Donkey Kong Bananza

Anyone who has been a child knows what this is. The cardboard box that gets broken down for recycling. The autumn leaf pile that gets jumped into and scattered. The sandcastle that gets stomped flat at the end of the afternoon. The snow fort that gets pushed over. The tower of blocks that gets demolished when the tower has been judged tall enough. The popping of bubble wrap. The breaking of a stick across a knee. The action across all of these is roughly the same action, and the action produces, in almost every body that performs it, a specific small pleasure that is structurally different from the pleasure of, say, building a sandcastle in the first place. The two pleasures are related, but they are not the same pleasure. The break-it pleasure is older, more reliable, and on the available evidence more widely distributed across the population.

Most three-dimensional video games of the past three decades have not been built to engage this pleasure. Most three-dimensional games are built to engage the navigate-the-level pleasure: the player encounters a space the designer has carefully arranged, and the pleasure comes from the body successfully traversing the space according to the design's intended path. This is also a real pleasure, supported by its own specific reward systems, and the contemporary game medium has refined the production of it to remarkable precision. The break-it pleasure, however, has been almost entirely absent from this lineage of design. It is available in Minecraft, in some destruction-focused niche titles like Red Faction: Guerrilla and Teardown, and in scattered demolition-derby modes inside larger games. It has been almost completely absent from the mainstream three-dimensional platformer, the genre Nintendo has been running for forty years.

Donkey Kong Bananza, the Switch 2 launch title Nintendo EPD released in summer 2025, has changed this for the first time in the mainstream platformer's history. The game's central mechanical proposition is that the player's avatar, DK, can take the level apart. The walls break. The floors break. The ceilings can be tunneled through. The geometry of the level is not a fixed substrate the player navigates; the geometry is a material the player works on. The shift sounds small in a paragraph description. The shift activates the break-it pleasure that the mainstream platformer has been structurally unable to activate for forty years. The result is one of the more substantial mechanical inventions Nintendo has produced in the genre since the 1996 introduction of Mario 64.

The frame this essay wants to give the reader: every game design choice activates specific psychological reward systems and lets others go quiet. The mainstream contemporary platformer has been activating one specific cluster (navigate-the-arrangement, master-the-traversal, perform-the-design's-intent) and letting the destruction cluster go quiet. The cluster the platformer has been ignoring is not a minor one. It is one of the more reliable sources of bodily pleasure in human development, present from infancy, well-documented in the developmental literature, and structurally available to almost every player who picks up a game.

This is a frame that travels. Once the reader has it, almost every game's design choices become legible as commitments to specific reward systems. The careful-traversal pleasure of Mario. The puzzle-solving pleasure of Portal. The collection-completion pleasure of Pokémon. The combat-mastery pleasure of Devil May Cry. The exploration-discovery pleasure of Outer Wilds. The build-and-shape pleasure of Animal Crossing. The destroy-and-disassemble pleasure of Bananza. Each of these is a specific psychological reward system the design has chosen to engage. The conversation about whether a game is "good" usually flattens these distinctions. The conversation about which reward systems a game activates and how well it activates them is sharper, more useful, and more applicable to choosing what to play.

The psychological foundation for the break-it pleasure is reasonably well-understood. James J. Gibson, the American perceptual psychologist whose 1979 book The Ecological Approach to Visual Perception founded the contemporary field of ecological psychology, gave the framework the analysis needs. Gibson's central concept was the affordance: the set of action-possibilities that a particular environment offers to a particular kind of organism. The chair affords sitting to a creature with the right body shape. The cliff affords falling-off. The tree affords climbing to a primate but not to a fish. Affordances are real properties of the environment-organism pair; they are not in the environment alone or in the organism alone, but in the specific relationship between them.

Gibson's framework has aged well into a useful tool for thinking about game design. Game design is, in the strict Gibsonian sense, the deliberate construction of an affordance environment. The designer decides which actions the world will offer the avatar's body; the body the player operates is the organism the designer is calibrating affordances for. The level is a structured set of affordances. The player's mastery is the player's increasing fluency at reading and exploiting the affordances the level has supplied.

