# Nostalgia Codec — Prototype 001: The Market After Summer

## Intent

Create a **60–90-second, mobile-first, playable memory**: the player discovers a small, inviting forest market, leaves briefly, then returns to the **same place, irreversibly changed**. Test whether recognition, absence, and a remaining mystery evoke attachment to somewhere the player visited less than a minute ago. Inspired by the *emotional and spatial signatures* of **Secret of Evermore**, **Secret of Mana**, and **Illusion of Gaia**—never their assets or layouts.

**Hypothesis:** A changed return produces a stronger feeling of personal history than an unchanged return.

**Form:** One small map · two states (summer → autumn) · three memorable landmarks · one optional discovery · one original musical motif in two variations · movement + one interaction · Phaser/web · no runtime AI.

## 15 DOs

1. **Do** make the complete experience playable in 60–90 seconds.
2. **Do** establish a welcoming, distinctive place before changing it.
3. **Do** use one compact map revisited, not two separate levels.
4. **Do** anchor memory with three visually recognizable landmarks.
5. **Do** make the initial market feel inhabited without extensive dialogue.
6. **Do** let movement and curiosity drive the first discovery.
7. **Do** lead the player naturally from market to grove and back.
8. **Do** trigger the seasonal change through one simple interaction.
9. **Do** preserve the layout while changing meaningful sensory details.
10. **Do** leave one small familiar element untouched after the change.
11. **Do** reveal one previously obscured possibility on returning.
12. **Do** use an original short melodic motif transformed on return.
13. **Do** support touch, keyboard, mute, and immediate replay.
14. **Do** ship a fixed, authored baseline before introducing generation.
15. **Do** test changed versus unchanged returns and ask open-ended questions.

## 100 DON'Ts

### I. Scope & product shape (1–10)

1. **Don't** expand the prototype into a complete RPG.
2. **Don't** add combat or enemies.
3. **Don't** add inventories, equipment, or currency.
4. **Don't** add quests, objectives, or quest markers.
5. **Don't** add leveling, upgrades, or character statistics.
6. **Don't** build a dialogue system.
7. **Don't** add more maps to compensate for weak atmosphere.
8. **Don't** turn the prototype into a procedural world generator.
9. **Don't** require accounts, saves, or a backend.
10. **Don't** spend effort on menus, achievements, or metaprogression.

### II. Space & navigation (11–20)

11. **Don't** make the market an empty decorative backdrop.
12. **Don't** make the world too large to recognize on return.
13. **Don't** make every area visible from the starting point.
14. **Don't** build a maze that requires trial-and-error frustration.
15. **Don't** let landmarks blend into repeated scenery.
16. **Don't** use identical-looking paths at every junction.
17. **Don't** obscure the return route without a good reason.
18. **Don't** teleport the player to a visually unrelated location.
19. **Don't** rely on a minimap to establish spatial memory.
20. **Don't** make the hidden detail visible before the return.

### III. Interaction & discovery (21–30)

21. **Don't** explain the mechanics in a tutorial paragraph.
22. **Don't** require precise timing or fast reactions.
23. **Don't** introduce multiple action buttons.
24. **Don't** make every interesting object interactive.
25. **Don't** require collecting items to progress.
26. **Don't** use puzzles that stall the 90-second experience.
27. **Don't** reward exploration with numerical points.
28. **Don't** mark the discovery with a giant visual notification.
29. **Don't** make the optional discovery mandatory for completion.
30. **Don't** prevent players from lingering or looking around briefly.

### IV. Change, return & recognition (31–40)

31. **Don't** replace the market with a different map.
32. **Don't** change every landmark beyond recognition.
33. **Don't** make the before/after difference merely a color filter.
34. **Don't** change only the background while all details remain identical.
35. **Don't** destroy everything to communicate loss.
36. **Don't** erase all signs that anyone lived there.
37. **Don't** tell players explicitly what they should remember.
38. **Don't** reset the world to summer during the first run.
39. **Don't** make the transformation feel like a random weather toggle.
40. **Don't** resolve every mystery on the return visit.

