Who’s Running the Greenhouse?
Social media may not have created the problems teenagers face today, but like a kind of greenhouse run for the wrong reasons, it may now be keeping them there.
Like many of you, I have followed Jonathan Haidt’s work on smartphones and young people, so I was struck by a forceful rebuttal from the less widely known but very eminent sociologist, danah boyd. Her recent blog post, provocatively titled “No one paid me to disagree with Jonathan Haidt. It was teenagers who taught me that he was wrong”, made me wonder how far apart the two really are, or whether they might be closer than either would care to admit.
Haidt’s bestseller The Anxious Generation has put smartphones and social media at the centre of the debate about declining adolescent mental health. After years studying teenagers’ digital lives, boyd comes at the issue from another direction: less interested in what social media does to young people than in what they are trying to do with it.
But what if social media began as a kind of greenhouse, a covered shelter for young people with nowhere safe left to grow outdoors, only to become the very thing that keeps them from going back outside?
A childhood that was already changing
boyd’s argument begins with the world that adults have built, intentionally or otherwise, for teenagers. Over recent decades they have steadily lost the freedom to roam and gather unsupervised that earlier generations simply took for granted, with public spaces becoming less welcoming, parents more anxious and childhood ever more scheduled. Little wonder, she argues, that teenagers moved indoors, into the digital equivalent of a greenhouse, where they could meet friends, experiment with identity and escape adult supervision. Indeed for those struggling with difficult family circumstances, exclusion or loneliness, it was often a lifeline.
She is no apologist for the technology companies, and openly scorns what many platforms have become, but boyd rejects the idea that removing the technology fixes the problems that lie beneath it.
The other half of the story
Haidt, I suspect, would agree with much of this. A central argument of The Anxious Generation is, after all, that we have overprotected children in the physical world while underprotecting them in the digital one, and he campaigns for more independent play and more risk.
Where he parts company with boyd is over the digital substitute. For Haidt, social media does not merely compensate for the disappearance of a freer childhood; it accelerates it, through disrupted sleep, social comparison, cyberbullying, compulsive use and the displacement of face-to-face friendship. In the 2026 World Happiness Report, he and Zachary Rausch argue that the evidence now shows harm at population level, although the same report finds the decline concentrated in English-speaking countries and Western Europe; elsewhere, young people are mostly happier despite heavy use.
The timing, though, hints at how both could be right. The erosion of independent childhood that boyd describes was under way long before the smartphone, yet the sharp deterioration Haidt documents arrives in the early 2010s, just as smartphones and algorithmic feeds became the norm. A slow loss followed by a sudden turn is exactly what you might expect if the place of refuge itself changed character. boyd asks why teenagers needed a digital refuge in the first place, while Haidt asks what happens to them once they get there, and both questions are highly relevant.
When the solution becomes part of the problem
Think of it like this. A gardener, worried about late frosts and careless feet, moves their young plants into a greenhouse. It is not a foolish decision, since some of those seedlings would not have survived outside, but plants that never feel the wind grow tall, soft and weak-stemmed, which is why gardeners know to “harden off” seedlings, putting them outside for a few hours a day before planting them out.
Now suppose the greenhouse is handed to a contractor who is paid not for the health of the plants but for how long they stay inside, and who automates the light, heat and feeding to that end. The plants grow fast and look lush, yet become ever less able to cope outdoors, until everyone agrees it would be cruel to put them out. So the doors stay shut, and the fragility the greenhouse created becomes the argument for keeping it.
Far-fetched? Consider the teenage version. Young people lose many of the opportunities that once built independence and confidence, so they turn to digital spaces, and at first the benefits are real: friendships maintained, communities formed, loneliness relieved. But platforms grow ever better at competing for attention, and algorithmic feeds, notifications and endless stimulation make them harder to leave, so that what once complemented ordinary social life gradually begins to displace it… with Covid acting as an accelerant. Boredom, once an invitation to invent something to do or find someone to talk to, becomes a mere inconvenience, instantly relieved by screen interaction. As offline life thins out, online life matters more, and the cycle reinforces itself.
