Roguelike vs roguelite: permadeath names two rules

Permadeath is used as if it were one rule. It is at least two. In the strict version, the run is the whole game: die, and the map, the items, and the knowledge in the character are gone. In the common modern version, the character dies and the account keeps a currency, a shortcut, or a tool. Both can be honest. Calling them the same word is how players arrive angry.

The design question is what the grave is allowed to keep. Keep nothing, and the tension is pure and the later hours must still be fun for a newcomer, because every player is a newcomer. Keep a meta layer, and you are designing two games, the run and the account, and the account can make the run trivial if you are not watching. Name the grave before you name the genre.

Permadeath is not a single rule. In one game it wipes everything, and in another it wipes only the character while the account keeps what it earned.

The label goes back to Rogue, the 1980 dungeon crawler, and originally meant games built closely in its image. Indie developers have since stretched it over a much wider field.

Two traits usually survive that stretch. Levels are built fresh by the generator every time you sit down, and dying sends your progress back to nothing.

Spelunky honors the second one literally. Die, and the game forgets you: the next run opens with the same health, the same bombs, and the same money the very first attempt had.

Rogue Legacy softens it. Gold you found on a doomed run buys upgrades to health, magic, attack power, and the rest of your stats, and those upgrades wait for the next character you roll.

Players tend to reserve roguelike for the first pattern and roguelite for the second. It is worth arguing about, because each pattern sets up a different bargain between failure, progress, and skill.

A roguelike with no persistent upgrades measures one thing: how good you are right now. Spelunky and Enter the Gungeon are built this way.

Because nothing carries forward, the difficulty across a hundred attempts is a flat line. The game you lost to yesterday is exactly the game waiting for you today.

Getting past that line means changing yourself instead of the game. You learn the rules, drill the controls, read enemy behavior, and build a working model of how the world fits together.

Between you and the final boss there is ability, knowledge, and whatever the generator decided to roll. That cuts the other way too: a player who is already excellent hits no gate and can finish tonight.

The cost is that some players never reach the end at all. The game asks you to clear something hard in a single sitting and offers no permanent way to make the ask smaller.

Fail to meet it and you keep failing to meet it. One player finds that electric; another finds it grinding.

There is a quieter cost as well. Death erases the run, so an attempt that ended badly can read as time thrown away.

You may in fact be improving the whole time. Improvement just does not produce the pleasure of a rising number, an unlock firing, or a collection filling in.

Roguelites answer that by making the run bank something. Rogue Legacy and Dead Cells both hand you upgrades that outlive the character who earned them.

The result is that nearly every attempt buys you something. A death in Rogue Legacy still counts if the gold you carried out pays for an upgrade that improves the next attempt.

Each run shaves the difficulty down a little. More players get to see the ending, because a character that keeps getting stronger eventually outgrows the obstacle.

For anyone who would have quit a pure roguelike early, that is a way in. Losing is still losing, but it is almost never nothing.

Stacking permanent upgrades inverts the usual curve. The hardest the game will ever be is the first hour, and it eases off from there as the upgrades pile up.

Most games do the opposite and push back harder as you get better. Here your skill climbs while the game's resistance drops.

The nastier consequence is that winning starts to look like a function of character level rather than play. In Dead Cells it can feel as though the run only became winnable once you had bought a few more health potions.

That puts a stretch of grinding between you and the part you came for. And when the ending finally arrives, the credit is murky. Did you get better, or did the game get softer?

Neither pattern is stuck with its worst version. Designers have found ways to keep what each one is good at and blunt the part that hurts.

The standard complaint about a pure roguelike is that nothing accumulates. One fix is to let things accumulate that are enjoyable to own but do not move your odds of finishing.

Enter the Gungeon pays out a special currency for killing bosses. It survives your death and spends in a hub shop, and what it buys is not extra health or fatter ammo crates. It puts more weapons into the pool the generator draws from.

