would be, and that's really just an empirical question.


---

=E[(x-E[x])(y-E[y])]