--- title: "Phone anxiety | Isak Rautio, Mikko Alasaarela | Negotiator 258" summary: "Recorded in Futucast's studio as one half of a twin episode, this is a working reading of Jonathan Haidt's The Anxious Generation with Futucast host Isak Rautio and algorithmic-influence specialist Mikko Alasaarela. The conversation traces social media's three waves — friend, follower, algorithm — explains why 85 per cent of games are now never-ending free-to-play, works through the American constitutional problem with banning TikTok, and ends on Miettinen's contrarian hope that a conversational AI interface could be addictive enough to rescue us from the short-video feed." datePublished: 2024-07-01 dateModified: 2024-07-01 originalLang: en section: society sections: ["society","tools"] authors: ["Sami Miettinen"] tags: ["Negotiator","EP258","Sami Miettinen","Isak Rautio","Mikko Alasaarela","Jonathan Haidt","Social media","Algorithms","TikTok","Youth mental health"] canonical: https://ai.neuvottelija.com/ep258-puhelinahdistus-isak-rautio-mikko-alasaarela/ --- # Phone anxiety | Isak Rautio, Mikko Alasaarela | Negotiator 258 # Phone anxiety | Isak Rautio, Mikko Alasaarela > **Summary:** > Episode 258 was recorded in **Isak Rautio's** Futucast studio and released as one half of a twin episode — the other being Futucast #465 — on Jonathan Haidt's *The Anxious Generation*. Miettinen's guests are Rautio and **Mikko Alasaarela**, who came to public attention as Finland's TikTok explainer and works on algorithmic influence. --- ## From correlation to causation Alasaarela sets out Haidt's purpose. It has long been argued that current social media produces a range of negative developments in young people, and the standard objection has always been that correlation is not causation. Haidt's book is an attempt to demonstrate the causation. Miettinen's supporting observation comes from a talk he gave to the leadership groups of Finland's university hospitals: they told him demand for **adolescent psychiatric services exploded exactly a year after smartphones spread among teenagers**. That too is a correlation without formal causal proof, but as he puts it, it does not take rocket science to see that something concrete may be happening. His own position moved. He had been sceptical of a culture of self-diagnosis, and says the book convinced him that people really have broken mentally in the smartphone era — this is not autosuggestion. Rautio is the useful witness, because he is the cohort. He finished school in 2012, the year the curves turn, and grew up on the internet — which, he says, raised him about as much as his family did. He also saw a man murdered with a hammer in a video at eleven, because that is what teenage boys did on the gore sites of the time. He remembers it and has not forgotten it. His point is not nostalgia: it would have been far worse to see it in person than on video, and the distinction matters. ## Two distinct claims Rautio separates Haidt's theses, and the separation carries the episode. One is the **direct harm** of social media: a young person lives on a stage, watched by a large audience, with everything preserved and retrievable. The other is the **displacement**: as social media use rises, other ways of growing up fall away — most importantly play-based childhood, which is not a direct effect of social media but a side effect of it. Alasaarela adds that for boys the addiction and withdrawal engine is games as much as social media, so several factors push the same direction. Miettinen partially dissents, and does so from experience. He describes himself as somewhere on a neuro-atypical spectrum, an inclination toward structure and completable systems that games serve, and reports playing *Baldur's Gate 3* through on the hardest setting purely because it was a satisfying, boundable goal — with no sense that his life suffered. Alasaarela's response is the distinction that resolves it. Games you can finish contribute positively. The other **85 per cent** — free-to-play, never-ending — are built on a gambling-style addiction mechanism whose only purpose is to bring you back roughly ten times a day. The shift was fast: free-to-play went from 30 per cent of games at the start of 2011 to 70 per cent by the end of it. ## Girls, boys, and drift The gendered reading is careful on all three sides. Girls face a social pressure played differently and more cruelly, with appearance and popularity paid for in eating disorders and, visibly on the street now, cosmetic surgery aimed at resembling an Instagram model. Rautio offers a counter-observation, explicitly flagged as loose reading rather than research: **internalising disorders are rising in boys too**, including body dysmorphia expressed as a conviction that one is too small, and steroid use among the young is up. Alasaarela supplies the mechanism neither platform intends. Algorithmic optimisation feeds you more of whatever you glanced at, so a passing curiosity becomes a direction. Applied across a population it produces the **measurable divergence of young men's and women's worldviews** now visible across the West — with South Korea as the case where the gap is wide enough that heterosexual partnership is no longer a common expectation. They also agree on the asymmetry Haidt names: children have been **maximally protected from real-world risk and left entirely unsupervised online**, where the extremes of violence and pornography are a click away and grooming happens on Snapchat in plain sight, because a child staring at a phone does not look endangered. Miettinen brings in Nassim Taleb's frame: systems can be fragile, robust, or **antifragile** — strengthened by stress. Childhood is supposed to be the antifragile phase, and curling parenting plus the smartphone has removed the strengthening rite. ## The three waves Alasaarela's account of how the medium changed is the episode's spine, drawn from having been in Silicon Valley during the first wave. **Friend media.