A kid who has spent a year or two on Scratch has internalized real things: event-driven thinking, state, the run-it-and-see iteration loop. Tell and Show isn’t a replacement for that. It’s a different artifact-shape. Scratch teaches a kid how computation works. Tell and Show lets them ship software a stranger can use. The two fit together, and a Scratch kid arrives in Tell and Show with most of the hard part already done.
What Scratch is great for.
Before anything else, the credit. Scratch is one of the most carefully designed pieces of software ever made for kids learning to think with computers. Mitchel Resnick and the Lifelong Kindergarten group at the MIT Media Lab shipped it in 2007 and have been refining it ever since.1
The design principle behind Scratch, articulated by Resnick’s group as low floor, high ceiling, wide walls, is the cleanest statement of what kid-creative software should be. The floor is how fast a beginner makes something happen. The ceiling is how far an expert can push the medium. The walls are how many kinds of projects fit inside one tool. Scratch is the platonic example of all three.1
A kid who has spent 18 months on Scratch has internalized a set of ideas that are genuinely durable, even if no one ever named them aloud. Events fire when something happens. Loops repeat behavior. Conditionals branch the flow. State is the thing the program remembers between frames. The fastest way to find a bug is to run it and watch what happens. None of these are toy ideas. They are the same primitives a professional software engineer uses, dressed in a vocabulary an eight-year-old can read.
That lineage runs back further than Scratch itself. Seymour Papert’s LOGO turtle in the 1970s was the original constructionist tool, and Papert’s argument was that kids learn most durably when they build something they care about for an audience they care about.2 Scratch carried that idea forward. Tell and Show carries it forward again. We’re in the same tradition, and we owe both Papert and Resnick the obvious debt.
What Scratch doesn’t address, without judgment.
Scratch is a sandbox, and that’s a design choice, not a flaw. The shareable output of a Scratch project is a Scratch project. It runs in the Scratch player, on the Scratch site, for an audience of other people who also have Scratch accounts. The medium is the museum. The kid’s work lives inside the curated world that Resnick’s team built around it.
For a lot of kids that’s the right shape. The gallery is one of the most important parts of Scratch’s design; other kids see the work, remix it, leave comments. It’s a peer-of-makers audience, which is a rare and valuable thing.1
But it isn’t a real URL a grandparent can open from a Gmail link. It isn’t a real app a friend can pull up on their phone without an account. The kid who finishes something in Scratch shares it inside the Scratch world. The kid who wants their stuff to reach further than that has to leave the sandbox to do it.
Yasmin Kafai and Quinn Burke draw a related distinction in their work on what they call connected gaming: the difference between making projects that live inside a learning tool and making projects that participate in the broader networked world kids actually inhabit.3 Both are legitimate. They are not the same.
The Scratch team is aware of this, and the studio has always positioned itself as a starting point. Resnick has written across his career about the eventual transition from Scratch to text-based languages like Python or JavaScript when a kid is ready to push beyond the block vocabulary.1 The honest framing is that Scratch is the first medium, not the only one.
Where Tell and Show fits. Not after Scratch, alongside it.
If you read the previous section as “so a kid should move on from Scratch,” that isn’t the argument. The argument is that Tell and Show is a different artifact-shape, and the kid’s Scratch fluency transfers directly to it.
The thing that transfers is the part of programming nobody can teach in a weekend. Event-driven thinking. Mental models of state. Debugging by running it and watching. Those are general-purpose habits, and a kid who has them already arrives in Tell and Show with the hard part finished.
The thing that’s new is shipping. In Tell and Show the artifact is real. It deploys to a real URL on real web hosting. A parent emails the link to a grandparent and the grandparent loads it in Safari. A kid’s friend pulls it up on the school bus. The kid’s work participates in the same internet adults participate in, which is a different experience from the Scratch player and a different relationship to one’s own work.
The other new thing is the AI partner inside the loop. Scratch deliberately doesn’t ship one; the kid translates ambition into the block vocabulary themselves, and that translation is part of the pedagogy. Tell and Show was designed around an AI partner that proposes scoped changes the kid keeps, reviews, or undoes. The pedagogy is different, and we’ve written about it at length in Visible AI is the whole pedagogy.
Here’s the comparison laid out cleanly:
A Scratch kid arrives in Tell and Show with most of the hard part already done. They’ve internalized how computation works. The new thing is making something a stranger can use. On what transfers and what’s new
What the transition looks like in practice.
