You Were There: Ueda's 'Ico' at 25

Twenty-five years at the gate

Today, 24 September 2026, ICO has reached twenty-five years from its original North American release. That date is historically tidy, although my own relationship with the game begins a little later and in a much more ordinary place: with a demo disc in Australia, a television, a PlayStation 2, and the peculiar excitement that used to gather around a magazine-mounted piece of plastic. My first encounter was Official PlayStation 2 Magazine Demo 19, the Australian and European disc identified as SCED-50154, a compilation whose playable lineup included ICO alongside Dead or Alive 2, Formula One 2001, Mike Tyson Heavyweight Boxing and Vampire Night, with trailers for games including Final Fantasy X, Virtua Fighter 4 and Herdy Gerdy. That last title matters because I remember the Herdy Gerdy video as part of the same object, one of those apparently trivial associations that survives for decades because memory does not catalogue culture according to importance. It stores the menu music, the order of icons, a trailer glimpsed once, the texture of a plastic case, the feeling of an afternoon, and then, somewhere among all of that, an image that quietly changes the scale by which later images are judged. For me, ICO entered life that way. Before I knew anything about Fumito Ueda, Team ICO, design by subtraction, Giorgio de Chirico, Piranesi, Michiru Oshima, the difficult migration from the original PlayStation to the PlayStation 2, or the extraordinary influence this modestly sold game would eventually exercise over other creators, I knew a boy with horns, a pale girl, a vast castle, a wind moving through stone, and a hand that could be held.

I have played, watched, read and listened to an enormous amount of art since then, and the Ueda lineage has become one of the fixed points in how I understand creative work. Shadow of the Colossus is a work I love deeply, a game whose scale, melancholy and moral unease expanded the vocabulary established by ICO. The Last Guardian stands above even that for me. It is my favourite work in any medium, not my favourite game within a list of games, but the creative work I value most among films, music, books, visual art, television, games and everything else that has moved through my life. I am now looking forward with immense curiosity to genDESIGN's gen ATLAS, formally named and shown again at Summer Game Fest 2026 after its first public glimpse at The Game Awards 2024. Yet twenty-five years gives a useful reason to return to the beginning, because the beginning remains astonishingly complete. ICO does not feel like a sketch for the masterpieces that followed. It contains the grammar already: companionship expressed through systems; monumental environments whose history is felt more strongly than explained; animation that gives bodies weight and hesitation; sound used with restraint; architecture used as narrative; ambiguous mythology; a fascination with creatures, children, ruins and thresholds; the belief that emotion in a game can arise from responsibility and touch rather than exposition. Shadow would change the scale. The Last Guardian would push the relationship between player and companion into more difficult, animal, unpredictable territory. gen ATLAS now appears ready to rethink that relationship again through a giant mechanical partner. The first gesture, however, remains a small hand reaching for another hand in a castle.

An anniversary can easily become an excuse for canonisation, a ritual of repeating that a beloved object was ahead of its time and therefore deserves reverence. ICO deserves something more precise. Its importance lies partly in influence, and the record is unusually strong: Hidetaka Miyazaki has described ICO as the game that awakened him to the possibilities of the medium and helped prompt his career change into game development; Neil Druckmann has repeatedly cited Team ICO's approach to relationship through gameplay as an influence on the thinking that eventually fed into The Last of Us. Those facts are remarkable, but influence alone cannot explain why returning to ICO still feels so immediate. Plenty of historically important works become mainly historical. ICO remains physically legible. The act of calling Yorda, waiting for her, helping her across a gap, feeling the controller respond as hands connect, watching her become distracted by birds or the environment, and hearing the room change when shadows arrive still produces a relationship between player, character and space that resists reduction to a design lesson. The game survives because its ideas are embodied. It does not ask us to admire its theory. It asks us to cross the room together.

Demo 19, April 2002, and the lost culture of discovery

There is something fitting about first meeting ICO through a demo, because the game itself begins as a fragment. You arrive without a complete account of the world, its history or its rules. The castle is already old. The ritual that brings Ico there clearly predates him. Yorda's imprisonment belongs to a structure of power whose full history is never itemised. The shadows behave with purpose before we understand their nature. A demo disc intensified that quality. I had no expectation that this short encounter was announcing a body of work that would remain central to me for the next quarter-century. It was simply one playable object among many on Demo 19, and the surrounding menu made the contrast stronger. Early PlayStation 2 demo discs were carnivals of incompatible things. Racing, fighting, licensed spectacle, trailers, technical showcases and odd experiments sat beside one another without algorithmic sorting. You could move from the hard surface of one commercial genre into something as hushed and visually unfamiliar as ICO with almost no preparation. That kind of juxtaposition mattered. It allowed strangeness to arrive before recommendation systems, prestige discourse or accumulated reputation could tell you that the strange thing was important.

The disc itself can now be reconstructed with almost archaeological exactitude. Demo archives list SCED-50154 as OPS2M Demo 19; surviving Australian copies are marked for the ANZ market; its playable contents include ICO; its trailer reel includes Herdy Gerdy. I like knowing that because it turns a private memory into a traceable object. The little constellation is real. The Herdy Gerdy video was there. ICO was there. A piece of magazine ephemera that might easily have been thrown away carried, among a dozen other pieces of software, the first opening in what became a lifelong engagement with Ueda's work. Physical media often creates these accidental archives. A modern storefront remembers purchases, achievements and timestamps, but a demo disc remembers adjacency. It remembers that two completely unrelated works once shared the same rotating menu. It remembers the editorial priorities of a particular month. It remembers a period in which exposure to a game could occur because a magazine editor had selected it and pressed it onto a disc, not because a recommendation engine had profiled the player. Twenty-five years later, that accident feels precious.

The Australian timing also gives the memory its own shape. ICO had been released in North America on 24 September 2001 and in Japan on 6 December 2001. The PAL European release followed on 22 March 2002, with Australia receiving the game in 2002 after it had already been classified locally in December 2001. Demo 19 belongs to that PAL moment, when ICO was becoming available to players here and the Japanese and European cover artwork gave the game a visual identity very different from the North American box. For an Australian player discovering it through Official PlayStation 2 Magazine, ICO arrived with the atmosphere that has since become inseparable from its reputation: the pale, painterly cover; the sense of a European or Mediterranean ruin that was nowhere specific; the title itself, short and strange; the quiet confidence of a game refusing to shout over the louder material surrounding it. It is easy, with hindsight, to imagine that the encounter was destined to matter. It did not feel destined. That is part of why I treasure it. A life in art is built from these small collisions, and sometimes a work becomes foundational before you have language for foundations.

Before the castle existed

ICO's apparent simplicity conceals a difficult production history and a remarkably concentrated original idea. Ueda had entered the games industry through visual work, including animation on Warp's Enemy Zero, and approached Sony with an unusually image-led sensibility. The central concept that became ICO emerged around a relationship between a boy and a girl, expressed through the physical act of holding hands. In a 2002 interview with Official PlayStation 2 Magazine, Ueda recalled an image from a television commercial he had seen in high school, a woman holding a child's hand while walking through woods, and also cited the relationship between Maetel and Tetsuro in Galaxy Express 999 as part of the imaginative background. These are useful references because they reveal that the game's core was relational before it was architectural. The castle would become one of the most memorable spaces in games, but the project did not begin with a desire to make a castle simulator or an abstract puzzle box. It began with two figures whose difference in age, scale, vulnerability and movement could create meaning when placed together.

Development began on the original PlayStation. Early material preserved and later released by genDESIGN shows how much of ICO's visual grammar already existed in embryonic form: the pair, the monumental spaces, the sense of pursuit, the idea that movement through architecture would communicate story. Ueda has described the PlayStation version as incomplete despite how developed surviving footage may appear, and the hardware imposed problems on the reactions and behaviours he wanted from the companion character. The project moved to PlayStation 2 during development, a transition that created additional work rather than a simple resolution. Ueda told OPS2 that development had begun in February 1998, initially with pre-rendered backgrounds before the team shifted to real-time 3D; the project was already well underway when PlayStation 2 became the more attractive target. The migration mattered because ICO depended on subtleties that were easy to underestimate: animation, light, depth, distance, camera framing, and a companion whose behaviour could make the player feel that another being occupied the same space.

