Adamastor: A Case Study on Generative Game Design

"Adamastor" is a design-focused paper presenting an exploration game with roguelike elements that treats procedural content generation (PCG) as a core experiential mechanism rather than merely a tool for producing assets. The game places the player in control of a plant seed traversing a sentient, living planet in search of a hidden sanctuary, with the planet's terrain, ocean currents, and geology reshaped by generators that both pre-compute the world offline and adapt it during play based on player behaviour. The paper frames its aims around the Discovery and Challenge aesthetics of the MDA framework and draws on Smith's dynamics of searching in a vast world, building generator strategies, and practicing in different environments.

The design rests on two procedural layers: an offline generator using pseudo-random number generators to produce layered world data such as height maps and ocean currents, and an online generator that analyses player playstyle to alter unexplored terrain, introduce natural barriers or moving obstacles, and manipulate the player's perception of the environment. Conceptually, the authors link the experience to the methodical, risk-tolerant exploration of Portuguese navigators during the Descobrimentos, as recounted in Camões's Os Lusíadas. Gameplay mechanics include wind as a regenerating but depletable resource, altitude control to avoid drowning or destruction, and fragile seed structures whose fragments can break, ending the run.

As a case study rather than an empirical evaluation, the paper offers no playtesting data, participant sample, or quantitative results; its contribution lies in illustrating how generative systems can be integrated into the core design logic of a game. Its scope is therefore limited to a single prototype and its theoretical grounding, with the authors positioning it as an experiment in using PCG to shape the experience itself.