What the mainstream three-dimensional platformer has been doing for forty years is calibrating its affordance environment to a specific narrow range. The player's body can jump. The player's body can run. The player's body can grab things. The player's body, in many games, can attack things. The player's body cannot, in any of the genre's mainstream entries, take apart the level it is moving through. The wall is, by design, an affordance for jumping-against (specifically: a barrier that constrains motion). The wall is not an affordance for breaking. The constraint of the wall's destructibility is the constraint of the genre's affordance vocabulary.

Bananza expands the vocabulary. The walls are now also an affordance for breaking. The floors are an affordance for digging through. The geometry of the level is now a material the player can work on rather than a substrate the player navigates around. The expansion is small in stated terms. The expansion is, in practice, a substantial shift in what the player's body can do in the world, and the shift activates the break-it reward system that the previous generations of the genre had been structurally unable to engage.

The reward system in question has been documented at the developmental-psychology and cognitive-neuroscience levels for some decades. The work on the sense of agency, gathered most accessibly in Daniel Wegner's 2002 The Illusion of Conscious Will, has demonstrated that the brain produces a specific reward signal when the body performs an action that visibly changes the environment in ways the actor expected. The breaking action produces this signal at unusual fidelity because the environmental change is large, immediate, and unmistakably caused by the actor's specific action. The breaking is, in the strict cognitive sense, a clean demonstration to the brain that the body has agency in the world. The breaking feels good because the body has, in performing it, received clear neurological confirmation that it can affect things.

A sharp reader could object at this point that breaking is not actually new to the platformer at all. Super Mario Bros., in 1985, let a powered-up Mario break a brick block outright by jumping into it from below. Donkey Kong Country, in 1994, built entire levels around TNT Barrels that exploded on the slightest contact, useful for blasting open a secret wall or finishing off an enemy no jump could touch, and DK Barrels that shattered open to release a captured Diddy Kong. Crash Bandicoot, two years later, filled its corridors with wooden crates that splintered apart on a single spin. The objection is fair, and worth taking at full strength rather than waving off: something has been breaking in this genre since close to its beginning.

What breaks in those games, though, belongs to a different category of object than what breaks in Bananza, and the difference is not decorative. A TNT Barrel announces itself before it is touched: stenciled, red, unmistakably a barrel and not a wall. The brick block in Super Mario Bros. is visually distinct from the solid block sitting beside it the moment the level loads. The crate in Crash Bandicoot is a crate, not a length of corridor wall built to resemble one. In each case the designer has pre-selected a small, bounded set of objects and flagged them, visually, as the exception to an otherwise fixed level. The breaking event itself is total and immediate, one hit and one state change, from present to gone. What the player manipulates, in every one of these games, is a stocked prop placed there for the purpose of being broken. The wall behind it, the floor beneath it, the level's actual substrate, stays exactly as fixed as it has always been.

Bananza's terrain carries no such flag. A wall looks like a wall: rock-textured, indistinguishable on sight from the wall ten feet away that will give a different amount, in a different shape, to the same punch. Nintendo's art director on the project, Daisuke Watanabe, has described the team working specifically against the blocky, obviously-cubed look of a game like Minecraft, aiming for a player's first reaction to a given surface to land closer to the reaction Watanabe named as the actual design target, "Wait, you can destroy this?", than an instant read of which textures are props and which are set dressing. That uncertainty matters more than it might first appear. Sense-of-agency research ties the strength of the reward signal to exactly this variable, whether the outcome was already knowable in advance or had to be produced by the action to be known at all. A TNT Barrel that explodes on the fortieth try exactly as it exploded on the first has more in common, in this one respect, with pulling a lever than with discovering an affordance. The wall that might give six inches or six feet, in a shape the player could not fully predict from the texture, sits closer to the leaf pile: foreseeable in kind, not in its particulars, and the particulars are what the signal is actually rewarding.

This is the pleasure children are demonstrating when they jump in leaf piles. The leaf pile reorganizes around the body. The body has, unmistakably, done something to the world. The reward signal fires. The same signal is firing when a teenager breaks a stick over a knee, when a frustrated adult slams a door, when a contractor demolishes a wall in a renovation project, and when DK punches through a wall in Bananza. The signal is one of the more reliable sources of small embodied pleasure available to humans. The signal is also, on the available evidence, increasingly less available to contemporary adults whose daily lives offer fewer and fewer opportunities to perform actions that visibly change the environment. The adult sits at a screen. The adult types. The typing produces small text outputs the screen has been designed to render in ways that minimize the felt sense of having changed anything. The breaking is, in many contemporary adult lives, structurally absent.

