IMO they should not only be opt-in, but should actually be required to publicly list the parameters and weights they’re using and allow users to tune those weights.
> IMO they should not only be opt-in, but should actually be required to publicly list the parameters and weights they’re using and allow users to tune those weights.
I wonder how many people here know that many of the popular apps have rolled out finer controls for recommendation algorithms so you can do this. On Instagram you can go in and see the topics your recommendation algorithm picked up and modify them manually if you like.
I think the goalposts will just continue to move, though.
People here know that they have finer controls (which are still not actually that fine and also don’t really make the parameters auditable). The problem is these settings are hidden away in places most people will never look. And also, I stress again, none of this is actually auditable because they treat these as some kind of trade secret special sauce and there’s really no reason society should feel obligated to support or enable this business model.
That actually wasn’t the objection in the article we’re discussing at all.
The objection is that recommendation algorithms show people more content they want to view, which leads vulnerable people (kids in this case) to consume more content.