A library is curated. Here's what didn't make the shelf.

A library is a place that has thought about what is worth knowing. Somebody chose what's on the shelf. Somebody chose what isn't.

The second part matters as much as the first, and most programs never talk about it. We're going to talk about it. Here is what we looked at hard and did not put in the room.

Leaderboards comparing your kid to a stranger

We tested a version of the activities room with leaderboards. Watched what happened. What happened was: the kids who were already winning kept winning, and the kids who were not yet winning stopped opening the activity. A leaderboard against strangers is a participation filter dressed up as motivation. It motivates the kids who don't need motivation and demotivates the kids who do.

If we ever do leaderboards, they will be your kid against her own past self. That's the one that works.

Streak counters

The streak counter is the most successful product mechanic of the last decade. It is also a guilt machine. You haven't visited in 3 days. You're about to lose your 47-day streak. Kids respond to it the way adults respond to it — they show up to preserve a number, not because they want to be there. We do not want kids in our room because they are scared of losing a number. We want them in our room because they want to build a thing.

We left the counter out. The cost is some daily-active-user metric we will never have. The benefit is that the kids who are in the room actually want to be in the room.

"AI buddy" chatbots that fake friendship

Several companies in our space have built kid-facing chatbots that pretend to be a friend. Hi, I'm Spark! What did you do today? Long conversations. Saved memory. Personality.

We rejected this category outright. A kid does not need a fake friend; a kid needs a real one and a working builder sitting next to them. The chatbot-as-companion product is, in our reading, a slow-acting harm. We will not build it and we will not put one in the room.

AI in the room is a tool. The kid prompts it, the kid edits its output, the kid ships the result. It does not greet the kid by name. It does not ask how her day was. It is a Pokédex page, not a Pokémon.

Mascots

We tested mascots. A friendly bot character. Owl with glasses. Cartoon dog with a lab coat. The whole thing.

Mascots work in marketing. Mascots do not work in a room where kids are supposed to be doing serious work. The eight-year-olds liked the mascot. The thirteen-year-olds rolled their eyes at the mascot. The mentors had to talk around the mascot. So the mascot lost.

The room has good lighting, decent chairs, finished projects on the wall with kids' names on them. That's the warmth. We didn't need a cartoon.

"Pro" gates halfway through an activity

You know the pattern. The kid is three problems into a math drill and a popup says unlock the next set for $4.99/month. Every time we have seen this in a kids' product, we have flinched. It teaches the kid that the world is full of small paywalls and that her curiosity is something to be charged for in five-dollar increments.

The activities in the room are not paywalled mid-activity. The room itself is a membership, which we are honest about. Once you're in, nothing in the room asks you for more money. That's the deal.

Gamified "XP" for ordinary work

Adding +10 XP for finishing a problem is the candy-coating model of learning. It moves the reward from the thing you built to the meta-progress bar. Over time it teaches the kid that the work is the boring part and the bar is the point.

We wanted the opposite. We wanted kids to learn that the work is the reward — that shipping the card game and watching your sister play it is the bar. So we left the bar out.

A whole bunch of subject-matter content

We're a library. Libraries don't carry everything. Things that did not make our shelf, on purpose:

• Test-prep drills. Other people do this well; the room isn't for it.
• Generic "intro to coding" curriculum that's been re-shot ten times on YouTube. If you can already get it free, we are not adding signal by re-recording it.
• Subject-specific apps for things the kid will encounter in school anyway. We focus on the part school doesn't reach.

The general rule

The rule we use, when we are deciding whether a thing belongs in the room, is the same rule we use for everything else: would a working builder, sitting next to a kid for an hour, recommend this thing?

Most things, the answer is no. That's why the shelf is short and that's why the things on it earned their place. When something new earns it, on it goes. When something on it stops earning it, off it comes.

A library that puts everything on the shelf is a warehouse. We are not a warehouse.