Imagine a kid who has been on Scratch for 18 months. They’ve shipped a side-scroller, a remix of a cat-clicker, and a multi-room interactive story. They open Tell and Show for the first time and pick the Game track.
What’s familiar. The iteration loop. The idea-test-fix rhythm. The instinct to run the project after every change, the muscle memory of catching a bug by watching the thing happen. The expectation that events fire, that conditionals branch, that state is the thing the program remembers between frames. Scratch kids arrive with all of this and we don’t have to teach any of it. Anecdotally, the onboarding goes faster; we’ve seen Scratch alums hit their first published artifact in noticeably less wall-clock time than Scratch-naive kids, though we don’t put a percentage on it.
What’s new. Three things in particular.
First, the AI partner. In Scratch the kid does all the translating themselves; ambition becomes blocks because the kid drags the blocks. In Tell and Show the kid says what they want in plain English and Inkie proposes a change. The kid reads the proposal on a ChangeDisclosure card and decides whether to keep it, ask for a different take, or undo. This is a new skill, and it’s the central skill of working with AI as a partner rather than a ghostwriter. We cover the design in detail in Iteration is the move.
Second, the artifact is real. There’s no Scratch player wrapping it. The output is HTML, CSS, and JavaScript that runs in any browser. The kid who wants to read the source can read the source; it’s a text file, not a block tree. For a kid who’s been wanting to peek behind the curtain, this is a significant moment.
Third, the deploy step. When the kid is happy with the project, they hit publish, a parent approves, and the artifact goes to a real URL. The grandparent loads it in Safari. The kid’s relationship to their own work changes the first time someone outside the household uses it. Papert called this the “public entity” effect; making something with a real audience is qualitatively different from making something for a class folder.2
Not graduating from Scratch. Adding a second medium.
The framing we’d argue against is the one that says a kid has to leave Scratch to use Tell and Show. They don’t. The two tools serve different moments and they coexist fine.
A kid might keep Scratch as the place for a weekend remix or a fast animation experiment, and use Tell and Show for the bigger projects they want their cousin to actually play. A kid might use Scratch in school where it’s the assigned tool, and Tell and Show at home where they have control of the artifact. A kid might gradually phase out Scratch as their ambitions expand past what the block vocabulary captures, or they might never phase it out. All of those are fine.
Does your kid already think in events, loops, and state?
Then the mental model is already there. The new thing is shipping.
Jump to where Tell and Show fits →Does your kid ask “what comes next?”
Scratch is honest about being a starting point. The transition is part of the design.
Jump to what Scratch leaves open →Does your kid want their friends to play their stuff?
The real-URL artifact is the part of the experience that changes a kid’s relationship to their work.
Jump to the transition →Is your kid curious about what’s under the blocks?
Tell and Show’s output is HTML, CSS, and JavaScript. Real text files, readable.
Jump to the comparison table →The decision isn’t a swap. It’s additive. We’d be skeptical of any tool whose pitch was “graduate from Scratch.” Scratch is the canonical kid-creative environment and it isn’t something to outgrow on a schedule. What we’re offering is a second medium for the kid who already has the first one, where the new things are an AI partner inside the loop and a real artifact at the end.
So no, it isn’t a step backward. The fluency a kid built in Scratch is exactly the fluency Tell and Show assumes. The new skill is shipping. The new partner is Inkie. The same kid does the choosing.
References
- Mitchel Resnick, Lifelong Kindergarten: Cultivating Creativity through Projects, Passion, Peers, and Play, MIT Press, 2017. See also the MIT Media Lab’s Lifelong Kindergarten group at media.mit.edu/groups/lifelong-kindergarten for the canonical low-floor / wide-walls / high-ceiling framing and the history of Scratch’s development since 2007.
- Seymour Papert, Mindstorms: Children, Computers, and Powerful Ideas, Basic Books, 1980. The origin of constructionism, the LOGO turtle, and the “public entity” argument that making something for a real audience is qualitatively different from making something for an assignment.
- Yasmin Kafai & Quinn Burke, Connected Gaming: What Making Video Games Can Teach Us About Learning and Literacy, MIT Press, 2016. Their distinction between constructionist gaming inside a learning tool and connected gaming in the broader networked world is the framing for why artifact reach matters.