Roguelike, Roguelite, or Run-Based? Ask What Survives the Reset

“Roguelike” and “roguelite” can identify a useful family of games, but the suffix alone cannot tell you what playing one will feel like. A turn-based simulation with permanent death, a rapid action platformer with unlocks, and a story-driven combat run may all invoke Rogue while asking for different skills.

For a practical description, start with the reset. Map what disappears, what persists, and what changes only because the player learned. The result is more useful than winning an argument about which label a game has earned.

The historical checklist was never a licensing board

The Berlin Interpretation emerged from discussion at the 2008 International Roguelike Development Conference. Its high-value factors include procedural generation, permadeath, turn-based and grid-based play, non-modal actions, complexity, resource management, and exploration. Its preserved text also says that missing some points does not automatically mean a game is not a roguelike.

That last detail is often lost when the Interpretation becomes a binary checklist. It was an attempt to describe a prototype—a cluster of traits associated with canonical examples—not an authority capable of approving later designs.

A Game Studies analysis by Jeremy Cartlidge applies prototype theory to the same dispute: genres can have porous boundaries, and a component list can overstate mechanics while under-describing the experience those mechanics produce. In plain language, games may resemble the center of a genre in several strong ways without sharing every feature.

The Interpretation remains valuable historical vocabulary. It tells us why turn structure, spatial reasoning, procedural worlds, and consequential death became central to “traditional roguelike” discourse. It becomes less useful when converted into a purity score.

Draw a four-layer reset map

For any game built around repeated attempts, inspect four layers.

Layer	What to ask	If it resets	If it persists
World state	Are maps, encounters, rewards, and routes rebuilt?	The next attempt demands fresh navigation and adaptation	The player can execute a learned world plan
Run power	Do equipment, cards, levels, health, and temporary abilities vanish?	Each run must rebuild its operating engine	The character can accumulate power across failures
Permanent state	Do unlocks, upgrades, currencies, story flags, or starting options remain?	Progress relies more heavily on player mastery	Future attempts may become broader, easier, or simply different
Player knowledge	Does the player remember enemies, systems, patterns, and risks?	It does not literally reset	Learning can be the main form of progress even when numbers vanish

The fourth layer is crucial. A game can reset every stored advantage and still produce cumulative progress because the player recognises a trap, values a resource differently, or understands a route. Conversely, permanent upgrades can grow while the player learns very little if repetition overwhelms decision-making.

This is why “you start over when you die” is not enough. Start over with what?

Traditional roguelikes foreground a world that must be read

Freehold Games describes Caves of Qud as a roguelike built around procedural regions, deep simulation, permanent death, and rolling a new character. It is a useful modern example of the traditional side of the family: the world is not merely a sequence of combat rooms but a system whose objects, creatures, factions, environments, and resources can interact.

That does not mean every traditional roguelike must perfectly satisfy the Berlin list, or that visual presentation determines membership. The more useful expectation is that turns expose information and permit deliberation; the generated world creates unfamiliar combinations; and death can erase a large amount of stored character state.

The player’s durable resource is therefore often interpretation. What does this creature imply about the region? Which unidentified object is worth testing? Is an apparently safe tile connected to a system that can propagate danger? A new character may be numerically weak, but the returning player is not equivalent to a beginner.

Permanent death gains meaning from the small decisions that carry real strategic weight. If an attempt ends through an unreadable surprise with no reusable lesson, reset becomes only repetition. If the world telegraphs enough to support inference, loss can update the player’s model.

Roguelites preserve different things for different reasons

“Roguelite” is commonly used for games that borrow run resets, procedural variation, or permadeath while changing other parts of the traditional pattern. That still covers several distinct persistence structures.

Hades: narrative and power cross the boundary

Supergiant’s FAQ calls Hades a rogue-like dungeon crawler while also explaining the looser “rogue-lite” distinction. Its repeated attempts combine randomized runs with carried-forward knowledge, permanent progression, relationships, story events, and optional difficulty modifiers.

