Adult smoking in the UK was about 45% in the mid-60s. It is now down to 10%. Getting to this point has involved a series of interventions in the form of:
- Health education
- Large media campaigns
- Higher prices and taxes
- Smoke-free laws
- Stigmatisation of smoking
- Availability of harm-reduction products.
Of course, once you get through the loose soil, you reach a bedrock of hardcore smokers who are less susceptible to stopping. Solving this problem will require more desperate measures, which is also true for youth smoking, which has dropped from 9% to 1% in around a decade.
Impacting that 1% will take something fairly remarkable. Still, Kaya Burgess at The Times has got exactly that for us, via a new paper published in Science, called Peer influence decay and behavioural diffusion in adolescent networks: A simulation approach.
The paper
I’d recommend reading the paper yourself, because it’s certainly a novel approach. The general gist here is that smoking is a very niche pursuit among school-aged youths. Targeting the whole class is inefficient, not least because it involves a lot of preaching to the converted. So, how can we optimise the impact of messaging? According to Burgess, it’s by targeting the “cool kids”.
Per the paper, teen behaviour is highly influenced by peers. The suggestion here is that the influence lasts about three friendship “steps”: in plain language, a friend of a friend of a friend. Then, that capacity to change people’s minds fizzles out.
If you want to change a school’s smoking habits, you don’t need the message to reach everyone. The most efficient way is to find the few smokers who sit at the centre of overlapping friendship circles and get them to quit. Per the paper, targeting just 10-30% of these smokers will have a ripple effect and therefore eliminate smoking in the school.
Strange intervention
While Burgess seemed happy enough with the paper, I found it a bit odd. Let’s not mince our words here. What this work describes is how to pick a tiny group of socially central teenagers and use them as levers to reprogram everyone else’s behaviour.
While the end might justify the means for some, I see a mix of public health “nudges” combined with the ruthless efficiency of social media logic flows. Indeed, one can’t help but think about things like those leaked Facebook internal memos about using dopamine to hook teens to the platform.
To be clear, I’m not directly comparing them. The researchers and Burgess have aims that are far more noble than selling targeted advertisements to kids. But on some level, we know that Facebook was wrong because they applied this “optimisation” on a vulnerable population who had yet to develop the skills to buttress themselves against this manipulation.
I don’t want to over-egg it, but turning kids and friendship groups into infrastructure and tools feels a bit cold. Teenage smoking is strongly correlated with existing mental health issues, so treating them as an engineering problem or, worse still, something to extract utility from, does not sit right with me.
Where are the ethics here? Is anyone concerned about how being used as a cog might affect the central smoker? What is the nature of these interventions? Will the kids feel pressured to succeed? Does anyone care why they started smoking in the first place?
Final thoughts
When you strip it back, the mechanism here is a playbook on how to run a soft behavioural‑engineering campaign on minors. As we’ve seen over the years, school-aged kids are susceptible to trends and contagions. If people feel justified in using this method to solve a basically non-existent problem, think of what that means for commercial influence operations, political micro‑targeting, state propaganda, and more.
