The rules we build by

The best study of expert word-game players found something surprising: playing word games trains you to recognise letter patterns — but not what the words mean. So in our word games we do the opposite: a word only counts once you've shown you know its meaning. You can't play a word you can't mean — and every word you meet joins your own list, on your device.

Every child's voice sits in a different, ever-changing range — and asking a kid to sing a note they can't reach is exactly how singing stops being fun (and it's hard on a young voice). So our singing apps do the opposite: we find the notes that feel just right for your child first, then sing every little song inside that range. Your child is never asked for a note they can't reach — and we remember only the notes, never a recording of their voice, right there on the device.

Grown-up games carry a hidden weight — how heavy they are to think about. We don't guess ours; we measure it. A small robot plays each game against itself hundreds of times and counts how many real choices a turn actually holds, and we print that honest weight — one to five — right on the game's page. A child meets each idea at the lightest weight it can be held. And every point a game pays out can be traced on screen back to the choice that earned it — we never dress a heavy game up as a light one, and we never hide where a score came from.

A feed can make swiping feel like learning. Ours can't — an idea only joins a child's shelf once they say it back in their own words, out loud or in their own typing. Nothing is kept until the child has made it theirs. We never grade what they say and we keep no recording of it; the saying-it-back IS the whole gate. A shelf full of a child's own words is a shelf they actually remember — the opposite of a folder of clippings no one ever reopens.

Every idea we hand a child traces back to a real source — and a summary can quietly betray it, stretching a careful claim into a bold one or adding something the source never said. So we audit our own writing the way we ask children to: every idea in our shared library is checked against its source and marked faithful, stretched, or invented, and nothing stretched or invented is ever kept. FidelityForge hands a learner the very same audit, because the surest way to trust a summary is to learn how to catch one that has drifted from what the source actually said.

Most kids' apps are built to keep you tapping — streaks that punish a day off, autoplay that never lets you stop, badges just for showing up. We do the opposite: no streaks, no autoplay, no engagement traps. A badge here only ever means you learned something — read a chapter, finished a mission — never that you came back. Kids play until they're done, then close the app and go be a kid.

Lots of games quietly run a clock against your child: a crop that takes hours unless you pay to skip the wait, a garden that wilts if you don't come back today, a streak that punishes a day off. We do the opposite: nothing you grow ever wilts, and no wait can ever be bought. Your child can come back in an hour or next month and find their garden exactly as they left it — or a little further along. Nothing rots, nothing nags, and nothing costs a coin to hurry.

Every list on our shelves has a visible end. The feed of ideas reaches a bottom, on purpose, and simply says so — nothing scrolls forever, and there is no streak asking a child to come back tomorrow. Reaching the bottom is a small, calm win: you have seen everything there is for now, and you are free to close the app. A stopping point a child can actually reach is the kindest thing a feed can offer.

When a child builds a world on the big platforms, it lives on a company's servers — so it needs an account, needs the internet, and the day those servers switch off, the world is gone. We do the opposite: a world your child makes is a file — a little code they keep, right on the device. It opens with no account and no server, works with the internet off, and is still there years from now. Build a place, save its code, and it's yours — to keep, or to hand to a friend.

In our island games the dice are honest. Seven comes up most often simply because there are more ways to roll a seven — not because the game wants you to win or lose. We never tilt the odds to keep a child playing. Instead we let them feel the real shape of chance: the taller a number's stack of ways, the more often it lands. And we score what a child understands about those odds, never how lucky the dice happened to be.

On the island you can only ask for what your next build actually needs — an ask with no purpose isn't a real ask. It keeps every trade honest and kind: a good deal has to help both sides. Our trading characters never bluff, never squeeze, and never talk a friend out of what's fair. The best offer on the island is the one that leaves everyone better off than before.

When a storm sets someone back, the island shares. There is no robber, no stealing a rival's goods, and no everyone-loses-half. What one player spills, the island gathers up and hands to whoever has the least. Helping the player who's furthest behind is the island's kindest habit — and it's how the whole board rebuilds the fastest, together.

A score from three tries and a score from three hundred are not the same, even when the averages match. So wherever we rank or rate anything, we show the prior — the quiet head start every newcomer is given until they have played enough to speak for themselves. A child learns to ask not just who is on top, but how sure we can be. That single habit is what separates a real conclusion from a lucky streak, and it is why we never let a handful of tries shout over a whole season of them.

Some of our games are won together, not against each other. In those, the score belongs to the whole table — there is no hidden tally ranking friends against friends. The helper who plays alongside a child never quietly wins the game for them. It suggests, it asks, and it steps back — but it never carries the team. Working something out together is the whole point, so the only score we keep is the one the whole table shares.

We never measure a child by minutes on a screen. We count what actually shows learning — ideas a child says back in their own words, circles turned (each back-and-forth with a grown-up or a character), and how often a session ends because the child finished, not because we tugged them to stay. Every number we show carries its plain prior — what it is measured against — so a figure can never quietly become a boast. And we never market a prodigy: no "2×", no "top 1%", no "your child is ahead". Time saved is time we hand back to the child — never a score to chase.