This is the cultural-historical situation Bananza enters. The game has, almost as a side effect of its mechanical proposition, recovered an embodied pleasure the surrounding adult environment has been progressively withdrawing. The player who has spent twelve hours operating DK across the campaign has been engaging the agency-confirmation reward system more times in those twelve hours than the player's surrounding daily life had been engaging it across the previous twelve weeks. The cumulative effect is not subtle. Players report that Bananza produces a specific kind of decompressing pleasure that the previous DK games, the previous Mario games, and most other contemporary platformers had not produced. The audience has been responding to the agency-confirmation signal being engaged at scale. Most of the audience has not had the vocabulary to name what is happening. The casual reception has clustered around words like "satisfying" and "cathartic" without quite reaching the underlying mechanism.

The reader who has this frame can apply it forward into evaluating any game whose design involves substantial environmental destruction. Some games activate the agency-confirmation signal well: Minecraft, with its mineable terrain that always produces clear feedback for the breaking action; Teardown, whose destruction physics have been calibrated specifically to produce maximally satisfying feedback for every break; Red Faction: Guerrilla, whose Geo-Mod technology engaged the signal at scale before the mainstream genre had developed the vocabulary for it. Other games include destruction without engaging the signal well: most cinematic action games' "destructible cover" mechanics produce small, contained, mostly-decorative breaks that do not produce the same reward as the genuinely-affecting destruction. The difference is whether the breaking has been calibrated to produce the agency-confirmation signal at full fidelity, or whether it has been calibrated to look impressive without actually engaging the reward system the visual impression is gesturing at.

None of this is a new discovery about pleasure, and the timeline is worth being precise about. Video games proved that breaking the world could carry an entire commercial product decades before Bananza, and decades before the three-dimensional platformer existed as a genre at all. Dig Dug, Namco's 1982 arcade release, gave its player a single verb: tunnel through packed earth toward the monsters living in it, the tunnels persisting, oddly shaped, wherever the player had already dug. Two years later, Boulder Dash had Rockford doing much the same thing through looser dirt, racing falling boulders toward a reward buried under the ground he was clearing as he went. Rampage, Bally Midway's 1986 cabinet, made destruction the entire premise: three mutated giants, one of them a gorilla named George, worked city blocks apart floor by floor until nothing was left standing. All three were commercial successes, and all three predate the Donkey Kong Country side-scrollers, which would let a barrel explode but would never let the level itself come apart, by the better part of a decade or more. The pleasure was proven in arcades before most of Bananza's eventual players were born. The genre chose, for a long time, not to use it.

The choice was not really a matter of taste, and treating it as one gives the platformer's design tradition too much credit for restraint and too little credit for what it was actually up against. Dig Dug and Boulder Dash could let their worlds break because those worlds were, underneath the sprites, a grid: a cell was earth or it was empty, a state flip with no physics to resolve and no way to make the level unwinnable that a designer could not test for by exhausting the grid's finite possibilities. A fully three-dimensional level that can be dug through anywhere raises a harder problem: whether the space just hollowed out still connects to everything it needs to connect to, whether a required item is still reachable, whether the player has carved an exit from the level nobody anticipated. That problem does not go away with better art. It has to be solved computationally, and for a long time it cost more to solve than a fixed level was worth. Nintendo's own account of building Bananza's version of it, reported from a 2026 CEDEC session by the engineers who built it, describes terrain assembled from more than one hundred million voxels, over eight hundred distinct materials each carrying its own destruction behavior, and more than two thousand separate decorative objects individually authored to break convincingly on impact. Director Kazuya Takahashi has said plainly that building the system "must have been an absolute nightmare for the programmers." The remark is not false modesty. A grid cell in Dig Dug either has earth in it or it does not. A voxel wall in Bananza has to know what it is made of, how much force just hit it, what shape the resulting hole should be, and what the level still needs to guarantee once the hole exists. Forty years is roughly how long it took the hardware and the tooling to make that guarantee affordable at the scale a mainline Donkey Kong game would require.