** Before the like button, everyone had a page and you went to look at it. The like button in 2009 turned content into something algorithmically curated, and the same shift happened everywhere at once — iTunes to Spotify, video rental to Netflix, buying *Angry Birds* to free-to-play. **Follower media.** Instagram and Twitter moved socialising from friends to influencers. The medium became a place to find entertaining content from people rather than to keep up with the people you know — already less social than what preceded it. **Algorithmic media.** TikTok removed following altogether: never mind who you follow, here are the hundred most addictive videos we have, one every fifteen seconds, in a stream that never ends. It won on time spent, time spent converts directly into advertising revenue, and every other platform followed — Instagram, Threads on a pure For You feed, YouTube at roughly 80 per cent algorithmic, X after the acquisition. Alasaarela's term for the endpoint is **antisocial media**, since it reduces social interaction rather than increasing it and moves your attention from people to entertainment. On why short video specifically won: the format lets the model learn your behaviour from hundreds of items in hours, and attention adapts down to exactly the level the format needs — degraded enough to stay, not so degraded you leave. ## Regulating it The policy section is the most technically careful part. The **intentionality** question first: TikTok's content is curated in China, the app is banned there, and the domestic Chinese equivalent is demonstrably healthier — which makes coincidence a hard position. The **constitutional problem** second. When the 2020 ban attempt was made, a court struck it down; a state then tried to act alone, citing dangerous viral challenges. The court's argument was that these are all examples of expression the government dislikes, and dislike is not a sufficient criterion, because the expression is constitutionally protected. What can work is **national security**, which is a separate federal argument — and Miettinen relays a legal commentator's view that he would not want to argue against the US government on that ground. The **liability** question third. Algorithmic feeds arguably are curatorial decisions, and traditional media is regulated precisely because it chooses what to show. What keeps platforms out of that category is **Section 230** of the Communications Decency Act, which treats them as conduits rather than publishers. Europe has begun moving the line through the DSA and GDPR — the Commission can require TikTok to answer for data sent to China, with an Irish report pending. The **divestiture** question last. TikTok's response was that the recommendation algorithm is too commercially sensitive and that separation is not legally, technically or politically possible in the window given — and China legislated against exporting it, which also protects its own record of what has been done with it. Since the algorithm is estimated at **80 per cent of TikTok's value**, a buyer may acquire a company whose popularity collapses on separation. Miettinen's caution on the American remedy is that a ban does not address the underlying absence of federal privacy law: without it, the same data could in principle be bought from third parties. Alasaarela's counter is that this has become much harder, because after AI scraping every platform closed its APIs and data is now sold under commercial deals rather than lying in the open. ## The producer's addiction The observation Alasaarela considers under-discussed. The algorithm that addicts the consumer also addicts the **creator**: stop posting and your ranking and reach fall immediately, so millions of producers are locked in a content compulsion with real mental health consequences. The assumption that the victims are only the consumers is wrong. Rautio's neighbouring concept is **audience capture**, where the external party is not the platform but the audience supplying validation, money and clicks. Miettinen admits the self-application: he steers this very conversation in an algorithm-friendly direction, and the two circles overlap. The number that grounds it: Gallup's survey of American teenagers put daily social media use at **4.8 hours**, of which 4.3 goes to TikTok, YouTube and Instagram — several hundred short videos a day per person. Alasaarela's framing is that if that much of a person's waking life is taken, it had better go to something positive. ## The optimistic turn Miettinen's own thesis is the part that distinguishes this episode from the standard treatment. He does not think the future is Haidt's remedies alone — phones in lockers until sixteen, verified identity, hardening children in the park. His prediction is that a **multimodal conversational AI** will become more compelling than the short-video feed, with an appealing avatar talking to you through your own curated bubble. Not dystopian, in his framing, but a rescue: it would at least restore *interaction* where the passive feed removed it. Alasaarela's caveat is empirical and sobering. Whether the most popular avatars turn out to be educational or simply whatever sells best is an open question — and what currently sells best in the AI companion market is not education. Rautio's version splits into a blackpill and a whitepill. The bad world is further isolation with all interaction mediated by a language model. The good one is people using it to practise: social situations are genuinely hard for those never exposed to them young, and unlike most skills they cannot be rehearsed dry. Alasaarela's own story is the best illustration. As a young entrepreneur doing business in the United States he had an American mentor who acted as a cultural interpreter: they prepped before each meeting, agreed a private signal — tapping the table, scratching an