Metroid and the locked door
Metroid (1986) on the Famicom Disk System was not the first game with a non-linear structure, but it was the first to make non-linearity functional through a specific mechanism: the ability gate. The planet Zebes was a connected map of rooms; the map could be explored in multiple directions from the beginning, but specific passages required specific abilities to traverse. The high jump required the High Jump Boots. The path into certain areas required Bombs. The Varia Suit was necessary to survive in high-temperature zones. Players who found an ability unlocked not just the area immediately beyond it but, potentially, areas they had encountered earlier in the game and been unable to enter.
The ability gate created a specific player experience: encounter a barrier, note it, continue elsewhere, find an ability, return to the barrier, pass it. The world was not simply sequential — it was a network of possibilities that expanded as the player's capability expanded. Backtracking was not failure but discovery: returning to an earlier area with a new ability revealed content that hadn't existed before the ability was acquired. The map that had seemed fully explored became unexplored again. The sense of a world opening continuously as the player became more capable was the source of the genre's distinctive satisfaction.
Super Metroid (1994) refined the formula to what most players consider its peak form. The map was larger, the abilities more varied, the hidden routes more numerous, and the environmental storytelling — conveyed through background design and enemy placement rather than text — more accomplished than the original. Super Metroid also introduced a mechanic that became central to the genre's design language: the ability to acquire skills unofficially, through player discovery rather than designated pickups. The wall jump and the mockball could be performed without official instruction, discovered by players who were willing to experiment with the movement system's mechanics. The game's world contained secrets accessible only to players who had internalized the rules deeply enough to extrapolate beyond them.
Castlevania's two paths
The original Castlevania (1986) was a linear action game: a sequence of stages traversed left to right, each with defined enemies and a boss, culminating in a confrontation with Dracula. The structure was closer to an arcade game than to what the genre name would later suggest. The game's design strengths — its precision platforming, its distinctive gothic visual style, its measured difficulty — had nothing to do with exploration or connected worlds.
Castlevania II: Simon's Quest (1987) attempted an open-world structure and was received poorly. The open world required players to find information — where to go next, what items to combine, what to say to which NPC — that the game provided through cryptic in-game hints that were additionally obscured by poor translation from Japanese. Players spent significant time unable to progress without external guidance. The game demonstrated that open-world structure without adequate player guidance produced frustration rather than exploration.
Castlevania: Symphony of the Night (1997) returned to the open structure with everything Simon's Quest had lacked. The game was produced by Koji Igarashi — who would become the primary author of the Castlevania formula for the following fifteen years — and combined the Metroid ability-gate structure with an RPG levelling system, equipment collection, and stat growth. The player character Alucard gained levels from combat, collected weapons and armour with different properties, and acquired traversal abilities that opened new areas. The RPG layer added a progression texture that pure ability-gating didn't provide: even without new traversal abilities, growing stronger through combat and equipment created a sense of continuous advancement that carried players through the exploration the map required.
What the combination created
The fusion that Symphony of the Night achieved — Metroid's connected, ability-gated world with Castlevania's RPG progression — created a structure with two simultaneously active engagement loops. The exploration loop: find the map, identify barriers, find abilities, return to barriers, pass them, reveal more map. The progression loop: fight enemies, gain levels, find equipment, become stronger. Each loop reinforced the other: becoming stronger through the progression loop made the exploration loop more accessible, while exploring the map through the exploration loop provided more enemies and equipment for the progression loop.
The two loops also provided different types of player agency. The exploration loop was spatial — players decided where to go and how to use the map's structure. The progression loop was incremental — players made decisions about equipment and could measure their growth numerically. Players who found the exploration loop frustrating could grind the progression loop until they were strong enough to proceed by force. Players who found the progression loop tedious could push into unexplored areas at risk of encountering enemies they weren't ready for. The genre's tolerance for different player approaches was an emergent property of having two interacting loops rather than one.
The genre's conventions — a large connected map, ability gates, progression systems, boss encounters that test accumulated capability, hidden rooms that reward exploration — were stable enough by the mid-2000s to be recognisable as a genre rather than as features of specific games. Hollow Knight (2017), Ori and the Blind Forest (2015), Axiom Verge (2015), and dozens of other independent games used the Metroidvania structure as a known design language, making games that were legible to players familiar with the genre's conventions without needing to establish them from scratch. The genre that Igarashi named in the late 1990s produced one of game design's most reproducible and commercially reliable structures.
Why the structure endures
The Metroidvania's durability as a design structure is explained by what it offers players that other structures don't. Linear games with defined progression provide narrative and difficulty control at the cost of agency: the player goes where the designer sends them. Open-world games with free exploration provide agency at the cost of directed engagement: the player can go anywhere, but the sense of purpose can diffuse. The ability-gated connected world occupies a middle position: the player has genuine agency over which areas to explore and in what order, but the map's structure channels that agency toward specific discoveries through barriers that can only be passed with specific capabilities.
The return visit is the structural mechanism that sustains the engagement. A player who sees an area they can't yet access doesn't experience a blocked path; they experience a marked possibility. Remembering barriers and returning to them with new abilities is a form of spatial reasoning — maintaining a mental model of the map and a catalogue of its blocked passages — that engages a different cognitive mode than the pure reaction of action games or the narrative following of linear adventure games. Players who enjoy this mode find Metroidvania games deeply satisfying because the satisfaction is cumulative: each ability acquisition immediately expands the world, producing a sequence of "aha" moments as barriers that had seemed permanent dissolve.
The genre's failure mode is the opposite of that satisfaction: getting lost in a map full of barriers with no clear indication of which barrier is currently passable, and no clear indication of what ability would pass it. The best Metroidvanias manage information carefully — hinting at which barriers are near-accessible and which require abilities far ahead, providing enough map detail that backtracking is efficient rather than random, giving players enough sense of direction that exploration feels purposeful rather than aimless. The worst produce the Simon's Quest experience: a world full of barriers and inadequate information about how to pass them. The design challenge is calibrating the gap between what the player knows and what they need to know to navigate productively — close enough to be solvable, wide enough to be satisfying.