Kenji Kaido's role as producer was crucial to the production culture around Ueda's first directorial project, and the team remained small by the standards of major console development. In a 2001 interview Ueda described a team of roughly twenty people, peaking at twenty-three. Smallness did not make the work simple; it appears to have sharpened the need for a coherent set of priorities. The castle's areas had to function as puzzles, architectural spaces and cinematic compositions. The two central characters required hand-authored animation capable of conveying speed, reluctance, imbalance and attention. Enemies had to create pressure without turning the game into a conventional action title. Music had to be withheld until it could carry real force. The camera had to show traversal while also making height, emptiness and distance emotionally legible. These problems all point toward the principle later discussed under the phrase 'design by subtraction': identify the central experience, then remove systems and conventions that distract from it.

That phrase can become too neat when repeated in design history, as though ICO were produced by simply deleting features from a normal game. The harder truth is that subtraction creates obligations. Remove a dense interface and the world itself must communicate. Remove constant music and environmental sound must sustain mood. Remove a large cast of speaking characters and animation must carry social meaning. Remove a broad inventory and the few available actions must remain expressive. Reduce the narrative explanation and the image, architecture and rhythm of play must support interpretation without collapsing into vagueness. ICO's minimalism required concentration, not absence. The work feels spare because so many elements point in the same direction. The player's attention is repeatedly returned to the pair, the castle, the route ahead, the sounds around them, and the question of whether both can reach the next safe place.

Design by subtraction and the courage to leave space

The most useful way to understand ICO's restraint is to consider what the game permits the player to notice. Conventional action-adventure design in the early 2000s often filled the screen with explicit goals, collectible structures, combat escalation, explanatory dialogue and systems that rewarded constant activity. ICO offers very little of that machinery. There is no traditional heads-up display constantly reporting health, ammunition or progress. The castle is not partitioned into a map screen full of named districts. Yorda does not provide a stream of tutorial chatter. The soundtrack frequently withdraws. Puzzles are embedded in physical spaces rather than announced as separate challenges. Save points are benches where Ico and Yorda sit together, one of the gentlest and most thematically coherent save metaphors in the medium. These choices slow perception. A player begins to read ledges, chains, ladders, windows, gates and shadows. Light becomes information. Distance becomes information. Yorda's position becomes information. Her hesitation becomes information.

Ueda's design method was presented at GDC as a process of protecting a central concept by removing what diluted it. The famous example is spatial. A more conventional adventure might move through village, forest, castle and other regions in a broad heroic itinerary. ICO concentrates almost everything within the castle. That concentration allows the building to accumulate memory. A bridge seen from below can later be crossed. A tower glimpsed in the distance becomes a destination. Openings reveal earlier spaces from new heights. The place begins to feel constructed rather than generated for the player's immediate convenience. Ueda told interviewers that he wanted players to sense structures that had existed before their arrival, and that ambition is one of the reasons the castle remains persuasive even when its engineering is fantastical. It has the coherence of a dream remembered repeatedly. The geography can be impossible in literal terms while remaining convincing in emotional terms.

Subtraction also changes the relationship between author and player. ICO does not constantly state what its symbols mean. Horns, cages, idols, shadows, the Queen, Yorda's luminous presence, the castle's machines and the transformation of children into dark figures all carry narrative weight without being assembled into an encyclopaedia. The player's interpretation therefore becomes part of the experience. This quality later became strongly associated with Ueda's work, especially once Shadow of the Colossus encouraged years of archaeological speculation about its forbidden land, ruins and absent civilisation. In ICO the ambiguity is already mature. The world is withheld in the right places. One understands enough to care about escape, danger and attachment, while the larger mythology remains suggestive.

The restraint extends even to emotional instruction. ICO rarely tells the player when to feel tenderness. It builds conditions in which tenderness can arise from action. Calling Yorda from another platform, waiting for her to find the path, catching her at an edge, returning to help when shadow creatures seize her, or simply running while holding hands all produce tiny rhythms of attention. The game trusts repetition to deepen meaning. By the time a later sequence threatens that connection, the player's response is built from hours of muscle memory. The controller has already learned the relationship. This is one of ICO's enduring achievements: it understands that a game can prepare an emotion mechanically, long before a cutscene asks the player to recognise it.

The castle as a memory machine

The castle is among the great architectural inventions of games because it operates simultaneously as setting, puzzle, antagonist, shelter and archive. Its scale is oppressive, yet many of its rooms are beautiful. It imprisons Yorda, yet it also contains gardens, courtyards, water, sunlight and views that make escape feel almost secondary to looking. The camera frequently pulls away from the characters, shrinking them against towers, walls and voids. These compositions tell the player that the building exceeds the present story. Ico and Yorda are small lives moving through an older system. The visual hierarchy matters: human figures can be emotionally central while physically marginal. That tension recurs throughout Ueda's later work, where Wander is dwarfed by the colossi and the boy in The Last Guardian is dwarfed by Trico and the Nest. Scale becomes a way of thinking about vulnerability.

Ueda's architectural influences resist a single national or historical category. He spoke about Giovanni Battista Piranesi, whose impossible prisons and monumental structures are an obvious point of comparison, while also mentioning Japanese rural scenery, including the melancholy of partially developed or abandoned places near the countryside. The cover art for the Japanese and PAL releases draws openly on Giorgio de Chirico, particularly the metaphysical unease of small figures placed beneath severe towers, arcades and empty plazas. De Chirico is useful here because his spaces feel both ancient and modern, familiar and uninhabitable. Their perspective appears rational until the atmosphere destabilises it. ICO works in a similar register. The castle contains recognisable engineering, masonry, doors, stairs, elevators and waterworks, but their arrangement belongs to a symbolic geography.

I have always loved that the game never reduces this architecture to background decoration. Traversal requires looking. The player must understand how spaces connect, where Yorda can follow, where she cannot, what can be moved, which gaps can be crossed, and how the environment changes when power or machinery is activated. Because the same space can be encountered from multiple angles, solving a puzzle becomes a form of architectural comprehension. The castle slowly enters the player's body as a route. One learns its rhythms through movement rather than through a minimap. That distinction seems small until one compares the memories produced by each approach. A map gives a representation of space. ICO gives remembered labour: climbing that chain, opening that gate, running back for Yorda, sitting on that bench, seeing sunlight enter from that opening.

The castle also demonstrates Ueda's instinct for controlled visual incompleteness. Surfaces are not overloaded with explanatory detail. There are no readable plaques describing dynasties, no codex entries tucked into corners, no audio logs from former residents. The building's history is communicated through wear, scale, function, repetition and abandonment. The player supplies time. This has become common enough in environmental storytelling that its radical quality can be forgotten, but ICO's version remains unusually pure. It does not scatter narrative collectibles across a ruin. It makes the ruin itself the narrative object. What happened here is partly the wrong question. The more immediate question is what the place feels like now that we are inside it, and how that feeling changes because two people are trying to leave together.

The hand

If ICO can be reduced to one action without diminishing it, that action is taking Yorda's hand. The brilliance lies in how literal the gesture is. Games routinely represent relationships through dialogue trees, affinity meters, quest completion and cinematic scenes. ICO places the relationship under the player's finger. A button connects the characters. Hold it and Ico pulls Yorda along; release it and the physical bond ends. The controller vibrates, movement changes, and the pair become a temporary unit with unequal momentum. Ueda and his team understood that the sensation needed imperfection. Yorda does not snap into formation like a tactical companion. She accelerates, hesitates, follows, looks around, catches up. Her body has its own timing.

The mechanic creates responsibility through inconvenience, although 'inconvenience' is too negative a word for what the game achieves. Yorda cannot traverse every obstacle that Ico can. The player often crosses a gap alone, works out a mechanism, and then creates a route for her. This means progress is measured twice: first by whether Ico can reach a place, then by whether the pair can occupy it together. That second requirement changes the puzzle from self-directed mastery into care. A ladder that solves Ico's problem may solve nothing for Yorda. A door that only she can open reverses the dependence. The two characters possess different forms of agency, and the game repeatedly rearranges which one matters.

