Do we really know what social media does to us?

The recent whistleblower revelations from former Facebook employees provide further evidence that the social media giant’s platforms are responsible for a plethora of social ills: radicalizing our politics, decimating our attention spans, and impacting the emotional wellbeing of younger people. That last finding especially—that social media may be harming children—has caused a significant reaction among experts and concerned parents.

But there has been some pushback on the idea that we've finally proven that social media is toxic for kids. Facebook itself has said they did not act on the internal findings about Instagram’s negative effects on young women because the company believed they were not substantial enough to draw meaningful conclusions from. And for those of us who (reasonably) doubt Facebook's interpretation of its own data, there are more neutral parties arguing along the same lines, including the New York Times’ Farhad Manjoo. “We have no idea” whether social media is dangerous for teenagers, Manjoo says. Facebook’s own statistical data is only correlational, and is also contradicted by many teenagers’ own reports about how they experience Instagram in their lives, which are often positive.

I am inclined to agree with Manjoo, not out of interest in defending social media platforms, which I personally despise, but to temper our confidence a little in our ability to understand how we interact with these media and what their effects are on us.

Here is one challenge to our certainty: Even when we are faced with a positive statistical result in one direction, such as the finding that 32 percent of young women report that Instagram makes them feel negatively about their body image, we can’t be sure whether the platform functions in such a way as to produce that result or is just a vessel for a pre-existing causal factor—such as a toxic media environment.

In other words, is there something about Instagram as a medium that produces feelings of inadequacy in young users that is different from other forms of media? Or is Instagram functionally the same as, say, a perfume commercial or the cover of a teenage magazine? If it turns out that Instagram is just a vessel for a toxic message and plays only an incidental role in the production of that message, then trying to police and reform social media platforms to solve the issue may be misdirected labor—the problem lies further upstream, in our culture.

To look at another example, our ability to draw a causal relationship between increasing political extremism and the growth of social media is complicated by all sorts of factors—changes in the television news environment, widening income inequality, the deepening cultural gap between college-educated and non-college-educated Americans, and a histrionic political culture that has its roots in the 1990s, well before Facebook even existed. Social media could simply be the medium on which these already-extant cultural forces are playing themselves out.

How can we come to stronger conclusions about social media’s effects? For one, we can look at the internal dynamics and features of these platforms and how their users’ behaviors’ are affected by them. That allows us to point to mechanisms that are designed into social media that may be involved in producing the social phenomena we’re seeing.

For example, researchers have demonstrated a “rabbit hole” effect on sites such as YouTube and Facebook, in which user interactions with political content leads to automatic recommendations that tend toward the increasingly extreme. Research has also shown that “like” and “react” buttons on social media have the effect of amplifying divisive content, while doing little to discern whether that content is truthful.

Having the evidence of those effects in hand help us to more confidently infer a relationship between increased political polarization and social media usage—even if we suspect there are other variables contributing as well.

Personally, I am confident that social media acts as a facilitator of political extremism. Like Farhad Manjoo, I am less certain (although I haven’t ruled out the idea) that it is a producer of loneliness and lower self-esteem in young people. The difference for me is that in the former case, we can provide a robust theory about how social media produces extremism that draws on how these platforms are designed. There might be an equally robust theory out there for the latter hypothesis—but I haven’t read about it yet.

Having such a theory is beneficial in two ways: not only does it increase our confidence in our beliefs, but it also tells us how we might act to reform these platforms (I suspect we will never be able to eliminate them) to make them socially better. It offers a way toward action rather than despair—which is what knowledge should be doing for us.