Idea still being shaped
Game Generation: Change One Rule
What changes when knowledge earns one move, claim, or steal in a familiar game?
A familiar game gets one changed rule and only enough prototype to test it. The record of actual play, confusion, choices, and rule exploits guides the next revision.
Why this idea may matter
Why this might be worth exploring
It puts human playtesting ahead of generated code and makes one mechanic’s effects inspectable.
Question to choice
One possible path
- 01
Choose a game
Choose a familiar game.
- 02
Change one rule
Change one rule.
- 03
Build a tiny version
Build only enough to test it.
- 04
Playtest
Playtest.
- 05
Record and revise
Record behavior and revise before expanding.
Possible participant artifact
What might leave the table
A tiny playable prototype and a one-page playtest record.
Keep the checking visible
Evidence and checking
- Observed play behavior, confusion, choices, and rule exploits matter more than generated code volume.
Where AI might help
Propose rule variants, generate a minimal prototype, or help summarize playtest observations.
What people still decide
Establish that a game is fun, fair, understandable, safe, or ready to publish.
Trust boundaries
- One mechanic at a time.
- No hidden data collection.
- Human playtest before expansion.
Still a working idea
What is still open
- First familiar game.
- Paper versus browser prototype.
- Answer-to-claim versus answer-to-steal.
Curiosity Capability Atlas