Early commentary sometimes described Yorda as an escort character, a label that places ICO beside a long history of disliked missions where the player protects an artificial intelligence from its own incompetence. The description misses the emotional architecture. Yorda is vulnerable to the shadows, but she is also the person who can open the sealed idol doors. She is the centre of the castle's power struggle. Her apparent fragility makes the player's physical help necessary while her mysterious ability makes escape impossible without her. The relationship therefore resists the simple fantasy of a competent hero transporting passive cargo. Ico is a child condemned because of his horns. Yorda is imprisoned by her mother and bound to the castle's future. Both are trapped by inherited systems. Their practical asymmetry gives them a reason to move together.

The strongest evidence that the hand mechanic works is the way the game later weaponises its absence. Once holding Yorda becomes habitual, separation feels wrong at the level of control. Ueda discussed the collapsing bridge sequence as a deliberate attempt to make the player understand the need to keep holding on. By that point the gesture has accumulated meaning through repetition. The scene does not invent attachment at the moment of crisis; it exposes attachment that the player's hands have already rehearsed. This distinction matters to me because it captures what interactive art can do with unusual directness. A film can show one character grasping another. A game can make the audience maintain the grip and then make the loss of that input feel like loss.

Yorda, language and the dignity of another mind

Yorda's presence depends on a difficult balance. She needs to be legible enough for cooperation and independent enough to feel separate from the player. Too much obedience would turn her into a tool. Too much unpredictability would make companionship frustrating. The final game often finds emotion in the narrow band between those extremes. Yorda wanders slightly, notices things, reacts, waits, and moves with a softness unlike Ico's more direct physicality. Ueda said that the team spent time on apparently unnecessary behaviours, such as Yorda strolling, following birds or pointing, precisely because they wanted players to feel that the two characters shared an adventure rather than merely functioning as complementary puzzle pieces. Those details are expensive in development and priceless in memory.

Language deepens that separateness. Ico and Yorda do not initially share speech. Her words are rendered in an invented language, and in the original presentation the subtitles themselves resist easy comprehension. The result is a relationship conducted through tone, gesture, proximity and repeated action. The game makes communication a bodily problem. Ico calls. Yorda looks, moves or does not. The player learns her pace. She learns, in the fiction, to follow him. Mutual understanding emerges without a conventional stream of banter. This silence protects mystery, but it also gives ordinary motions unusual significance. When characters cannot explain themselves, turning toward someone becomes expressive.

There is a wider principle here that I think explains much of Ueda's later work. His companions are compelling because they retain opacity. Agro is loyal but remains a horse with movement that must be anticipated rather than directly puppeteered. Trico becomes one of the most astonishing creatures in games because he is affectionate, frightened, curious, stubborn and sometimes difficult to command. The robot head in gen ATLAS is already being described by Ueda as partner, vehicle, navigator and tool, with a relationship that players may interpret differently. That lineage begins with Yorda. Ueda's companions do not exist to flatter the player with instant compliance. They create attachment by maintaining enough otherness to require attention.

This is one reason The Last Guardian eventually became my favourite work in any medium. It takes the proposition of ICO and makes the companion far less human, far larger, more physically consequential, and less directly controllable. The emotional wager becomes more extreme: can a player love an artificial creature partly because it will not always do exactly what is wanted? ICO's version is quieter, but the philosophical seed is present. Caring for another being means accepting a gap between intention and response. The game makes that gap playable.

Silence, wind and the music that waits

ICO is remembered as a quiet game because it understands the dramatic value of withholding music. Many large games use continuous scoring to regulate emotional temperature, signalling danger, melancholy, triumph or discovery before the player has time to decide what the space feels like. ICO often leaves the castle open to wind, footsteps, chains, water, mechanisms, distant environmental sounds and the small vocalisations of its characters. The absence of a constant score makes rooms feel larger. It also makes the player listen for practical information. Shadow creatures are heard as disturbances. Yorda's calls matter. Machinery acquires texture. The acoustic emptiness becomes part of the architecture.

Ueda spoke about keeping background music to a minimum so that music could be effective where it did appear. That decision also aligns with the project's broader resistance to redundancy. If the image and action already communicate solitude, a continuous melancholy score might overdetermine the experience. ICO allows the player to inhabit the feeling. This confidence is rare. Silence in commercial media is often treated as a vacuum to be filled, especially in works concerned about losing attention. ICO uses silence as attention.

When music enters, the contrast gives it enormous force. Michiru Oshima's contribution, alongside pentagon and the game's sound team, creates a score that feels suspended between folk memory, dream and ritual. 'You Were There', composed by Oshima with lyrics by Lynne Hobday and vocals by Steven Geraghty, remains inseparable from the game's ending for me. I do not need to quote the lyric to describe its effect. After hours in which melody has been rationed, the human voice arrives with the feeling of something remembered from far away. The title itself, 'You Were There', has always seemed like an accidental description of how ICO survives in memory. It addresses a presence that may be absent now but remains emotionally active.

The OPS2 interview contains another small piece of lineage I love: Ueda said that during early development he used Simon and Garfunkel's 'Scarborough Fair' as temporary music and that it became an inspiration for ICO's theme. The connection makes intuitive sense. Both have the air of an old song whose origin is difficult to place, a melody carried forward through repetition and variation. ICO constantly feels like folklore from a culture that never existed, and its music shares that quality. The score does not decorate the castle. It seems to have leaked from the same imaginary history.

The image: de Chirico, Piranesi and the art of the impossible familiar

The Japanese and PAL cover art has become one of ICO's defining images, and its relationship to Giorgio de Chirico is more than an art-history footnote. Ueda designed the cover with de Chirico's metaphysical painting in mind, particularly works such as The Nostalgia of the Infinite and Mystery and Melancholy of a Street. The visual relationship is clear in the severe tower, long shadows, tiny figures and dreamlike public space. More importantly, the cover identifies a psychological register that the game sustains: a place can be bright and unsettling at the same time. Sunlight does not guarantee safety. Emptiness can be beautiful without being peaceful. Architecture can feel meaningful before its meaning is known.

Piranesi enters from another direction. His imagined prisons and architectural fantasies exaggerate stairways, arches, bridges and vertical depth until built space becomes psychological space. ICO's castle carries a similar pleasure in impossible construction, though the game is gentler than Piranesi's darkest visions. Ueda also cited Japanese rural imagery, including the strange melancholy of development meeting countryside. This mixture helps explain why the castle cannot be reduced to 'European fantasy architecture'. It is assembled from visual memories rather than documentary reconstruction. The result feels familiar because its components recall many places, and unreal because no single place could contain them all.

The camera completes the visual system. ICO's viewpoints often prioritise composition and spatial understanding over the intimacy of a shoulder-mounted action camera. Characters can become tiny within the frame. The player is encouraged to see the route, the gap, the height, the larger structure. In 2001 this had practical consequences for navigation, but it also established a distinct relation between player and avatar. Ico is ours to control, yet the camera sometimes observes him as one small figure in a painting. The game therefore alternates between embodiment and contemplation. I run, jump and pull; then a framing decision reminds me how little of the castle I occupy.

This visual distance is one of the reasons the game has aged gracefully. Photorealistic ambitions can become attached to the technical assumptions of their moment. ICO's stylisation turns limitation into language. Bloom, haze, simplified facial detail, high contrast and carefully controlled colour produce an image that still feels authored. Later higher-resolution versions clarify surfaces, but the original's identity was never dependent on texture density. It depended on proportion, light, movement and composition. Those survive resolution changes because they are decisions rather than specifications.

Shadows and the violence the game refuses to celebrate

Combat in ICO is deliberately awkward. Ico swings a stick or sword against shadow creatures with none of the virtuosity expected from an action hero. Encounters are urgent because Yorda can be taken, not because defeating enemies produces an elaborate progression economy. The shadows emerge from dark apertures, seize her, and attempt to drag her away. Their design is abstract enough to avoid becoming a competing spectacle. Ueda explained that smoke and shadow helped distinguish them visually and allowed them to appear and disappear in ways suited to their behaviour. They are threat in concentrated form.