The hardware story is not the whole story, and it is worth being honest about the part that triangle counts and voxel budgets do not explain. Retro Studios' own Donkey Kong Country Returns, in 2010, and Tropical Freeze, in 2014, both arrived after Red Faction: Guerrilla had already shown a mainstream audience that large-scale environmental destruction was commercially viable on hardware from the same console generation as the Wii. Both games kept the barrel and the crate exactly as Rare had specified them in 1994. A substrate-destruction system risks something a barrel never does: it can turn a carefully telegraphed platforming challenge into a pile of options with no clear solution, and a studio that has spent two decades refining exactly how a Donkey Kong side-scroller telegraphs its challenges has a great deal to lose, and comparatively little to gain, by handing the player a way to dig around a challenge instead of solving it. The constraint, by the 2010s, was as much a bet about what a reliable and beloved formula could survive changing as it was a bet about processing power. A three-dimensional collectathon built around open exploration, which is what Bananza is, offered the first genre context where that particular bet finally looked worth making.

Andy Clark, in his 2008 Supersizing the Mind, made the broader cognitive-philosophy argument that bears on this. Clark's central claim was that the human mind is not, in the strict sense, confined to the brain. The mind extends into the body, the tools the body uses, and the environment the body interacts with. The relevant cognitive unit is the body-tool-environment system, not the brain alone. Clark's argument has been influential in the broader cognitive sciences and remains useful for thinking about what happens when a game expands the affordances the player's body has available.

In a conventional platformer, the cognitive unit is the player-controller-avatar system as it relates to a fixed level. The level is not part of the cognitive unit; the level is the fixed environment the cognitive unit moves through. In Bananza, the level is part of the cognitive unit. The player still moves through the level, but the greater share of the session is spent reshaping it, in collaboration with the design's initial setup, through the destruction-and-reshaping work. The cognitive unit has, in Clark's vocabulary, grown to include the material the player has been given to work with. The pleasure the game produces comes from that expansion.

The implication for the medium is more substantial than the casual conversation about Bananza has yet recognized. The mainstream three-dimensional platformer has been operating, for forty years, inside a specific affordance vocabulary that has been excluding one of the more reliable sources of bodily pleasure available to its players. The exclusion was not necessary. The exclusion was a specific design tradition that the genre had stabilized around without examining whether the stabilization was serving the medium's range of possible reward systems. Bananza has demonstrated, commercially, that the vocabulary can be expanded, that the audience responds when it is, and that the trade-offs (the loss of the careful-pacing the conventional platformer was good at) are real but not, on the audience's evidence, prohibitive.

The reader who finishes this essay can carry several useful frames forward. The first frame: when a game produces a specific kind of satisfaction the surrounding category does not usually produce, ask what reward system has been engaged that the category usually leaves quiet. The answer is often, on a careful look, available to anyone willing to look for it. The second frame: in one's own daily life, ask whether the contemporary environment has been activating the agency-confirmation reward signal at the rate the human nervous system was structured for. For most adult readers, the answer is going to be no, and the absence is one of the small specific things the contemporary cultural environment has been costing its participants. The activities that activate the signal (gardening, cooking, repair work, woodworking, even the small destructive pleasure of correctly recycling cardboard or breaking down a piece of furniture for disposal) are worth recognizing as the small embodied goods they actually are.

The third frame: when a game expands the affordance vocabulary of an established genre, the expansion is worth paying attention to. Most genre conventions in commercial games are stable not because they are uniquely correct but because they have been stable for long enough that they have stopped being examined. The games that examine them, when they succeed, demonstrate that the medium had been excluding things it did not need to exclude. Bananza is one of these. The next one will, on the available evidence, be worth watching for.

The cave wall in any game from the previous thirty years would have been a wall. In Bananza, the cave wall is a material. The body finds out which is which by trying. The trying turns out to be one of the more reliable sources of small embodied satisfaction the medium can offer, and the medium has been underproviding it for forty years. The fact that Nintendo has chosen, with this entry, to start providing it is an opening other studios should notice. The audience is ready. The reward system has been waiting.