"We need more from academics and industry," a policy-maker tells me. I've heard this before. Lawmakers are pressing forward with social media regulation, but if it's full of strange and bad ideas it's because the people who should know aren't putting forth better ones. 1/

Unfortunately, it's not obvious what to do here. For example: one definition of transparency would be when users' "folk theories" of how a recommender works are a good match for the underlying algorithm. For which algorithms is this true? That's a major research question! 2/

Or, maybe you want to require platforms to carry higher quality news. What does that mean? A list of sources you personally like? Newsrooms over a certain age or size? Articles approved by some external body? Would that be prior restraint? 3/

Or let's take the polarization. Important consider the role of media in a conflict! But so far, evidence is social media users see *more* diverse news than non-users. Experimentally, little causal effect on polarization. We don't know the effects yet, let alone the solutions. 4/

Or take hate speech. International law says states have an obligation to act on "incitement" to hate or violence. It also says states must not interfere in freedom of expression. What happens with automated systems -- with unavoidable error rates -- applied at scale? 5/

In short:

- There are numerous potential policy goals
- I've never seen a law that considers them *all at the same time*
- We are years away from having a solid understanding on many topics

I don't think we are ready to write good laws. But we better be researching, fast.

FIN