The clumsiness of combat protects the game's identity. If fighting were deep, rewarding and frequent, the player would begin to seek it. ICO needs the opposite relationship. Combat interrupts movement through space and the developing intimacy of the pair. Violence is something to clear so that the journey can resume. This principle evolves dramatically in Shadow of the Colossus, where combat becomes the entire visible structure of the game while the emotional meaning of each victory grows increasingly uneasy. There, mastery and guilt are placed in tension. In ICO, the tension is simpler: violence is necessary because something is trying to take Yorda away.

The horned shadow children give the imagery a deeper cruelty. Their resemblance to Ico links the castle's supernatural machinery to the ritual that brought him there. The story never needs a long explanatory monologue to make the implication disturbing. Children marked as different are sacrificed to a system that converts them into part of its own darkness. Ico escapes the assigned fate and, by helping Yorda, disrupts another inherited cycle. The mythology remains ambiguous, but the emotional structure is clear enough: both children are being used by an order older and more powerful than they are.

I find that aspect more affecting with age. When I first encountered ICO, the castle was primarily a place of mystery and wonder. Twenty-five years later, the institutional shape of the story is harder to miss. Adults and systems have decided what these children are for. Ico is designated a sacrifice. Yorda is designated a vessel. The adventure begins when they reject those assigned functions through the simplest possible alliance: they leave together. The game never turns this into a political speech. Its ethics are expressed through movement.

The ending and the strange endurance of ambiguity

ICO's final movement works because the game has spent so long teaching the player to understand togetherness as a physical condition. Separation therefore becomes more than a plot complication. The bridge, the fall, the return through altered spaces and the confrontation with the Queen all draw power from the earlier hours of ordinary cooperation. The player knows what it means for Yorda to be missing because the game's basic movement pattern has changed. The castle becomes lonelier once the second body is absent.

The ending itself remains one of the reasons the game stays alive in discussion. It resolves enough to satisfy the escape story while preserving uncertainty around Yorda, the Queen, the castle and the exact status of what the player finally sees. Ueda has consistently shown little interest in collapsing interpretation into a single authorised explanation. His later games would intensify this habit. Connections between ICO and Shadow of the Colossus exist, but they are delivered in ways that encourage players to infer history rather than receive a chronology. The Last Guardian similarly offers an emotionally clear relationship inside a world whose deeper cosmology remains partially veiled.

Ambiguity can be lazy when a work withholds answers because it has not developed them. ICO's ambiguity feels different because the immediate experience is so exact. We know how it feels to move through the castle. We know how Yorda behaves. We know the rhythm of saving on a bench, the anxiety of a shadow encounter, the relief of a door opening. The concrete layer is strong enough to support uncertainty above it. This is an important distinction. Mystery becomes persuasive when the world beneath it has weight.

The beach at the end has stayed with me for the same reason. After an entire game of stone, mechanisms, height and confinement, open shore and natural light feel almost unreal. The location is simple, yet the contrast makes it enormous. Then the song begins. The game releases the player from the castle without completely releasing the castle from the player. That sensation is still present twenty-five years later.

From ICO to Shadow of the Colossus

Shadow of the Colossus is often treated as the moment Ueda's ideas became monumental, which is fair in the literal sense. ICO's castle is huge, but Shadow turns scale into the primary mechanic. The colossi are enemies, landscapes, puzzles and moving architecture. Wander climbs them the way Ico climbs towers. The lineage becomes obvious once the spectacle is set aside. Both games are concerned with bodies navigating structures that exceed them. Both use sparse worlds, restrained exposition and invented languages. Both ask the player to form attachment through action. Both distrust easy heroism.

Yet Shadow also reveals how willing Ueda was to transform rather than repeat. Yorda's companionship is removed from moment-to-moment play. Mono is the goal rather than the partner. Agro becomes the continuing companion, a horse whose imperfect responsiveness gives travel physical character. The game replaces the intimacy of handholding with the intimacy of riding, calling and trusting. The castle's accumulated geography becomes a forbidden land organised around immense encounters. Combat, which ICO kept deliberately secondary, becomes the explicit task. The moral confidence of rescue becomes unstable as each victory darkens Wander and empties the landscape of another living giant.

I love Shadow of the Colossus for all of these reasons, and its twentieth anniversary has rightly renewed attention to how singular it remains. There are moments in Shadow that no other game has reproduced, especially the combination of awe and reluctance that can accompany a successful climb. Yet I cannot see Shadow as a replacement for ICO. It is a sibling that inherits a grammar and speaks it differently. The smaller earlier game still possesses forms of tenderness that Shadow deliberately abandons.

The relationship between the two works is also a reminder that artistic progression does not have to mean increasing complexity. Shadow is larger in spectacle while remaining structurally austere. Sixteen colossi, a horse, a sword, a bow, a landscape, a repeated ritual. The same instinct toward concentration survives. ICO therefore matters to Shadow at the level of method as much as mythology. It proved that a game could build a strong identity by committing to a small number of actions and images with unusual seriousness.

The Last Guardian, and why the line becomes personal

The Last Guardian is where this lineage reaches the point that, for me, exceeds the usual language of preference. I can rank games, films, albums and books for convenience, but The Last Guardian occupies a different category in my life. It is my favourite work of any media of any kind. That statement includes every creative form I know and every work I have encountered. Part of the reason is historical: by the time it arrived in 2016, I had already lived with the ideas of ICO and Shadow for years. The Last Guardian felt like a return, an extension and a culmination. The deeper reason is Trico.

Trico transforms Ueda's companion philosophy by making the relationship radically asymmetrical. The creature is enormous, animal, emotionally expressive, physically dangerous, frequently helpful and never reducible to a vehicle with perfect inputs. The player must observe mood, attention and environment. Commands can fail or be delayed. Progress often depends on understanding what Trico is perceiving. Critics sometimes described this as unreliable AI, yet the friction is fundamental to the bond. A creature that responded like a cursor would cease to feel like Trico.

ICO made the companion relationship tangible through a hand. The Last Guardian expands that proposition into a whole ecology of touch: climbing feathers, removing spears, feeding barrels, soothing fear, calling, pointing, being caught, being carried, watching Trico squeeze through architecture built at another scale. The game's most powerful moments emerge because the creature has been present as a problem, a friend, a body and a mind for hours. The emotional ending therefore lands on accumulated behaviour rather than on script alone.

When I return to ICO at twenty-five, I can see The Last Guardian everywhere in seed form without diminishing either game. Yorda's pauses anticipate Trico's autonomy. The castle anticipates the Nest. The priority placed on animation anticipates an entire game built around making an impossible creature feel alive. The quiet anticipates long stretches where environmental sound and bodily movement do more than dialogue. The relationship between small human figure and monumental environment anticipates the visual logic of every Ueda game that followed.

That continuity also explains why I look toward gen ATLAS with such unusual hope. I do not need it to reproduce The Last Guardian. I would rather Ueda and genDESIGN surprise me. The June 2026 reveal already suggests a deliberate shift: science-fiction machinery, a colossal robot head, firearms used within a broader action-adventure framework, an open world, more visible communication, and a partner that can function as vehicle, navigator, tool and companion. The vocabulary is changing while the old question remains recognisable: what happens when a player must live alongside another presence rather than merely control a character through a world?

The influence that travelled outward

ICO's commercial history is part of its mythology because its influence greatly exceeded its initial sales. It became one of those works passed between creators, critics and players as evidence of what games could communicate when they trusted interaction, atmosphere and restraint. The most striking testimony comes from Hidetaka Miyazaki. Before FromSoftware, while working outside the game industry, he played ICO on the recommendation of friends and later said that it awakened him to the possibilities of the medium and made him want to create games. That sentence has acquired almost absurd historical weight. The designer who would go on to shape Demon's Souls, Dark Souls, Bloodborne, Sekiro and Elden Ring found a path toward game development partly through Ueda's first title.

