To make things worse, for 99% of people things might look normal but for 1% it may show a large skew. The net can have memorized this.
Obviously, this becomes an arms race. The usual strategy is to add random checks, friction to contribution and severe-enough punishments for offenders (lifetime RL bans, etc.)
Expert opinions welcome - I'm a tourist.
So they also have to give up all their user data?
Meaning if I record my own inputs, I should be able to recreate the suggestions I see using the published software. This doesn't mean the inputs of other users are exposed to me, but it does imply that "the algorithm" can only use data that is either publicly exposed or is my own inputs.
That seems pretty restricting given that even the most basic recommendation would try to cluster users based on interests and use that to recommend stuff (eg. Amazon's "customers who viewed this item also bought..." from decades ago).
That would be very hard/impossible to recreate in order to test the algorithm.
In whatever form it takes such that a regulator can scrutinise it. If that seems arduous or impacts the social media business model, too bad. Nothing of value will have been lost. I am not typically a person who favours over regulation, even my views on AI are likely too liberal for HN, but on the matter of social media impacting children, I'm all in for regulation.