Since those guns are not flatly better than the ones you started with, the balance holds. What later runs gain is range, not a standing discount on difficulty.

Nuclear Throne applies the same idea to characters. The unlockable ones are not upgrades over the defaults; each trades a strength for a weakness, so you get new ways to play rather than rungs on a power ladder.

Presentation works too. Downwell hands out new color schemes for long-term gem collection, changing how the game looks without touching how it plays.

Persistence does not have to be a stat. Hades keeps permanent upgrades, but its most interesting thing that carries forward is the story.

Every trip back to the hub can open up more of the characters, their histories, and how they feel about each other. Items you turn up on a run can be handed over to unlock further scenes.

So a losing streak still moves something. The action may keep defeating you while the world, the cast, and the conversations keep advancing.

That is enough to stop repeated failure feeling hollow, and it never touches the numbers.

A roguelike can lower its wall for a new player without permanently lowering its difficulty. The trick is help that expires or help that does not count.

Spelunky uses shortcuts. Feed the tunnel man what he asks for and he opens a door that starts your run deeper in, in the jungle, with the mines skipped.

Practicing later areas gets much easier that way. The game is also blunt that this is not a real clear: shortcut runs are struck from leaderboard times, and the true ending demands a chain of items hidden across every area from the top.

Help can also be temporary. Into the Breach lets you bring a single leveled soldier into your next attempt, which is a real leg up and a good reason to start again, without discounting the game forever.

The suspicion aimed at roguelites is that hours spent beat skill. The fix is a rule that stops you converting patience directly into power.

Rogue Legacy stations Charon at the door, and he takes whatever gold you have not spent before the next run starts. Scraping together a little from many bad runs and cashing it in later is off the table.

An expensive upgrade therefore has to be earned inside one run. Which means surviving long enough to demonstrate you can play.

Dead Cells writes the rule differently. Money does not survive your death, so you have to reach an upgrade station and bank it first. Progress is available, but only to a player who performs well enough to secure it.

A roguelite can settle the argument in the open. Most of them are tuned so that a good enough player can clear the game having bought nothing at all.

That turns into a challenge of its own: what is the smallest number of upgrades you can win with? The upgrade tree becomes scaffolding for players who want it rather than the only road to the ending.

The appeal of the pure roguelike sits right there. Random generation and permadeath make memorized routes and slow stockpiling useless, so what is left is skill, knowledge, adaptation, and nerve.

The roguelite trades that for a different pleasure: power that builds, and the assurance that a bad night still counted. The price is the inverted curve and a doubt about whether the win was yours or the game's.

Neither is the correct answer. The question a designer has to settle is what a loss is supposed to mean, and whether the progression system protects the value of getting better. You will not settle it on paper. Build both versions, play them, and trust what your own hands tell you.

The difference is easy to state in code. A roguelike has one kind of state, the run, and throws it away on death. A roguelite has two: the run, rebuilt every time, and the meta state, which survives. Keep them in separate objects from the first day, because a value that drifts from one to the other is how a strict roguelike ends up quietly carrying power between runs.

In Godot 4 the usual shape is an autoload that holds the meta state as a custom Resource and saves it to user:// with ResourceSaver, or a ConfigFile if it is only a handful of numbers, and a run scene that is instanced fresh and freed on death. The run reads the meta Resource when it starts and writes to it only at the moments you choose: a death, a boss kill, a shop.

Then test the rule instead of trusting it. Die on purpose, start a new run, and list every number that differs from the first run. For a roguelike the list should be empty. For a roguelite it should match, line for line, the sentence you wrote about what the grave keeps.

In the Flockbay app, describe a run and write down, in the same sentence, what a death deletes. Build that and die on purpose. If something you meant to lose is still in the next run, the sentence was a wish. If something you meant to keep is gone, the sentence was incomplete.

An AI roguelike maker and a roguelite RPG template lean toward the account. Say so, or forbid the account.