Amrit Kaur Purba and colleagues argue that social media restrictions operate within a wider system of adolescents, families, schools, governments, and commercial actors - and therefore should be treated as complex systems interventions rather than isolated behavioural policies.
Without this broader approach, they warn that "governments risk introducing highly visible policies that are poorly understood and may cause unintended harm while leaving root causes unchanged."
They outline how lessons from other commercial determinants of health such as the tobacco and alcohol industries can help predict how social media companies may adapt politically, scientifically, technologically, and economically after regulation.
For example, companies may try to redefine what counts as "social media" so that it falls outside new regulations, invest more in related or less regulated spaces, and shape policy through lobbying, public messaging, research funding, and marketing.
Adolescents themselves may also adapt by moving to more private or harder to monitor spaces, such as encrypted messaging apps or AI based chat systems.
The authors also note that restrictions may not affect all young people in the same way, suggesting that those with supportive families, strong digital skills, access to high quality educational resources, and opportunities for safe offline activities may benefit more than those facing isolation, unsafe environments, or limited support.
One young person's perspective, who is also an author on the paper, seems to support this view. While acknowledging that social media can be both helpful and harmful, they describe it as "a place where friendships are made, where people can find communities, express themselves, learn new things, and sometimes a place to escape difficult situations."
They add: "I have had friends reach out to me on social media about things they aren't comfortable talking to family members about, and I have done the same. Without social media, what could we have done?"
The authors suggest using systems mapping to anticipate these effects and design more balanced, evidence informed approaches.
While this approach cannot predict exactly what will happen, it helps show how different parts of a system connect, how these parts may respond to change, and where effects may feedback on one another, they write.
As such, they recommend that evaluations move beyond standalone measures like screen time or short term changes in mental health to capture wider factors such as school engagement, social connections, industry and platform responses, and longer term effects.
And they conclude that taking this broader view need not delay action. Instead, "it will help ensure policies are balanced, flexible, evidence informed, and improve over time."
