RollerCoaster Tycoon Codexery

Guest AI

Every guest is a tiny, opinionated, hunger-driven critic of your park.

Guest AI is the simulation system that brings the park's visitors to life in RollerCoaster Tycoon. Rather than being static decorations, every guest is an autonomous agent with internal needs, preferences, and a decision-making loop that governs how they enter the park, choose rides, buy snacks, react to their surroundings, and ultimately decide whether to leave satisfied or furious. They are the living feedback loop of the game: the player builds, and the guests respond.

The system spans pathfinding, utility-based ride selection, a layered 'thought' and emotion model, and a money-and-need economy. Understanding how guests think—what makes them happy, scared, sick, or lost—is as central to mastering RollerCoaster Tycoon as designing a great coaster itself.

Type
AI-driven visitor entity
First appears
RollerCoaster Tycoon (1999)
Role
Simulated park guest; primary player-feedback mechanism
Affiliation
Park visitors (non-player characters)
Status
Core mechanic across all mainline RCT titles

Lore & Background

In the world of RollerCoaster Tycoon, a park is only as good as the people who walk through its gates. The guest AI was designed to make each visitor feel like a real person with a day to fill: they arrive with a certain amount of money, a tolerance for intensity, a preference for thrill or gentle rides, and a stomach that will demand a hot dog by mid-afternoon. They chatter, they complain, they get motion-sick on a particularly loopy coaster, and they tip their hats (or throw them) at the end of the day.

The system rewards the player for thinking like a guest. A beautifully themed entrance means nothing if the queue for the flagship ride snakes past the only water fountain in the park. A five-star food stall is useless if the pathfinding grid funnels every visitor past a muddy patch. The AI does not judge your park the way a critic would; it judges it the way a tired, slightly nauseous, mildly hungry person would, and that makes it far more honest.

Over the course of the series, the guest model grew richer. Parties of friends arrived together, shared a single ride target, and splurged on ice cream in unison. Weather systems made them shiver and grumble. The 'thought' queue—those little text bubbles that pop up over a guest's head—became the game's most beloved micro-narrative, turning a simulation into a thousand tiny, overlapping stories about one very ordinary Tuesday at the theme park.

In Their Own Story

The rain started at 2:47 p.m., thin and cold, the kind that soaks through a cheap park T-shirt and makes the asphalt paths slick. A party of three—two teenagers and a younger kid with a lollipop—had just finished the Tornado, the one with the four inversions and the drop that makes your stomach lurch into your throat. The kid was green. The older teen was laughing, pointing at the sky as if the clouds owed them an apology. The third, a lanky guy in a bandana, was already scanning the map board for the nearest burger stand.

They found it, but the path was blocked by a cluster of guests arguing about a lost set of keys, and the AI shuffled them into a slow, meandering orbit around the obstacle. The kid's lollipop dripped onto the path. The bandana-guy's 'thought' icon flickered: a little angry face, then a hungry face, then a small, resigned shrug. They ate. They paid. They walked on, the rain still falling, the park still humming with the distant shriek of someone who had just decided—against every rational instinct—to ride the Tornado again.

Reader's Guide

Guests are spawned at park entrances with a set of internal state values: happiness, energy, nausea, hunger, thirst, and a personal intensity tolerance. Each tick, the AI evaluates its current state against available actions (walk to a ride, buy food, sit on a bench, leave) using a utility-style scoring system. The ride they choose is filtered first by their 'want to ride' queue, then by whether the ride's intensity, nausea, and excitement ratings fall within their personal comfort window. A guest who is already nauseous will skip a high-nausea coaster even if it's at the front of their list.

Pathfinding is grid-based. Guests compute a route to their target and follow it, but they can become 'lost' if the path is blocked or the target becomes unavailable. Lost guests wander with reduced efficiency, which is why dead-end paths and poorly connected walkways are the silent killers of park throughput.

Practical strategies: keep food and drink stalls within a comfortable walking distance of high-nausea rides (guests will seek them out when nausea spikes). Provide benches near long queues to let energy recover. Vary your ride portfolio so guests with different intensity tolerances all have something to do. In RCT2, remember that a party shares a single ride target, so a cluster of four at a ride entrance can create a bottleneck that a solo guest would not.

Monitor the 'thought' queue in the guest list. A spike of 'I'm scared' or 'I'm getting sick' thoughts across many guests is your AI telling you that a ride's stats are misaligned with your guest base. Adjust the ride's design or its signage, and watch the thoughts shift.

Did You Know?

More in Game Mechanics & Systems

Elsewhere in the RollerCoaster Tycoon universe

Spotted an error? Know more?

This is a living reference — every entry is fact-audited, and reader corrections feed straight into our audit queue. Suggest an edit · See this site's audit record

Comments

Loading…
Open in the interactive codex →