Colossal Cave and the navigable space
The fundamental concept of an open world is a space that the player navigates at their own discretion — a place with multiple paths, where the player's choices determine the sequence of experience. Colossal Cave Adventure (1976) was the first implementation of this idea that reached a significant audience. Will Crowther's cave system was based on the real Mammoth Cave in Kentucky, and the navigational language — "you are at the bottom of a pit with a narrow east-west passage. There is a large boulder here." — described a space that existed independently of the player's position in it.
The critical design feature of Colossal Cave was that the world had a coherent geography that the player had to learn. Items existed in specific locations and stayed there until moved. Structures — the grate, the wellhouse, the axe — had relationships to each other that, once understood, made the world navigable without notes. The map in the player's head was the game. Building that map — learning where things were, how the passages connected, which routes were safe and which were dangerous — was the primary activity.
This is the foundational design principle of every open world game since: the world has a geography that the player discovers through exploration, and discovery is the primary reward. The content of the geography — the enemies, the puzzles, the narrative — is secondary to the act of learning the space itself.
Zork and the hostile world
Zork (1977–79) extended the Colossal Cave model with more sophisticated language parsing, more complex puzzles, and a tone that was darker and more challenging. The Great Underground Empire was not a friendly space. The thief stole items from your inventory and used them against you. The grue killed you in the dark. Navigating Zork required not just building a spatial map but understanding the world's rules — which objects could be manipulated, which creatures could be defeated, which puzzles required what items.
Zork also made explicit something that Colossal Cave had implied: the world was larger than any single session would exhaust. You could return to it, try different approaches, explore sections you'd missed. The world had a totality that was only revealed through repeated, varied engagement. This property — the world as larger than any single experience of it — is essential to what we now call "open world" design. A world you could exhaust in one play session wasn't really a world; it was a level.
The text parser that mediated access to Zork's world was both its strength and its limitation. Natural language interface meant the world could be described with richness and variety that visual graphics of the period couldn't approach. But it also meant the world existed only in language — there was no picture of the Great Underground Empire, only words. The mental image was the player's own, constructed from the text and therefore personal and vivid in a way that external graphics couldn't be. The trade-off between linguistic richness and visual immediacy was one that the industry took decades to resolve, and some designers would argue it never has been fully.
Ultima and the persistent world
Richard Garriott's Ultima series introduced a specifically new idea: the world that existed between sessions, that changed over time, and that had an internal consistency across multiple interactions. Ultima I (1981) was largely linear. Ultima III (1983) introduced a world where NPCs had schedules — they were in different locations at different times of day, and the time of day was tracked by the game. Ultima IV (1985) had a world where every action had moral consequences that accumulated across the entire game, not just the current session.
The Ultima world's persistence — the sense that it continued to exist between your play sessions, that the city of Britain was not instantiated fresh each time you loaded the game but was waiting for you exactly as you'd left it — was a qualitative change in what a game world was. It was no longer a set of rules that produced an experience when activated. It was a place that had a reality of its own, that you visited rather than generated.
This sense of the world as real — as having an existence that preceded and would outlast any particular play session — is what Miyamoto meant when he described Zelda as "a miniature garden." The garden exists. You visit it. You find things there that were left by the designer for you to discover. The experience is of discovery in a pre-existing world, not of constructing a world through your decisions.
The Legend of Zelda and the console open world
The Legend of Zelda (1986) is the game that brought the open world concept to a visual, non-literate form that mass audiences could engage with without gaming experience. Crowther and Lebling's text adventures required literacy in gaming language and comfort with reading. Garriott's Ultima required patience with menus and statistics. Zelda required only that you move a character on a screen — the most accessible possible interface — through a world that rewarded exploration in the most direct possible way: you found things.
The overworld of Zelda was a 16×8 grid of screens, each containing enemies, items, or passage to dungeons. Most screens could be visited in any order. The nine dungeons could be entered in any order, though the items found in earlier dungeons — the raft, the power bracelet — were required to reach later ones. This structure — open enough to allow individual routes, constrained enough to create a meaningful discovery sequence — is the design template for every open world game since.
The specific genius of Zelda's overworld design was that the geography itself communicated information without stating it. The forest that concealed a dungeon entrance wasn't labelled as hiding something — but a forest in a context where secrets were embedded in unexpected places suggested to alert players that it was worth investigating. Burning a tree to find a passage, bombing a specific wall in a dungeon, pushing a gravestone to enter a hidden room — these were solvable by players who had learned the world's logic, not by players who had been told the solution. Learning the language of the world was the game.
What was established and what came next
By 1990, the open world as a design concept was fully established. The specific technical and design vocabulary — the persistent geography, the player-directed exploration, the world that rewards attention and punishes inattention, the sense that the world exists independently of the player's presence in it — had been developed over fifteen years of text adventures, graphical RPGs, and action-adventure games. What changed after 1990 was scale and technology.
The move from 2D top-down to 3D first-person and third-person perspectives, which became possible with consumer hardware through the mid-1990s, changed the experiential quality of open world exploration without changing its fundamental design logic. Standing at the edge of the world in The Elder Scrolls: Morrowind (2002) and seeing a mountain range on the horizon, then walking to that mountain range and finding it genuinely there, populated with content and history — this is the same experience as entering a cave in Colossal Cave Adventure and finding it connected to rooms you'd visited from a different direction. The world is coherent. It existed before you arrived. You are discovering it, not generating it.
The designers who built the concept had hardware that could barely display it. Crowther ran Colossal Cave on a PDP-10 mainframe. Garriott sold his first games in plastic bags from a computer store. Miyamoto programmed Zelda into a console with 2 kilobytes of RAM. The idea — that a world could be made, and that exploring it could be the point — proved robust enough to survive the transition to hardware of almost incomprehensibly greater capability. It's still the idea at the centre of the medium's most ambitious work.