Why Other People Are Not Background Noise
The study that started the argument, the critics who took it apart, and the four mechanisms that survived.
In 2007, two researchers named Nicholas Christakis and James Fowler published a paper in the New England Journal of Medicine that made headlines all over the world. They had gotten their hands on something remarkable: thirty-two years of data on 12,067 people from the Framingham Heart Study, including records of who was friends with whom.
What they reported was startling. If your friend became obese, your own risk of becoming obese rose by 57 percent. If it was a close mutual friend, 171 percent. And the effect appeared to travel out three degrees, to your friend's friend's friend, someone you had never met.
Obesity, they suggested, spreads through social networks like a virus.
It is a great story. And I am going to do something here that most programs will not do, because I would rather you trust me than be impressed by me.
The part where I tell you the argument
That 57 percent figure has been contested for over fifteen years, and the criticism is serious. It is not from cranks. It came from other careful researchers.
What came after: the experiments
The way you settle an argument about whether something causes something else is not to argue harder about correlations. It is to run an experiment. And that is exactly what happened.
What is actually new in the science
The field moved from observation to experiment.
Randomized network trials (Honduras, India) have largely replaced arguing about the Framingham data. The experiments support influence, at somewhat more modest magnitudes and typically out to two degrees rather than three.
Biology entered the conversation.
Microbiome strain-sharing across social networks (Nature, 2024) opens the possibility that some of this is not purely behavioral. Researchers have begun asking whether conditions linked to the microbiome, including obesity, might spread through a partly biological route.
The focus shifted from "who you know" to "what they do and say."
The strongest predictor is no longer the body size of your friends. It is social norms (what you believe is normal and approved of in your group) and behavioral modeling (what you literally watch people do). This is much better news, because norms and modeling are things you can actually work on.
What nobody argues about
Set the contagion debate aside entirely. Here is what is not in dispute, and it is more than enough to justify everything in this module.
Relationships predict how long you live.
Meta-analyses of hundreds of thousands of people find that strong social relationships are associated with roughly a 50 percent increased likelihood of survival over follow up periods. Social isolation carries mortality risk in the same neighborhood as smoking. The Harvard Study of Adult Development, running since 1938 and now the longest study of adult life ever conducted, arrived at essentially one conclusion after eight decades: good relationships keep us happier and healthier. Not money. Not fame. Relationships.
Support quality beats support quantity, dramatically.
Winston's research found that helpful, supportive ties predicted greater weight loss, while harmful, unsupportive ties predicted weight gain. The number of people supporting you mattered far less than what kind of support it was.
Doing it together works better than being cheered from the sidelines.
This one is worth tattooing somewhere. It comes back in Segment Two.
Recruiting people you already know into your effort changes outcomes.
In classic work by Wing and Jeffery, participants who enrolled with friends and received social support training completed treatment at dramatically higher rates and maintained their full weight loss at long-term follow up at roughly two thirds, compared to roughly a quarter of those who came in alone.
Undermining is measurable and it hurts.
Leahey and colleagues followed 633 employed adults. Friend support and coworker support for healthy eating both predicted lower weight at 24 months, and family support for physical activity helped as well. Meanwhile, family social undermining of healthy eating predicted weight gain. They also found that having more casual friends who were overweight predicted poorer weight loss, but that social norms for unhealthy eating were the stronger predictor. It was not the bodies. It was the norms.
The formal interventions produce real but modest effects.
A 2024 systematic review and meta-analysis of 24 randomized trials involving 4,919 adults found that social-support-based weight loss interventions produced significant benefits at the end of treatment and at three and six month follow up, though effects during treatment itself were uncertain. A 2023 meta-analysis of social media based interventions found small to medium effects on weight, BMI, waist circumference, and body fat. A 2025 network meta-analysis of digital weight loss components found peer support to be one of nine active ingredients, though not the strongest one on its own.
The four mechanisms
Forget the word contagion. It is a bad metaphor, because it suggests something invisible floating through the air. What is actually happening is four specific, nameable, entirely visible mechanisms. And every one of them is something you can get your hands on.
1. Behavioral Modeling (Mimicry)
This is the most reliably demonstrated of the four, and it happens entirely below your awareness.
Bring two strangers into a lab. Have them eat together. Instruct one of them, secretly, to eat more than usual. The other one eats more. Instruct them to eat less. The other one eats less. Change how fast they eat, and the other person’s fork speed changes too.
This is not a weakness. This is ancient hardware. For most of human history, eating in sync with your group was how you avoided being poisoned, how you got your share, and how you stayed part of the tribe. Your brain is doing something it has done for two hundred thousand years.
Where the leverage is: Who is at the table.
2. Social Norms (Descriptive and Injunctive)
Researchers split norms into two flavors, and the distinction is genuinely useful. Descriptive norms are what you believe people actually do: "Everybody in my family goes back for seconds." "All the guys at work get the large." Injunctive norms are what you believe people approve of: "If I order a salad, they’ll think I’m being difficult." "Nobody wants to be the person who brings up calories."
Both push on your behavior, and here is the disturbing part: norms operate on your perception, not on reality. You are responding to what you believe the group thinks. You are frequently wrong. And you never check.
There is modeling work (Crielaard, 2020) suggesting that social norms can substantially counteract health awareness. In one analysis, BMI decreased about 50 percent less when norms were driving behavior compared to health awareness alone. In men, overweight social norms held group-level BMI near overweight levels despite individual health awareness. People knew better. The norm won.
Where the leverage is: Changing which group’s approval you are performing for, and checking whether the norm you think exists actually exists.
3. Availability and Opportunity
The most boring mechanism, and probably the most powerful.
Other people put food in front of you. Other people choose the restaurant. Other people schedule the meeting at 7pm so you skip your walk. Other people bring donuts. Other people say "let’s just order in."
This is where the Social OS and the Environmental OS shake hands. Other people are the single largest modifier of your convenience zone.
Where the leverage is: Getting a vote on the where, the when, and the what before the plan is made.
4. Identity and Belonging
The deepest one, and the one that broke Denise at that birthday party.
Every group you belong to has an unwritten story about who its members are. "We’re a family that eats." "We work hard and we drink hard." "We’re the fun ones." "We’re not those health nut people."
When you change your behavior, you are not just changing your behavior. You are proposing a revision to the group’s story about itself. And you are implicitly suggesting that the way they have been doing it is not the only way. Nobody signed up for that. Nobody voted on it. And the group, being a system, will push back toward homeostasis.
This is why the sister said "too good for cake now." She was not attacking Denise. She was defending the story.
Where the leverage is: This is the one where you go back to your Betterness OS.
The four questions, translated
You already know the four questions from the Environmental Operating System. Here they are again, pointed at people instead of rooms.
| In a room | In a relationship |
|---|---|
| What's here? | Who brought it, and who asked for it? |
| What's seen? | Whose plate am I looking at? |
| What's convenient? | Who is making the easy thing easy? |
| How does this space make me feel? | Who is in the room when I feel that way? |
Before you go on
The rest of this module is three segments and an assessment. You do not need to accept the contagion claim to get value from any of it. You only need to accept that the people around you are modelling behaviour, setting norms, controlling availability, and shaping your identity — and none of that is controversial.
- The Navigator asks who is teaching you.
- The Crew asks who is on the boat with you every day.
- The Lighthouse Keeper asks who you are helping, and why that is the part that makes everything permanent.