The connection should not be flattened into a claim that ICO directly designed Souls. Their systems and tonal priorities differ substantially. What matters is permission. A work can influence another creator by demonstrating that a medium can support a particular seriousness of atmosphere, ambiguity and authorship. ICO showed that a console game could be quiet, visually rigorous, mechanically focused and emotionally suggestive without apologising for any of those qualities. Miyazaki's later worlds share the confidence to let architecture, absence and incomplete history speak, although his combat design and role-playing structures are his own.

Neil Druckmann and Bruce Straley's interest in ICO offers another line. Druckmann has discussed Team ICO's ability to create relationships through gameplay as an influence on work that preceded and eventually informed The Last of Us. Again, the important inheritance is methodological. The goal is to make the player care through what they do with another character, not only through what a cutscene tells them. Tenzin in Uncharted 2, Ellie in The Last of Us and many later companion designs across the industry exist in a landscape where ICO helped establish that a relationship could be systemic.

The broader influence is harder to enumerate because ICO's ideas have become diffuse. Minimal interfaces, environmental storytelling, companions with strong behavioural identities, games built around vulnerable partnerships, ruins that imply history without codices, deliberate silence, small protagonists against monumental spaces: none belongs exclusively to ICO, and many have independent origins. Yet the game became a reference point through which these values were discussed. Its name entered the vocabulary of developers. That kind of influence is less visible than a copied mechanic and more important than imitation. It changes what people believe is possible.

The irony is that ICO itself drew from a broad field: de Chirico, Piranesi, rural Japanese landscapes, hand-drawn animation, older cinematic and literary images, Galaxy Express 999, and games such as Another World. Ueda's originality was never isolation. It was synthesis disciplined by a clear sensibility. This is one of the most useful lessons the game still offers. Original work often emerges from recognisable influences arranged around a personal conviction strong enough to transform them.

What twenty-five years does to a game

A twenty-five-year-old game exists in at least three forms: the software itself, the historical object it became, and the private version preserved by each player. These forms overlap without becoming identical. The software can be emulated, remastered, analysed and replayed. The historical ICO can be placed within the PlayStation 2 launch era, Japanese game development, art-game discourse and the careers it influenced. My ICO still contains Demo 19. It contains the specific surprise of encountering this strange thing before I understood its reputation. It contains the later discovery that the director's next game was Shadow of the Colossus, the long wait for The Last Guardian, and the eventual recognition that those works had become a personal artistic lineage.

Replaying an old game therefore becomes a conversation between versions of oneself. Some mechanical details that once seemed invisible become obvious. Camera speed, combat roughness and companion pathfinding can feel more exposed beside modern standards. Other qualities grow stronger because contemporary design has made their rarity clearer. ICO's refusal to fill every second with prompts, dialogue and rewards feels almost radical now. Its length feels elegant. Its limited verbs feel coherent. Its absence of an explicit upgrade economy allows the player to finish with essentially the same body with which they began, altered by experience rather than statistics.

The technical age also reveals the difference between fidelity and presence. ICO's faces are simple. Animation occasionally shows its era. Textures are spare. Yet Yorda can feel more present than characters rendered with orders of magnitude more detail. Presence is relational. It depends on how a character responds to space, how the player must account for them, how sound and timing give them continuity, how the game reacts when they are threatened. This is why Ueda's work remains relevant in an industry capable of extraordinary graphical realism. He has always been interested in the harder problem: how to make something seem to exist.

genDESIGN's later discussions of The Last Guardian reinforce this continuity. Ueda has argued that caricature and omission can heighten reality, because realism is not equivalent to maximal information. ICO demonstrates that principle at the beginning of his directorial career. A simplified figure can feel alive if motion, context and interaction are convincing. A castle can feel ancient without thousands of written lore entries. A relationship can feel profound without extensive dialogue. Twenty-five years has made these claims easier to see, because technology has advanced far enough to prove that detail alone never guaranteed them.

gen ATLAS and the horizon beyond the castle

For years after The Last Guardian, the next Ueda project existed mainly as an object of faith: genDESIGN was working, Epic Games had signed on as publisher, concept images appeared, and the absence of a release window became part of the familiar rhythm around a creator whose games take time. The first trailer at The Game Awards 2024 finally provided moving images: a lone figure, a colossal mechanical body, an approaching wall of destruction, separation and escape. In June 2026, Summer Game Fest transformed that unnamed project into gen ATLAS and showed substantially more of its world.

The new game is immediately recognisable as a departure. Ueda has confirmed that formal design work began in 2020 after several years of prototyping and building genDESIGN as a studio. gen ATLAS is the largest game he has made, structured as an open world, with a giant robot head serving multiple functions and a firearm available to the player without turning the experience into a conventional shooter. Ueda has discussed influences including Giant Robo and WALL-E, and has described the robot as partner as well as tool, navigator and means of transport. He has also said that the game is a new world with no deliberate direct connection to ICO, Shadow of the Colossus or The Last Guardian, while leaving open the possibility that players may perceive echoes once they experience it.

Those details make the twenty-fifth anniversary of ICO feel less like a closed retrospective and more like a point on a living line. The designer who built a game around a boy taking a girl's hand is now building one around a humanoid character travelling with a severed giant robot head. The materials have changed from stone, wood and shadow to machinery, weapons and science-fiction ruins. The obsession with scale remains. So does the companion question. So does the interest in worlds that appear to have histories larger than their protagonists. Even the shift toward an open world seems less like capitulation to a market format than an attempt to find the spatial form required by these new ideas.

I am especially interested in Ueda's recent comments about communication. gen ATLAS appears likely to contain more textual communication than the earlier games because the science-fiction setting makes logs and recorded exchanges believable. That could alter the balance of silence and interpretation substantially. I welcome the risk. A creator becomes boring when the audience's love hardens into a demand for self-imitation. The most respectful hope I can have for gen ATLAS is that it follows the conviction that produced ICO rather than the surface of ICO itself: identify the experience that matters, build the world around it, cut what weakens it, and trust the player enough to leave room.

Waiting for gen ATLAS also gives me a strange feeling of continuity with that Demo 19 disc. Once again there is a Ueda work ahead of me that I do not fully understand. Once again images exist without the whole. The difference is that I now carry twenty-five years of expectation into the encounter. That makes surprise harder, but not impossible. Ueda has already surprised me three times by refusing to make the same game twice. I hope the fourth does it again.

The bench, the gate and the small rituals that become enormous

ICO's most memorable ideas are often attached to small objects. The save bench may be the clearest example. Saving a game is normally an administrative act that sits outside the fiction: open a menu, choose a slot, confirm, return. ICO turns it into an image of companionship. Ico and Yorda sit beside one another on a stone bench and fall into stillness. The transition is quiet, almost domestic. It gives the player a reason to see rest as part of the journey rather than as a break from it. A bench also has an ordinary social meaning outside games. People sit together because they are waiting, recovering, talking, watching, sharing time. ICO borrows all of that without explaining it. The save point therefore becomes one more place where a conventional piece of game machinery is translated into the language of the relationship.

The idol gates work similarly. They are locks, progression checks and pieces of worldbuilding at once. Ico cannot open them through strength. Yorda can, and the strange light that moves through her hands makes her agency visible in a way that does not require dialogue. The player quickly learns that reaching the door alone is insufficient. Progress belongs to the pair. This is one of the game's recurring structural jokes at the expense of heroic individualism: Ico can climb, leap, fight and manipulate much of the environment, yet the castle repeatedly reminds him that his own abilities do not constitute escape. He needs another person whose powers are different from his.

Even the call button becomes a ritual. Calling a companion in another game can feel like issuing a command to a unit. In ICO the call has shades of invitation, impatience, reassurance and fear depending on context. When Yorda is close, the sound is ordinary. When she is distant, it measures separation. When shadows arrive, calling can become panic. When a route has finally been prepared, it becomes an announcement that it is safe to come. The same input acquires emotional variation because the situation changes around it. Ueda's design is full of this economy. Rather than inventing a separate mechanic for every emotional state, the game lets a small set of actions gather different meanings over time.