The reset is not a return to the same narrative zero. Failure moves the protagonist back to a hub where conversations and upgrades can change what the next attempt means. Mechanical progress, story progress, and player learning cross the boundary together, but not at the same rate.

Dead Cells: action mastery inside an unlock structure

Motion Twin describes Dead Cells as a rogue-lite action-platformer with an ever-changing castle, permadeath, no checkpoints, and nonlinear unlocks. Real-time execution sits beside route knowledge and the management of a temporary build.

An unlock does not preserve the exact weapon carried at death. It changes the system of future possibilities. That distinction matters: a broader pool can increase variety without guaranteeing a stronger opening, while other permanent systems may alter power more directly.

Rogue Legacy: inheritance makes persistence explicit

Cellar Door Games calls Rogue Legacy a genealogical rogue-lite. Death passes play to an heir, while collected gold can improve a manor and produce permanent advantages. The fiction and the upgrade economy both make the boundary visible: this character ended, but the lineage did not.

These examples do not reveal one feature that defines every roguelite. They show why the reset map works. Hades preserves an evolving home and narrative relationships; Dead Cells foregrounds action mastery and nonlinear unlocks; Rogue Legacy turns inheritance and permanent upgrades into the premise. The same label points to different contracts with failure.

Metaprogression is not one thing

“The game has metaprogression” sounds precise until two systems are compared. Permanent state can do at least four jobs:

• Power: raise health, damage, resources, or starting strength.
• Breadth: add weapons, cards, characters, routes, or modifiers to the future pool.
• Control: let the player influence randomness or choose starting conditions.
• Context: advance story, relationships, codex knowledge, quests, or presentation without increasing combat strength.

These jobs create different difficulty curves. Power can let persistence compensate for imperfect execution. Breadth may make the game more varied and, in some cases, harder to predict. Control reduces variance without necessarily raising the maximum. Context can make a failed run meaningful without making the next one easier.

When evaluating a game, replace “Does progress carry over?” with “Which category carries over, and does it make the next attempt stronger, more flexible, more legible, or simply more interesting?”

“Run-based” is a useful modest description

“Run-based” is not a formal standard, and it should not be presented as an agreed replacement for roguelike or roguelite. It is a practical fallback when the repeated-attempt structure matters more than lineage.

The term tells a reader that play is organised into attempts with a beginning, an end, and some reset. It does not promise procedural generation, permanent death, turn-based movement, permanent upgrades, or any particular combat style. Those details still need to be supplied.

A concise description can therefore use the broad term and add the reset contract:

A real-time run-based action game with generated routes, temporary builds, permanent option unlocks, and story that advances between attempts.

That sentence carries more purchasing information than an unsupported argument over one letter.

Five questions to ask before buying

1. Is play turn-based, real-time, or both?

This changes the dominant pressure. Deliberate spatial calculation and rapid execution can coexist, but the input clock still shapes the experience.

2. What is rebuilt each run?

Maps may be fully generated, rooms rearranged from authored pieces, rewards shuffled through fixed encounters, or routes left mostly stable. “Procedural” can describe any of these at a distance; the amount of fresh interpretation is what matters.

3. What does death remove?

Ask about character, inventory, route, quest state, currency, story, and difficulty settings separately. A game can use permanent death for the current body while preserving most campaign progress.

4. Does persistence add power or possibility?

A numerical upgrade and a new item entering the random pool are not equivalent. One raises a baseline; the other changes future decisions.

5. What makes the next run different?

The answer may be a new world, a new build, a new story event, a stronger starting state, a difficulty modifier, or only the player’s improved understanding. This is the heart of the reset contract.

Use labels as coordinates

Call a game a traditional roguelike when that history genuinely clarifies its turn structure, systemic world, consequential death, and procedural exploration. Use roguelite when the lineage is useful but important structures—often real-time action or persistent progression—move away from that prototype. Use run-based when repetition and reset are the main promise and a historical claim would add more heat than information.

Then explain the map. A genre label should take the reader to the right neighbourhood; it should not pretend every building there has the same floor plan. What survives the reset tells us what failure costs, what progress means, and which skills the next attempt will actually ask the player to bring back.