Invent a Board Game and Argue About the Rules

Every homemade board game has a rule that breaks the moment somebody tries to win with it. Finding those before you play is a real design job, and a chatbot is a tireless pedant.

How it goes

1. Decide how someone wins before anything else. Not the theme, the win condition. Get to the end, collect six things, be last standing.
2. Draw the board. Big, on card, by hand. A track with squares is fine for a first game.
3. Write the rules on a separate sheet, numbered. Between eight and twelve rules is the workable range.
4. Let the AI hunt for loopholes: Here are my board game rules. Find three ways a player could exploit them or get stuck forever. Do not rewrite the rules. This is genuinely useful; models are good at spotting under-specified cases.
5. Your child patches the rules themselves, in their own words.
6. Playtest with a real person who has only the rules sheet and no explanation.

Who does what

AI finds loopholes, spots rules that contradict each other, and asks what happens in cases nobody considered.

Your child sets the win condition, draws the board, writes and patches every rule, and runs the playtest.

You play to win, exploit anything exploitable, and take the resulting rule change gracefully.

Make it easier or harder

For a six-year-old: use a plain track of twenty squares and one die, and let the design work go into what happens on six special squares. Inventing a whole board is overwhelming; inventing six events is a good afternoon.

For a twelve-year-old: require a catch-up mechanism, so the player in last place has a route back. Games without one stop being fun at the halfway point, and noticing that about their own design is a genuine step up in thinking about play rather than rules.

Take it off the screen

Everything after step four is cardboard and arguing. Make a proper box for it from a cereal packet with the title on the front.

Take it to a friend’s house. A game that survives contact with someone else’s children is properly finished.

When it goes sideways

The first playtest often ends in an argument about a case the rules do not cover. That is the game working as intended. Stop, write the missing rule on the sheet in front of everyone, initial it, and carry on. A rules sheet with six handwritten amendments is a finished design document.

If the AI’s loopholes are all trivial, the rules are probably too vague to exploit rather than watertight. Ask it instead: What is the fastest possible win under these rules? A three-turn win reveals more than a list of edge cases.

One thing to watch

Do not let the AI write the rules. Rules written by a model read smoothly and hide the thing that matters, which is that your child thought through every case themselves. Loophole-hunting is the role that keeps the design theirs.

Related: creature battle cards is the card-game version, and invent a toy and write its manual uses the same “test it on a real person” step.