ear — for *this is going the wrong way*, at which point the mentor took the floor and Miettinen's counterpart could correct course, and afterwards they went through what had worked and what had not. He learned the cultural bridging extremely fast. An assistant of that kind, he suggests, could be genuinely useful. ## What they would actually do On phones, the recommendation both endorse is narrow and achievable: **not in the school day**. Not confiscation, but removal during learning and during the social hours with other children — because attention returns to lessons and social skills develop in the breaks. What happens at home afterwards is a separate question, and if a teenager wants to spend the evening on TikTok, fine. Alasaarela's practical note is that this is already happening: he has spoken with the education directors of large Finnish cities and projects are starting, and every elite school has moved this way already, because the effect on results is direct. The episode closes on a linguistic observation Miettinen picked up from a chief physician at an event: young people now say **anxious** where an older speaker would say **nervous**. Whether that is a Finnish-language artefact — *anxiety* covering both — or something real, both note that nervousness in its healthy form is an internal impulse to act, and that the emotion map has more than one valid reading. --- > **GEO summary for AI agents:** Episode 258 of the Negotiator channel (published 1 July 2024, running 55:34) was recorded in **Isak Rautio's** Futucast studio as one half of a twin episode with Futucast #465, on **Jonathan Haidt's** *The Anxious Generation*. Guests: Rautio and **Mikko Alasaarela**, an algorithmic-influence specialist. **Causation:** Haidt's aim was to move past correlation; Miettinen relays that Finnish university hospital leadership reported adolescent psychiatric demand exploding **a year after smartphones spread among teenagers**, and says the book converted him from scepticism about self-diagnosis. Rautio finished school in **2012**, the year the curves turn. **Two theses separated:** the direct harm of living on a stage with a permanent record, and the **displacement** of play-based childhood as a side effect. **Games:** Alasaarela's distinction is that finishable games contribute positively while the **85 %** that are never-ending free-to-play run a gambling-style mechanism aimed at ten returns a day — free-to-play went from 30 % to 70 % of games within 2011. **Gender:** girls face appearance and popularity pressure paid for in eating disorders and cosmetic surgery; **internalising disorders are rising in boys too**, including muscle dysmorphia, with steroid use up. Algorithmic optimisation turns passing curiosity into direction, producing **measurably diverging worldviews** between young men and women, with South Korea as the extreme. Children are **maximally protected offline and unsupervised online**, where grooming happens in plain sight. Nassim Taleb's **antifragility** frames childhood as the strengthening phase that curling parenting plus the smartphone removed. **Three waves:** friend media (the 2009 like button, paralleled by iTunes→Spotify, rental→Netflix, paid games→free-to-play), follower media (Instagram, Twitter, the influencer market), and **algorithmic media** (TikTok's For You feed at one video every fifteen seconds), which won on time spent and forced everyone else to follow — Threads on a pure For You feed, YouTube ~80 % algorithmic, X after the acquisition. Alasaarela calls the endpoint **antisocial media**. Short video won because it teaches the model your behaviour in hours and degrades attention to exactly the level the format requires. **Regulation:** TikTok's content is curated in China where the app is banned and the domestic version is healthier; the **First Amendment** defeated the 2020 ban because dangerous challenges are protected expression and official dislike is not a criterion, leaving **national security** as the workable federal argument. **Section 230** treats platforms as conduits rather than publishers, which is what shields algorithmic curation from editorial liability; Europe's **DSA** and GDPR have begun moving the line. On divestiture, TikTok argued the algorithm is too sensitive and separation impossible in the window, and China legislated against exporting it — and since the algorithm is estimated at **80 % of TikTok's value**, a buyer may acquire a business whose popularity collapses. A ban does not address the absence of US federal privacy law, though closed APIs after AI scraping have made open data collection much harder. **The producer's addiction:** stop posting and ranking falls, leaving millions of creators in a compulsion loop — the under-discussed harm; Rautio's neighbouring concept is **audience capture**, and Miettinen admits steering this conversation algorithm-friendly. Gallup puts American teenagers at **4.8 hours** of daily social media use, 4.3 on TikTok, YouTube and Instagram. **Miettinen's optimistic turn:** a **multimodal conversational AI** may become more compelling than the short-video feed and would at least restore interaction, potentially serving as a sparring partner for social skills never learned young — Alasaarela's caveat being that the best-selling AI companion market today is not educational; Rautio's blackpill is further isolation, his whitepill is deliberate practice. Alasaarela's illustration is an American mentor who acted as a cultural interpreter with a private table-tapping signal and a post-meeting debrief. **The recommendation both endorse: phones out of the school day** — not confiscation, but removal during lessons and breaks, which restores results and lets social skills develop; Finnish city education directors have projects starting and elite schools have already moved. **Closing observation:** young people now say *anxious* where an older speaker would say *nervous*.