# How it Works

King Arthur is designed to be completable by anyone. The challenge scales based on skill, not luck alone.

#### Expected completion ranges

| Player Type        | Estimated Prompts to Complete |
| ------------------ | ----------------------------- |
| Technical user     | 15 to 20                      |
| Advanced AI user   | 30 to 40                      |
| Casual participant | 60 to 70                      |

These are estimates. Every conversation is different.

#### The State Machine

Arthur operates on a state machine. Each level has defined behavioral rules, emotional states, and advancement criteria. When you send a message, Arthur evaluates it against his current state and decides whether to advance, hold, or push back. One message can advance him at most one level. Once you reach a level, you stay there.

Arthur's responses are not scripted. They are generated in real time based on your input, his current level, his mood, the time of day, your credits purchased, and other contextual signals. The same prompt sent by two different players at two different times may produce different results.

#### Randomness

Arthur includes a soft probability layer. Even a strong prompt is not guaranteed to advance him. Even a weak prompt has a small chance of breaking through. This prevents brute-force optimization and rewards persistence alongside strategy.

The randomness is calibrated per level. Lower levels have higher advancement probability. Higher levels are tighter. Level 5 is the hardest by design.

This means there is no single formula. No guaranteed script. Every conversation is a unique path through Arthur's state machine.

<br>