The bombs, chains, blocks, ladders, switches, ropes and mechanisms are similarly plain. They belong to the grammar of puzzle games, but ICO's presentation keeps them materially grounded. A block has weight. A chain occupies space. A bomb must be carried. A mechanism makes noise and changes the architecture. The player rarely feels as if a puzzle has materialised solely to test abstract intelligence; it feels as if a strange old building contains systems that can be exploited. This is partly an illusion created by careful staging, yet all game worlds are illusions of this kind. ICO's achievement is to make the underlying design disappear into place.

These small rituals matter because memory rarely preserves an entire playthrough with equal intensity. It keeps fragments: the bench music, the light of an idol door, the little tug as Yorda's hand connects, the sound of sandals on stone, a chain swinging in a high chamber, a patch of grass in a courtyard, the sudden stain of shadow opening beneath her feet. Twenty-five years later I do not think of ICO as a sequence of puzzles solved in order. I think of these sensory units. The game has become a collection of physical memories, even though none of the physical spaces ever existed.

There is a useful lesson here for any creative work concerned with scale. Monumentality does not come only from huge objects. ICO certainly has towers and abysses, but the castle becomes believable through tiny repeated behaviours. The grand view works because the player has also pushed the block, waited at the ladder and sat on the bench. The Last Guardian later takes the same principle into the body of Trico. The creature's size is spectacular, yet attachment is built through small acts: feathers, eyes, wounds, barrels, calls, scratches, pauses. Shadow of the Colossus gives Agro similar dignity through the simple fact that riding has rhythm. Ueda's worlds become large because their details have consequence.

The PAL ICO, the Australian object and the strange luxury of getting the later version

My first encounter with ICO came through the PAL ecosystem, and that matters beyond nostalgia. The North American version arrived first in September 2001, while the Japanese release followed in December and the European PAL version in March 2002. The later Japanese and European versions incorporated refinements and extras that the earlier American release did not contain. After completing the game, PAL players could begin a second playthrough in which Yorda's language became translated and, remarkably, a second controller could take control of her. There were additional secrets and visual options as well. These features are curiosities beside the central game, but they reveal that the version circulating through Australian magazine culture in 2002 belonged to a project that had continued changing after its first commercial release.

The second-player mode is especially fascinating because it alters the philosophical premise in a playful way. For most of ICO, Yorda's partial independence is essential. The player cannot reduce her to a second avatar. On a completed PAL save, another human can inhabit that role. The camera and structure remain centred on Ico, and Yorda still has limited abilities, yet the relationship can literally become cooperation between two people on a couch. A mechanic designed to simulate attention toward another mind can, after the first playthrough, be handed to another mind. It is an elegant bonus because it exposes the architecture of the original relationship without pretending that the two experiences are equivalent.

The PAL cover also matters. The American release used an image created under different production constraints and has long been criticised for misrepresenting the game's tone. The Japanese and PAL art, drawn by Ueda, presents the game through the de Chirico-inflected language that now seems inseparable from ICO. For a player in Australia, that meant the object on the shelf already communicated distance, architecture and loneliness before the disc loaded. Cover art can be dismissed as packaging, yet packaging determines the threshold through which a work enters life. Ueda understood that the picture outside the game could belong to the same world as the picture inside it.

The Australian classification record adds another mundane piece of the object's history. ICO was classified here on 3 December 2001, with Sony Computer Entertainment as applicant, months before many Australian players would see the PAL game in stores and magazines. Bureaucratic records have no romance, but I enjoy the contrast. Somewhere a file records a date, a platform, an applicant, a rating and the phrase 'low level animated violence'. Somewhere else, my memory records a pale girl in a cage. Culture is always both things at once: paperwork and revelation, product code and private meaning.

Demo 19 sharpens that duality. Its code, SCED-50154, is almost comically dry. The disc is an industrial object, one numbered piece of Sony's promotional apparatus. Its menu contained products scheduled for sale. Yet that industrial object became the accidental carrier of a life-changing aesthetic encounter. There is a temptation to romanticise the pre-digital era as though every demo disc was magic, which would be false. Many were forgotten immediately. The point is that physical distribution created particular forms of chance. You bought or borrowed a magazine for one reason and discovered something else because it was on the same disc.

I miss that form of adjacency. Digital abundance gives us access to far more material, often instantly, but it can reduce the productive arbitrariness of a finite compilation. Demo 19 did not know my preferences. It did not predict that ICO would become part of an artistic lineage I would follow for decades. It simply placed ICO beside Formula One 2001 and Mike Tyson Heavyweight Boxing and trusted curiosity to do the rest. The randomness now feels almost curatorial.

That is why I want the demo-disc detail preserved in this anniversary piece rather than treated as disposable autobiography. It says something about the way art arrives. The official history begins with development dates, E3 showings and release territories. A personal history begins wherever the work first becomes visible. For me, the beginning is not 24 September 2001 in North America or 6 December in Japan. It is a 2002 Australian magazine disc, a menu, a Herdy Gerdy trailer somewhere nearby, and the decision to select ICO. The date of a masterpiece's publication and the date it enters a life are different anniversaries. Both deserve to be remembered.

Preservation, access and the future memory of ICO

An anniversary also raises the less romantic question of access. ICO was created for a console that is now several generations old. It received the excellent PlayStation 3 HD collection with Shadow of the Colossus in 2011, but the original game has never enjoyed the kind of continuous modern availability that would make recommending it effortless in 2026. Shadow of the Colossus received a full PlayStation 4 remake from Bluepoint in 2018. The Last Guardian remains a PlayStation 4 title playable on later hardware through compatibility. ICO itself is still more dependent on old hardware, old discs, the PS3 collection, or emulation and preservation practices. For a work this influential, that situation is frustrating.

Preservation is sometimes discussed as if the only thing at stake were code. Code is essential, but ICO's history shows how many surrounding objects matter: regional versions, demo discs, magazine interviews, concept footage, cover art, manuals, promotional trailers, development builds, soundtrack releases, classification records and memories from the people who made and played it. genDESIGN's decision in 2021 to release footage from the original PlayStation development era was valuable because it made visible an earlier form of the project that could otherwise have survived mainly through second-hand description. Demo archivists preserving the contents and serial numbers of magazine discs perform a different version of the same cultural work. Neither archive is glamorous. Both prevent history from flattening.

ICO particularly deserves preservation because its development process is part of its meaning. Seeing the PlayStation prototype clarifies how persistent the central idea was across technological change. The pair, the scale and the atmospheric ambition existed before the final hardware made them fully workable. Understanding that continuity prevents a simplistic story in which PlayStation 2 power suddenly created the game. Technology enabled the final form, but the artistic problem preceded it.

I hope that Sony eventually gives ICO a careful modern release that respects the original image while making the work easy to access. I do not need a remake that replaces every surface with contemporary realism. The game has already proved that its visual identity does not depend on that. A high-quality preservation release, with stable performance, appropriate options, the PAL and Japanese extras, archival materials, original and restored presentation choices, and perhaps the surviving development footage would be more valuable to me than an aggressive reinterpretation. The ideal edition would treat ICO as a work of art with variants and history, not as outdated software requiring cosmetic correction.

The reason is straightforward. Twenty-five years from now, when ICO turns fifty, the players encountering it for the first time should have the opportunity to meet the game rather than merely its reputation. Essays like this one can describe the hand mechanic, the castle, the quiet and the influence. None can substitute for the moment the player presses the button, hears Ico call, watches Yorda approach and feels the controller acknowledge contact. Preservation matters because some meanings live only in interaction.

There is a beautiful symmetry in that. ICO is a game about helping something vulnerable cross a dangerous structure without losing contact. Preservation asks us to do something similar with the work itself. Hardware generations fail. Storefronts close. Licensing changes. Physical discs decay. Knowledge fragments. The answer is continued care: hold on, carry what can be carried, build a route, make sure the other side remains reachable.

Returning to the castle after a lifetime of other worlds

There is another reason ICO remains difficult to dislodge from memory: so many later games have taught me how unusual its priorities were. I have spent years in enormous open worlds, dense role-playing systems, cinematic adventures, action games of extraordinary mechanical sophistication and independent works capable of formal experiments that would have been commercially improbable in 2001. I am grateful for that abundance. Yet ICO's scale of attention still feels distinct. The game assumes that a few hours in one place with one other person can be enough. It does not need a continent, a skill tree, a hundred quests or a season of post-release expansion to justify its existence. Its confidence is almost architectural. Build the castle correctly, place the two characters inside it, establish the terms of their dependence, and let the experience deepen through traversal.