Get James Capon’s stories in your inbox
Join Medium for free to get updates from this writer.
In this account, technology did not cause the original problem, but it has become remarkably effective at sustaining it. This is a plausible story rather than a proven one, I admit, but it reconciles two apparently contradictory positions.
The distinction we keep missing
Too often we talk about social media as though it were a single activity. A teenager using WhatsApp to arrange a football match is doing something quite different from the same teenager spending two hours on an algorithmically curated video feed: one supports real-world relationships, while the other may displace them. Even that line blurs, since group chats can breed pressure while some online communities offer genuine belonging.
Shouldn’t we therefore judge technology by what it enables, what it displaces and how much control the young person retains? That would move the debate from screen time towards design. In greenhouse terms, the problem is less the glass than who controls the thermostat, and for what they are rewarded.
You can check out any time you like, but you can never leave
Haidt’s latest intervention, published with three colleagues in November 2026, adds a further dimension. Parental consent, he argues, is no answer, because families are caught in what economists call a collective-action trap. If all your 13-year-old’s friends are on Instagram or TikTok, refusing permission excludes your child from an important part of their social world, while allowing it exposes them to precisely the environments you would rather they avoided.
In a 2025 American survey Haidt cites, 54% of parents said they felt obliged to allow social media because other families already had, which is why he advocates enforceable age limits and placing responsibility for platform safety on the technology companies rather than on individual parents.
In a way, this reinforces boyd’s point, since teenagers are not necessarily choosing digital spaces because they prefer them to real-world relationships, but because that is where their friends already are. Yet it also exposes a weakness in simply defending teenage autonomy, for how meaningful is individual freedom when exercising it risks social exclusion?
Have we become the anxious generation?
Behind the disagreement between Haidt and boyd lurks a larger question. Are today’s teenagers unusually anxious, or have they inherited the anxieties of their parents?
We are reminded of danger every day, as each crime, accident or tragedy becomes a national story, endlessly repeated through news and social media. Parents understandably want to protect their children, but somewhere along the way, has protection become confused with the elimination of risk? And with streets dominated by traffic and public spaces that rarely welcome teenagers, the old freedoms cannot be restored simply by telling children to go outside.
Rebuilding what we’ve lost
So where does that leave us? Haidt favours coordinated restrictions that spare teenagers the pressure to join social media before they are ready, while boyd worries that restrictions take agency from young people who already have too little. That is a genuine disagreement, but both want young people to have more independence and richer lives beyond adult supervision, and both see the current digital environment failing many of them.
It seems to me that restricting ‘harmful’ technology without restoring real-world social life treats the symptom and neglects the cause… effectively taking the plants out of the greenhouse without hardening them off or giving them a garden worth growing in. Yet investing in the garden while ignoring products designed to monopolise attention sounds equally inadequate. We need to do both.
Encouragingly, the garden is already being replanted in places. Let Grow, which Haidt co-founded, helps children rediscover independence through unsupervised play and small adventures; in Britain, Playing Out has helped communities reclaim residential streets for children’s play; and other initiatives invite teenagers, particularly girls, to help design the parks and meeting places they actually want to use. Independence cannot be taught entirely under supervision, and at some point adults have to let go. Perhaps here Haidt and boyd really do agree: the answer being not simply the removal of phones, but giving young people somewhere worthwhile to go, where they can make friends, make mistakes and learn to manage their own lives.
The aim is not to recreate a rose-tinted childhood from fifty years ago, but to build a modern world in which young people can once again exercise meaningful independence. Social media offered a refuge from a childhood that was already becoming less free; the trouble is that the refuge is now sustaining the very conditions that made it necessary. And if we want our children to become less anxious, we may have to start by becoming a little less anxious ourselves.
Note: danah boyd styles her name in lower case, which I have followed here.
