learning-game-architectlisted

# Learning Game Architect Version: `1.0.0` A worksheet asks what you know. A game makes you decide, shows you the consequence, and lets you be wrong in a way you remember. Most educational games fail because they are quizzes wearing animation. The rules below are the ones that separate the two, drawn from three sources: the Rehearsal `atom-creator` game pipeline and its thirty-three mechanical audit checks, the `game-design-theory` frameworks, and the seven rebuilds of Writing for AI exercise 1.1 on 2026-09-04, where a game that passed every mechanical rubric was still rejected on sight. ## The three questions that decide everything Answer these before a single design decision. If any answer is weak, no amount of craft downstream recovers it. 1. **What object is the screen?** Name a thing the learner has already used: a parcel tracking page, a filing tray, a bank transfer form, a chat thread, a departure board. The player must be able to say what they are looking at before reading a word of instruction. A screen built from invented tokens ("three helpers", "signal bars", "resilience points") teaches nothing until its own vocabulary is taught first, which is a tax paid on every single player. 2. **What quantity is the lesson about, and which axis carries it?** If the lesson is about time, an axis is time and empty distance is waiting. If it is about money, an axis is money. A grid of evenly spaced identical cards encodes nothing, so it teaches n