A Design Where Love Is Never Directly Increased

 

In the previous part, we designed the three-layer emotion model. Among its principles, one stands above the rest. The variable called love does not exist in the system.

There is no love += 1. Instead, love "emerges" from a combination of other variables. This is the core of this system and the biggest difference from conventional affection-meter systems.

Why Love Should Never Be a Direct Variable

When I discussed this topic with AI, it offered an interesting perspective.

"Love is not on the same level as other emotions."

Sadness, anger, jealousy, fear. These are momentary emotions. They rise or fall in response to specific events. But love is different. Love is closer to the result of accumulated relationship interpretation than a momentary emotion.

Trust is high. Dependency exists. Intimacy has built up. There is a touch of jealousy. Fear of loss is present. The player repeatedly chooses the other person as a priority. The player reads this combination as "love."

Expressed systematically, it looks like this:

love_state = (
    trust >= 70 and
    intimacy >= 60 and
    fear_of_loss >= 30 and
    chosen_priority >= 5
)

Instead of directly increasing love, we design the conditions under which love emerges. This distinction matters enormously.

Why This Approach Feels More Natural

Think about real life. Can anyone pinpoint the exact moment they fell in love with someone? Most people cannot. One day you simply realize, "Oh, I have feelings for this person." Love is not about pressing a start button — it is about recognizing something that has already accumulated.

A love += 1 system cannot reproduce this experience. Because love goes up with every choice, the player starts calculating, "My love score must be around X by now." It becomes score management, not emotion.

The condition-combination approach is different. The player does not know the love value. The variable called love does not even exist. Instead, sensations accumulate separately: "I've come to trust this character," "Being with them feels comfortable now," "I wouldn't want them to disappear." Then, at some point, the story presents that combination as "love."

This is far closer to how it feels in real life.

When I asked the AI to "express the process of falling in love systematically," it showed a sequence like this. At first, there is only curiosity. Over time, comfort builds. At some point, the absence of the other person becomes uncomfortable (fear_of_loss rises). Then the player repeatedly makes choices that prioritize the other person (chosen_priority accumulates). Somewhere along this process, you realize, "Oh, this is love." Not from a single event, but as the result of accumulation.

Translating this into story rather than code is the scenario writer's job.

The Diverse Relationships That Combinations Create

Another strength of this design is that the same "liking" manifests differently.

Attracted but unable to trust — high affection, low trust. Being together is exciting, but you cannot rely on them at a critical moment.

Precious but not romantic — high trust, low affection. The conviction that this person will never change, no matter what. But the heart does not race.

Strongly drawn but dangerous — high affection, high tension. Passionate but full of conflict. Intense when together, but unstable.

Someone too precious to lose — high respect, high fear_of_loss. The boundary between romance and friendship is blurred. But you cannot imagine them disappearing.

In an affection-meter system, all four of these are the same "affection level 70." In the three-layer system, they are entirely different relationships. And these different relationships create different stories and different emotions.

State Transitions: The Flow of Emotion

The AI proposed one more important concept. Design state transitions before numerical values.

What matters is not the numbers in emotion variables, but which state a relationship moves from and to.

Positive transitions:
Wariness → Curiosity → Relief → Trust → Dependency → Attachment → Love

Negative transitions:
Wariness → Interest → Expectation → Disappointment → Anger → Conflict → Disconnection

Even from the same starting point, paths diverge depending on the events experienced. The right order is to design these transition flows first, then define "which variable combination corresponds to which state."

This resembles the State Machine pattern — a familiar concept for developers. The idea of managing emotions as a state machine felt mechanical at first, but when you think about it, the flow of a story is itself a series of state transitions. A relationship deepening is a state transition, and a relationship breaking down is also a state transition.

The Many Forms of Love

In this system, love is not monolithic either. Different condition combinations produce different kinds of love.

Trust-based love — high trust and intimacy, low fear_of_loss. Stable and comfortable love. Not dramatic, but solid.

Anxiety-based love — high affection and fear_of_loss, low trust. Intense but anxious love. It can border on obsession.

Admiration-based love — high respect and admiration, moderate intimacy. Love that originates from looking up to someone. It can be difficult to transition into an equal relationship.

Love amid conflict — high affection and tension simultaneously. A relationship where you care for each other but do not mesh. A cycle of reconciliation and collision.

Each of these creates a different story and leads to a different ending. The answer to "We were in love, so why didn't it work out?" changes. One falls apart because of anxiety. Another because admiration prevented equality. Another because conflict eroded trust.

Points of AI Collaboration

Three aspects of this design made AI particularly useful.

First, consistency checks between variable relationships. AI provides logical answers to questions like "If trust increases in this event, what should happen to tension?" Humans rely on intuition but miss things. AI does not miss them.

Second, simulating the combinatorial possibilities. If six variables each range from 0 to 100, the combinations are practically infinite. When you ask AI, "What relationship state does this variable combination represent?" it produces a realistic interpretation.

Third, writing definition documents. Documenting the meaning, change rules, and interactions of each variable is a massive task, and AI drafts it quickly.

What humans must do is this: "Does this relationship actually feel right?" Judging whether the system's output is emotionally convincing. This is something AI cannot do. No matter how correct the numbers are, if the feeling is wrong, it is meaningless. The final validation of an emotion system must always be human intuition.

In the next part, we will cover the third element of this emotion system. The structure by which scenes push emotions forward — a context system where the same dialogue carries entirely different meaning depending on the situation.

---

Next: Scenes Push Emotions Forward — Designing the Context System