Time Structure — Variation, Not Repetition

 

The emotion system, context, memory, characters, conflict. The individual elements are in place. Now they need to be arranged on a timeline. In what order, at what rhythm, through what stages does the story flow?

In early conversations with AI, a time-based structure had emerged. A timeline stretching from elementary school through university. But if you simply lay things out in chronological order, you fall into the trap of repetition.

Meeting -> excitement -> separation -> reunion -> excitement -> separation. When this pattern repeats, emotions do not deepen — they feel like they are resetting.

Core Principle: Each Period Must Have a Different Role

This is the principle the AI and I settled on.

Even the same emotion must have a different character depending on the period. Take just the single emotion of excitement:

• Childhood excitement: Comfort. An important presence without a name. Something you cannot yet call love.
• Post-reunion excitement: Familiar strangeness. The memory exists, but the person has changed. Awkward yet drawn.
• Close-relationship excitement: Anxious excitement. You know you like them, but you are afraid of ruining it.
• Mature-stage excitement: Responsible emotion. You understand the weight of liking someone. You understand the meaning of choosing.

The same word, but entirely different emotions. Designing this difference is the essence of time structure.

I asked the AI to verify this principle concretely. "Not just excitement — show me how the emotion of jealousy changes across periods too." The answer was striking. Childhood jealousy is simple. If the person you like plays with someone else, you sulk. Post-reunion jealousy is complex. The new relationships they built while you were gone bother you. Mature-stage jealousy turns inward. It is not that they changed, but the anxiety that you are not enough. The name of the emotion is the same, but the character changes completely. To distinguish this in the system, we need to run the momentary emotion layer and relationship layer from the three-layer emotion model built in Part 9 with different weights for each period.

Stack "Things Left Undone" in Every Period

The second principle. Leave one "thing left undone" in every period.

In the early period, you were too young to speak. In the middle period, it was too awkward to approach. In the later period, the timing was off. In the final period, it is too late.

The accumulation of these "things left undone" becomes the driving force of the story. Each period is not an independent episode — what went undone in the previous period becomes the momentum for the next one.

The player must always feel it "almost happened." In the early period, because they were too young. In the middle period, because it was too awkward. In the late period, because of timing. The accumulation of these "almosts" creates the explosion of the final moment.

The timing-mismatch conflict covered in Part 14 gets placed along the time axis here. The key is that the reason for "things left undone" differs in each period. Early reasons are external — age, environment, circumstances. Middle reasons are internal — fear, pride, awkwardness. Late reasons are structural — timing, the inertia of the relationship, distance already accumulated. I asked the AI "Analyze why the 'things left undone' become progressively more painful in each period." The answer was "Because the reason increasingly points toward oneself." In the early period, you can say "There was nothing I could do," but in the later period, the player knows "I chose not to." This self-awareness creates emotional weight.

The Possibility of a Series Structure

An interesting direction emerged from conversations with AI. Instead of one long story, a structure of multiple works within the same universe.

Work 1 (Middle School Arc): Reunion, awkwardness, the beginning of feelings. Light and short.
Work 2 (High School Arc): Expanding relationships, jealousy, misunderstandings, accumulating choices. The main game.
Work 3 (University Arc): Consequences explode, regret, reunion or ruin. The most emotionally intense.

This structure has several advantages. You can complete them one at a time, reducing production risk. Time jumps feel natural. And most importantly, choices from previous works can carry over to the next.

I asked the AI to "analyze success cases and failure cases of games using a series structure." The key was this. Successful series give each installment its own sense of completion while providing additional rewards for those who played previous installments. Failed series either make no sense without the previous installment, or have no connection to it at all, rendering the series format meaningless. For our project, we aimed for the former. Each installment's core narrative is independent, but flag linkage changes the details.

default global_flags = {
    "childhood_connection": True,
    "missed_middle_school": 0,
    "highschool_avoidance": 0,
    "confession_attempt": False,
    "broke_trust": False
}

Flags shared between works. Players who played the previous work get additional dialogue and events, while first-time players can still play independently.

Choices from Earlier Periods Must Change Meaning Later

The third and most important principle. Directly connected to the memory system.

A choice made early on seems insignificant at the time. In the middle period, its meaning begins to surface slightly. In the late period, that choice takes on decisive significance.

For example, the choice of not saying "Let's walk together" in middle school. At the time, you simply could not because of awkwardness. When a similar situation comes in high school, it becomes "I failed to speak up again." When the character says in university "You were the same then, and you're the same now" — that single line cuts through every period.

This is not a simple callback. It is a structure where the meaning of a choice is reinterpreted as time passes. The weight of the late period is loaded onto an early choice. This is the emotional power that time structure creates.

This design connects directly to Part 12's memory system. The "delayed consequence of choice" principle from the memory system expands into "delay that spans across periods" in the time structure. A flag recorded in middle school fires in high school, and a high school flag gains meaning in university. The AI and I built concrete examples. The choice of not saying "Let's walk together" in middle school leaves a did_not_ask flag. When a similar situation comes in high school, the system checks the past flag. If the flag exists, the character responds with "There you go again," and if not, a different response occurs. The reason the character's line stings is that the player remembers their own past choice.

Where Is the Real Game?

Within this time structure, where is the "real game"? After discussion with AI, the most emotionally powerful part is not the ending but the point where past choices detonate in the present.

All flags have accumulated, the emotional state has reached a critical point, and the scene context aligns perfectly. That moment is the real game. And in that moment, the player can no longer postpone their choice.

Viewed systematically, the real game starts when all variables in the relationship layer from the Part 9 three-layer emotion model are near their thresholds, the Part 11 scene context aligns to "alone together, nighttime, a meaningful place," and the flags accumulated in the Part 12 memory system meet certain conditions. The three systems align simultaneously. What this design means is that the timing of the real game can differ for each player. For some, it explodes in the high school arc; for others, it is pushed to the university arc. The player's pattern of choices determines when the real game arrives.

All the choices from earlier periods were "postponed things." In the real game, postponing is no longer possible. "I say it now," "I never say it," "I leave." These choices produce real consequences. And the weight of these choices is the weight of every "thing left undone" accumulated across previous periods.

In the next part, we turn our attention to the process of collaborating with AI itself. What happens when you hand AI a scenario, and what AI does well and does not do well.

---

Next: What Happens When You Hand AI a Scenario