That confidence has changed how I understand creative ambition. Ambition is often mistaken for quantity because quantity is easy to advertise. More square kilometres, more hours, more enemies, more systems, more dialogue, more endings. ICO's ambition lies in coherence. Every major decision is asked to support the same emotional proposition. Even the castle's scale serves intimacy because the larger the space becomes, the more meaningful it is to cross it with someone. Even the shadows serve intimacy because their threat is defined by separation. Even the doors serve intimacy because Ico needs Yorda beside him. The game's ambition is therefore centripetal rather than expansive: ideas are pulled inward toward a centre.

This is also why the rough edges rarely bother me. ICO has awkward combat, old camera conventions and moments when companion behaviour exposes the limitations of early-2000s technology. A work does not become great by eliminating every friction. It becomes great when its essential decisions remain stronger than the friction surrounding them. In some cases the friction contributes texture. Ico's imprecise fighting makes him feel like a child rather than a trained warrior. Yorda's pace prevents her from becoming a perfectly synchronised extension of the player's body. The camera's distance sometimes makes modern control instincts impatient, but it also preserves the game's visual composition. I would not argue that every limitation is secretly intentional. I would argue that the work's identity is robust enough to absorb them.

The difference between tolerance and appreciation becomes important when older games are discussed. Nostalgia can excuse anything, and historical respect can become a reason to pretend that a work has no age. ICO has age. It belongs to a particular machine, controller, display technology, development culture and period of 3D game design. Its image was built around cathode-ray televisions and the rendering assumptions of PlayStation 2. Its controls reflect a time before certain third-person standards hardened. Its regional releases contain differences that a modern simultaneous worldwide launch would likely eliminate. I value those facts because age is evidence of survival. A work can remain powerful while visibly belonging to another period.

The same principle applies to my own memory. I do not want to polish the Demo 19 encounter until it resembles a predestined artistic epiphany. I was a young person playing a magazine demo disc. I did not emerge from the session with a vocabulary of metaphysical painting and interactive companionship. Recognition came gradually. I played more. I learned who Ueda was. Shadow of the Colossus arrived. Years passed. The Last Guardian became an almost mythological unreleased project and then, wonderfully, a real game. The meaning accumulated. This slower process is more valuable to me than a dramatic conversion story because it resembles the way most lasting art actually works. It keeps returning until one day you realise it has become part of your internal furniture.

At twenty-five, ICO now sits on both sides of that realisation. It is historically distant enough to belong to another era of PlayStation, yet emotionally present enough that a screenshot of the castle can collapse the distance immediately. The game has crossed from something I once discovered into something I use to measure discovery itself. When I encounter a new work with unusual quiet, a convincing companion, a world that trusts absence or an image that makes architecture feel like memory, some part of me still hears ICO in the background. That is influence at the scale of a life rather than an industry. It is the standard by which an old surprise continues to make room for new ones.

Why ICO remains

When people describe ICO as beautiful, they are correct, but beauty is only the visible edge of its achievement. The game is durable because form and feeling are unusually aligned. The architecture makes the characters small, so their connection matters. The lack of language makes gestures important. The lack of constant music makes sound meaningful. The companion's limitations make attention necessary. The sparse interface makes the world readable. The minimal mythology makes imagination active. The awkward combat makes violence feel like interruption rather than reward. The repeated act of holding hands turns a button into an emotional object.

That alignment is difficult to produce and easy to romanticise after the fact. Development was long, technical problems were real, the PlayStation version had to be abandoned, spaces were built and rebuilt, the team had to solve companion behaviour and animation with limited resources, and the game did not become a massive commercial phenomenon on release. The final elegance should not erase the labour. In some ways, the labour is what makes the elegance admirable. Simplicity in a finished work often hides the largest number of discarded possibilities.

I also think ICO persists because it is generous toward the player's imagination. It does not treat ambiguity as a puzzle with a canonical answer waiting in a wiki. It gives enough form for personal meaning to gather. The castle can be a literal place, a fairy-tale prison, an image of inherited violence, a childhood dream, an architectural memory, or all of these at once. Yorda can be read through mythology and symbolism, yet she also remains a character whose hand must be reached. The symbolic and mechanical layers reinforce one another without requiring a single thesis.

My own meaning has changed as my life has changed. As a young player I was struck by mystery, atmosphere and the sheer difference of the game. Later I became more interested in design, authorship and how Ueda's visual education shaped the work. Later still, the themes of care, vulnerability and inherited responsibility became more powerful. A work that can survive these changes without becoming exhausted has achieved something rare. It continues to meet the person who returns to it.

A quarter-century without closure

Twenty-five years is long enough for a game to pass through several technological eras and still remain younger than the questions it asks. ICO is concerned with fear of difference, inherited obligation, confinement, trust, communication across barriers and the decision to help another person even when no institution around you rewards that decision. These are not themes that expire with hardware. They also explain why the work can feel more serious now than it did to me on first contact. As a younger player, escape was the obvious goal. As an adult, the fact that Ico and Yorda choose one another inside systems that have already chosen their fates for them carries more weight. Their alliance is small and private, yet it is powerful enough to break the structure around them.

I also value the absence of triumphalism. ICO is not a fantasy of becoming the strongest person in the castle. Ico remains physically modest. Yorda remains vulnerable. Their progress depends on persistence, cooperation and understanding a place that is larger than either of them. Even the final victory does not turn the game into a coronation. The ending returns to quiet, distance and uncertainty. There is humility in that shape. The characters survive the ordeal without the work pretending that survival has made the world simple.

Perhaps that is the deepest reason I can still approach ICO with gratitude rather than mere nostalgia. Nostalgia asks for a vanished time to return. ICO does not need 2002 to return. Demo 19 matters because it was the door through which I entered, but the work on the other side of that door continues to stand. I can honour the magazine, the console, the period and the person I was then without requiring the game to remain trapped there. The castle is old by design. It was always waiting for someone to come back.

You were there

Twenty-five years after 24 September 2001, and more than twenty-four years after that Australian Demo 19 encounter, I am grateful that ICO entered my life before I knew what to call it. There was no need for a theory. The disc went into the console and the castle appeared. A boy and a girl moved through light and stone. I held a button and he held her hand. That was enough to begin.

Everything that followed now makes the beginning look almost mythic: Shadow of the Colossus, with its impossible scale and sorrow; the long, improbable gestation of The Last Guardian; the arrival of Trico, who became part of the work I value above all other creative work; genDESIGN's independence; the years of waiting; the first 2024 glimpse of a new mechanical giant; the 2026 revelation of the name gen ATLAS and the knowledge that Ueda is still pursuing the same broad artistic problem from a new direction. Yet I do not want the later history to turn ICO into a prologue. A prologue exists to lead somewhere else. ICO is complete in itself.

The greatest tribute I can offer is therefore simple appreciation. Appreciation for Ueda's conviction before there was a reputation to protect. Appreciation for Kenji Kaido and the small team that made an unusual project viable inside a major platform holder. Appreciation for animators who hand-shaped movement, artists who built a castle from remembered influences, programmers who made a companion believable, sound designers who understood the force of quiet, and musicians who waited until melody could matter. Appreciation for the magazine editors and disc producers whose Demo 19 placed this strange game in front of an Australian kid among racing games, fighters and trailers. Appreciation for the fact that a commercial medium built around hardware cycles can still produce works that outlive the machine that ran them.

Most of all, I am grateful for the hand. It remains one of the clearest ideas in the history of games because it collapses theme, mechanic, animation, sound and emotion into one act. Hold the button. Wait for Yorda. Feel the connection. Move together. The game does not need to announce that companionship matters; it makes companionship the condition of progress. Twenty-five years of technical advancement have not made that idea smaller.

There is a line from ICO to every later Ueda work, but I prefer to imagine it as a route through one of his spaces rather than a straight chronology. Somewhere far behind is a sunlit courtyard. Somewhere above is a bridge. Somewhere ahead is a giant shape we cannot yet understand. The route between them is built from doors, pauses, separations, returns and moments of trust. I first stepped onto that route through Demo 19 in 2002. I am still on it now.