### V. Art direction & references (41–50)

41. **Don't** reproduce copyrighted sprites, tiles, maps, or characters.
42. **Don't** trace recognizable environments from the reference games.
43. **Don't** confuse pixelation alone with nostalgia.
44. **Don't** use a generic fantasy asset pack without adaptation.
45. **Don't** overdecorate every tile with equally strong visual noise.
46. **Don't** saturate every object with the same color intensity.
47. **Don't** use modern glossy UI over a restrained pixel world.
48. **Don't** make the autumn state uniformly gray and lifeless.
49. **Don't** sacrifice landmark readability for atmospheric effects.
50. **Don't** pursue elaborate animation before composition works.

### VI. Sound & silence (51–60)

51. **Don't** reuse melodies or samples from existing games.
52. **Don't** use familiar game sound fonts as a nostalgia shortcut.
53. **Don't** add continuous music that overwhelms ambient cues.
54. **Don't** make the return motif completely unrecognizable.
55. **Don't** signal sadness with an obvious dramatic soundtrack.
56. **Don't** use loud stingers for the seasonal change.
57. **Don't** attach an audio cue to every step or touch.
58. **Don't** depend on stereo headphones for comprehension.
59. **Don't** let audio fail block progression.
60. **Don't** start sound before the browser permits playback.

### VII. Story & emotional restraint (61–70)

61. **Don't** explain the place's history in exposition.
62. **Don't** introduce named characters who require backstories.
63. **Don't** manufacture tragedy through death or catastrophe.
64. **Don't** state that the place is supposed to feel nostalgic.
65. **Don't** use sentimental on-screen narration.
66. **Don't** make an NPC explicitly mourn the earlier world.
67. **Don't** force the player to perform a morally weighty choice.
68. **Don't** reveal the cause of every environmental change.
69. **Don't** make the secret a lore dump or cutscene.
70. **Don't** mistake melancholy for the only possible nostalgic emotion.

### VIII. Controls & accessibility (71–80)

71. **Don't** require a gamepad.
72. **Don't** depend on hover interactions on touchscreens.
73. **Don't** position controls against mobile browser navigation edges.
74. **Don't** require tiny touch targets.
75. **Don't** lock progression behind a precise gesture.
76. **Don't** make the camera drift during directional movement.
77. **Don't** use camera shake as the primary transformation effect.
78. **Don't** require sound to find the way forward.
79. **Don't** encode critical state only through color differences.
80. **Don't** hide restart or mute behind multiple screens.

### IX. Engineering & runtime (81–90)

81. **Don't** add an LLM call to the experience runtime.
82. **Don't** require network requests after initial asset loading.
83. **Don't** add physics where tile collision is sufficient.
84. **Don't** build a custom engine when Phaser can handle it.
85. **Don't** mix world state with rendering-specific side effects.
86. **Don't** let the two seasonal states diverge in collision layout.
87. **Don't** preload unnecessarily large textures or audio files.
88. **Don't** allow mobile frame drops to interrupt the return moment.
89. **Don't** make transition or replay depend on page refresh.
90. **Don't** create a reusable content framework before proving the scene.

### X. Experiment & handoff (91–100)

91. **Don't** ask participants whether they felt nostalgia first.
92. **Don't** tell testers which classic games inspired the prototype.
93. **Don't** assume visual resemblance proves emotional success.
94. **Don't** compare only against players who already love SNES games.
95. **Don't** ignore players unfamiliar with the references.
96. **Don't** change multiple variables between test conditions.
97. **Don't** evaluate only completion rate or session duration.
98. **Don't** claim statistical proof from a tiny qualitative sample.
99. **Don't** feed the generative codec untested aesthetic guesses as facts.
100. **Don't** expand scope until the changed-return effect earns it.

## Acceptance / test

- **A:** Enter summer market → pass three landmarks → reach grove → touch stone → return through same geography → recognize changed landmarks and notice an altered musical motif → discover one new visual possibility.
- **B (control):** Identical journey, timing, and interactions, but the market is unchanged on return. Hold other variables constant as far as practical.
- **Observe:** Where players pause; whether they recognize landmarks; whether they voluntarily inspect familiar details; what they remember unprompted.
- **Ask afterward:** “What did the place make you feel?” “What do you remember most?” “Would you want to return? Why?”
- **Success signal, not proof:** Players in A spontaneously describe recognition, longing, loss, warmth, or a changed sense of the place more often than in B.

**Next step:** Hand this file to a coding agent and build **A and B**. Distill the effective scene into a portable Prosemap Flow seed *only after* observing the response.