Why we don’t track screen time.

Screen time is a category error.

The phrase "screen time" comes out of television research from the 1960s and 1970s, when the screen in question was a CRT in the living room and the activity on it was almost entirely passive. A kid watched the screen. The screen did not watch them back. The unit of analysis was duration of consumption, and within that frame, duration was a reasonable proxy for risk: more hours in front of the TV meant fewer hours doing almost anything else.

That research frame got transplanted, mostly without rethinking, onto every kid-and-pixels situation that followed. Game consoles, computers, phones, tablets. The American Academy of Pediatrics issued screen-time guidelines that, until a major revision in 2016, treated all screens as roughly equivalent. The revised 2016 guidance acknowledged that the medium had changed and that "all screen time" was no longer a useful frame.¹

The category error is straightforward once you say it out loud. Watching a kid’s show, reading an e-book, video-chatting a grandparent, playing a console game, and shipping a website you built are five different activities that happen to share a piece of glass. A unit that measures only the piece of glass is measuring the wrong thing. It’s the cognitive equivalent of measuring nutrition by counting hours spent in the kitchen.

What the research actually says.

Common Sense Media has published an annual Census of kids’ media use since 2011. Their reports, especially the 2021 and 2023 editions, draw an explicit distinction between passive media (watching, scrolling) and active or creative media (making, building, video-chatting).² The thrust of their findings, sustained across years: kids consume substantially more passive media than active or creative media, and the ratio has gotten more lopsided as short-form video apps have grown. The Census doesn’t argue all screen time is bad. It argues the composition matters more than the duration.

Anya Kamenetz, who covered education and tech for NPR for over a decade, made the same case in her 2018 book The Art of Screen Time. Her one-line summary: "Enjoy screens. Not too much. Mostly with others."³ The book reads like Michael Pollan’s food rules transplanted onto screens, and the move is the same. Stop counting calories. Look at what’s on the plate.

A unit that measures only the piece of glass is measuring the wrong thing. On why duration is the wrong proxy

The deeper learning-science literature backs this up. Mitchel Resnick’s work at MIT’s Lifelong Kindergarten group has spent two decades arguing that creative-screen-time and consumption-screen-time produce very different outcomes for kids.⁴ A kid building something on a screen is doing constructionist work, in Papert’s sense; a kid scrolling is doing something closer to TV. Resnick’s framework for kid-creative environments (Projects, Passion, Peers, Play) isn’t a screen-time framework at all. It’s an active-engagement framework, and it predicts that the kid’s outcome depends on what they did, not how long the screen was on.

None of this is fringe. It’s the consensus inside developmental psychology and educational research. It just hasn’t propagated into consumer dashboards, which still mostly show a timer.

The metrics we celebrate instead.

The Tell and Show family dashboard surfaces a different set of numbers as the headline. They’re the numbers we think actually correspond to whether the studio is doing what it’s supposed to do.

• Projects shipped. The count of public URLs your kid has published, with your approval. This is the durable artifact. A shipped project is the proof the loop worked: an idea your kid had, that they iterated on, that they decided to put their name on, that you signed off on.
• AI decisions reviewed. The count of ChangeDisclosure cards your kid reviewed before clicking Keep, Review, or Undo. This is the AI-literacy metric. Every reviewed card is a moment your kid sat with a proposed change and decided what to do with it.
• Iterations per project. Most published projects are shipped at v3, v4, v5. The iteration count tells you whether your kid is treating the project as a one-shot or a craft. The studio celebrates v7s.
• Wizards used. The count of recipe-style helpers your kid invoked to solve a specific problem. Higher numbers correlate with more ambitious projects.

None of those metrics measure duration. A kid who spends two hours and ships v4 of a game with thirty reviewed AI decisions is doing different work from a kid who spends two hours and ships nothing. The dashboard reflects the difference. The timer would not.

Why we still ship time-cap controls.

Parents asked for them. So we built them. The parental-controls surface includes daily caps, per-session limits, an optional weekly AI-request budget, and a lights-out window. If you want to cap your kid at 90 minutes a day, the dashboard does that. If you want the studio to refuse to open between 8pm and 7am, the dashboard does that too. The controls are real and they’re respected by the studio at runtime, not by a browser plugin that a determined kid can route around.

The controls live on a sub-page (/parents/controls) rather than on the family dashboard’s headline view. That placement is deliberate. The headline metric we want a parent to see when they open the dashboard is what their kid made, not how long they spent in front of the screen. The duration data is available in two clicks. It’s just not the lead.

We’re not the parent. We can’t tell you what the right cap is for your kid. What we can tell you, with some confidence, is that a cap by itself isn’t the whole conversation. A kid who spends their capped 90 minutes shipping a game is doing different work from a kid who spends their capped 90 minutes on TikTok. The 90-minute number is the same. The outcome isn’t.

The honest trade-off.

Some parents really do want a screen-time meter as the headline. We’ve heard the argument many times in cohort intake calls. It usually comes down to: "I don’t have time to evaluate what my kid is doing on the screen. I just need a number that tells me whether to call them down for dinner."

That argument is real and we take it seriously. Our answer is that the family dashboard tries to do better than a number, by surfacing a small set of signals that a busy parent can absorb in a glance. Three shipped projects this month. Forty-two AI decisions reviewed. A flagged-safety incident, if there was one (there usually isn’t). The dashboard is built to be readable in fifteen seconds, the same time a screen-time number would take. The signals are just richer.

If, at the end of all this, what you want is a timer, the timer is there. Set the cap. Set the lights-out window. We won’t fight you on it. What we won’t do is design the rest of the studio around the assumption that duration is the metric that matters. The kid who ships their first game at v5 doesn’t remember how long it took. They remember the URL. So do their friends. So does, eventually, the version of themselves that goes off to do harder work with the same tools. That’s the metric we’re building for.