The anniversary leaves me with gratitude rather than a wish to freeze the past. I can remember the plastic demo disc, the magazine culture around it, the particular strangeness of discovering ICO without knowing what it would become, and still look forward with real excitement to gen ATLAS. That continuity matters to me. The child who encountered a castle on Demo 19, the adult for whom The Last Guardian became the most treasured work in any medium, and the person waiting now to see where Ueda goes next are connected by the same curiosity. Twenty-five years have added context, but they have not exhausted the feeling of standing before one of these worlds and wanting to know what waits beyond the next door.

There is one more line I need to acknowledge, because for me the influence of ICO does not end with the works that followed it inside Ueda's own career. It runs into my own. The Quantumyth, as I understand it now, would not exist without ICO. Some of its stories and images began before I encountered the castle, but ICO altered the scale of what I thought an interactive work could be. It demonstrated that architecture, silence, movement, companionship and player responsibility could carry meaning with the density of literature, cinema, painting or music while remaining irreducibly interactive. That possibility became part of the creative foundation beneath everything I have tried to build since. When I imagine The Quantumyth eventually becoming interactive, I am not imagining conventional games wrapped around stories I have already written. I am imagining works in which interaction itself becomes expressive material, where the act of moving, waiting, touching, protecting, choosing, noticing or simply being present can carry the story. ICO helped teach me that this was possible before I had the language to explain why.

I eventually tried to put that belief into words. In 2021, in the ArtsHub essay adapted from my Melbourne International Games Week keynote, I argued for thinking beyond the narrow cultural category of 'gaming' and toward interactive art as a mature creative form with extraordinary unrealised potential. I was trying to articulate a conviction that had been forming for years: that this medium can sit beside humanity's oldest and most serious arts while also doing something none of them can quite do, because the audience does not remain entirely outside the work. ICO is one of the reasons I believe that so deeply. It does not ask us merely to observe care. It gives care a button, a rhythm, a distance, a vulnerability and a consequence. The lesson travelled with me into The Quantumyth, into every argument I have made about the future of interactive art, and into the stubborn belief that the medium's next evolution will come when creators trust interaction as a language rather than treating it only as a delivery mechanism for spectacle, competition or content.

That belief also makes the loss of Bluepoint Games particularly painful to consider at this anniversary. Bluepoint helped give ICO and Shadow of the Colossus a durable HD life on PlayStation 3 in 2011, then returned to this lineage with the 2018 Shadow of the Colossus remake, a work that demonstrated how extraordinary technical renewal can coexist with reverence for an older game's atmosphere, scale and emotional identity. Sony's decision to close Bluepoint in March 2026 ended a studio whose craft had become unusually associated with the careful restoration of PlayStation history. I mourn that loss, and especially the disappearance of a team of artists, engineers and craftspeople whose expertise was built around understanding where preservation ends and reinterpretation begins. The best remake work requires humility. Bluepoint understood that fidelity does not mean freezing every polygon in amber; it means recognising what the original was trying to make the player feel, then using new technology in service of that feeling.

I still believe ICO can return. A modern remaster would already be valuable, simply making the original readily available at contemporary resolutions and frame rates while preserving its visual character, sound, timing and strange luminous softness. A full remake could be beautiful too, provided the guiding principle remained restraint. The castle does not need to become busier. Yorda does not need more dialogue. The silence does not need filling. The mystery does not need explanation. The hand does not need improvement as an idea. The task would be to allow another generation to enter that space without sanding away the qualities that made it singular. If another studio someday inherits the kind of responsibility Bluepoint once carried, I hope it approaches ICO with the same seriousness that the 2018 Shadow of the Colossus remake brought to its source: not as an old object to be modernised out of embarrassment, but as a work of art whose original intentions deserve the best contemporary craft available.

And because anniversaries permit a little dreaming, I will allow myself one final hope for later this year. When The Last Guardian reaches its tenth anniversary in December, my dream announcement would be two projects side by side: The Last Guardian Remastered and ICO Remastered, with the latter perhaps opening the door one day to the remake I have imagined for so long. There is no need for such preservation to compete with the future. gen ATLAS is ahead of us, and I want Ueda and genDESIGN to keep moving forward into worlds none of us have seen. At the same time, the path behind them deserves to remain walkable. I would love to see a new generation discover Trico without technical barriers, then walk farther back through the lineage and find Ico and Yorda waiting in the castle. For me that would complete a beautiful circle: the work that helped define my understanding of interactive art, the work that became my favourite creation in any medium, and the work I am still waiting to see next, all allowed to exist together rather than replacing one another.

If The Quantumyth ever becomes what I have spent so many years imagining, some part of its ancestry will always lead back here. Back to a magazine demo disc, back to a castle I did not understand, back to the discovery that two small figures crossing an enormous space could contain an entire philosophy of interactive art. That is the inheritance I carry from ICO. It is artistic, personal and ongoing. Twenty-five years later, I am still trying to honour it by making something of my own that trusts the medium as completely as ICO did.

Happy twenty-fifth anniversary, ICO. You were there. You are still there.

Research notes and sources

• Research note: Factual details in this retrospective were cross-checked against contemporary interviews, official genDESIGN material, release databases, archival demo-disc catalogues and later creator interviews. The personal reflections and evaluations are my own.

• genDESIGN official site and news archive: ICO development footage (2021), gen ATLAS first trailer (The Game Awards 2024), second trailer and title reveal (Summer Game Fest 2026), studio history.

• Official PlayStation 2 Magazine interview with Fumito Ueda, February 2002, archived by Cats Under A Tree: development start, PlayStation-to-PlayStation 2 transition, hand animation, architectural influences, Yorda behaviour, Scarborough Fair temp music.

• Gaming Intelligence Agency interview with Fumito Ueda, August 2001, archived via Team ICO interview collections: team size, sound restraint, enemy design, environmental philosophy.

• GDC / Game Developer archives: Ueda's design-by-subtraction method and the concentration of ICO around the castle and central relationship.

• DemoPals / Crimson Ceremony OPS2M Demo archive: OPS2M Demo 19, SCED-50154, playable ICO and trailer list including Herdy Gerdy; surviving Australian listings identify the ANZ edition.

• GameFAQs and GameSpot release records: North American release 24 September 2001; Japanese release 6 December 2001; PAL European release 22 March 2002; Australian release in 2002.

• Australian Classification database: ICO classified 3 December 2001 for PlayStation 2, Sony Computer Entertainment applicant.

• Fondazione Giorgio e Isa de Chirico, Metaphysical Art journal: discussion of ICO's cover and Ueda's explicit de Chirico influence.

• The Guardian interview with Hidetaka Miyazaki, 31 March 2015: ICO awakening him to the possibilities of games and prompting his desire to make them.

• The Guardian / Washington Post interviews with Neil Druckmann and reporting on The Last of Us: Team ICO's influence on relationship-through-gameplay thinking.

• MusicBrainz, Tower Records Japan and soundtrack reference material: ICO ~Melody in the Mist~, Michiru Oshima, pentagon, Lynne Hobday and Steven Geraghty credits.

• Game Informer interview with Fumito Ueda, June 2026, and official genDESIGN materials: gen ATLAS development since 2020, open-world structure, giant robot head as partner/tool/vehicle/navigator, firearms within a broader non-shooter design, and Ueda's comments on continuity and new-world creation.

• ArtsHub, 'How games can become our greatest artform', 4 October 2021: edited version of my Melbourne International Games Week keynote, arguing for interactive art as a mature creative form and describing The Quantumyth as an interactive-focused multimedia project.

• Sony / Bluepoint Games closure reporting, February-March 2026, including Sony's statement confirming the studio's closure following a business review; Bluepoint's preservation work included The ICO & Shadow of the Colossus Collection (2011) and the Shadow of the Colossus remake (2018).

• The closing hopes for The Last Guardian Remastered, ICO Remastered and a future ICO remake are personal wishes, not announced projects as of 24 September 2026.