How would you do a roguelike game?

[Shinizak]

I have a casual thought experiment. If you HAD to implement roguelike features into your game, how would you do that?

https://en.wikipedia.org/wiki/Roguelike

[Thrudd]

Roguelike comes from a video game that was trying to replicate elements of old D&D.
So, I would start with the base of Basic/AD&D, at least in the sense of how they are run and implemented, if not the rules themselves.
To add old D&D elements to another game: Create multi-level dungeons, possibly using a random dungeon generator/table. Decide what this place is supposed to be and stock it with monsters and treasure (much of which could also come from random tables). Exploration of the dungeon takes place in "turns", check for wandering monsters at regular intervals. Each level of the dungeon includes stronger creatures. Don't fudge dice, don't spare PCs from death. If some of them die, the surviving party can retreat from the dungeon, find some new members in town, and then try again (with some new monsters possibly having moved into the parts that were already cleared). Generally, award XP for treasure returned from the dungeon in addition to the slaying of monsters, and only award it after the party returns to town.

[Vahnavoi]

Roguelikes are computerizations of old school D&D - some very literally so, with for example (unfinished) Incursion roguelike being an outright rules port of 3rd edition. Any tabletop roleplaying games using random terrain generation for dungeons and wilderness, where player characters can die and means of recovering characters are limited, already fit the definition of a roguelike.

So you might as well ask me how would I do an old-school game.

I've used several procedural map generators through the years, and my advice would be: if it's dice and chart-based, to do the randomization step and complete the map before the game proper, at least to some extent. Modern computers are fast enough to get away with procedurally generating terrain as the player goes forward, you as a human aren't that fast and will slow the game down when looking up charts.

There are faster alternatives. Drawing tiles or cards, with pieces of map drawn on them, can be done on the fly. There are many board games with sufficiently sophisticated tilesets for this very purpose - get those games and steal their pieces. Or draw map symbols on post-it notes yourself. It's possible to use abstract notation, similar to ASCII display of roguelikes, for this purpose. I used to draw my tabletop maps in notation similar to Angband roguelike. Last, and possibly fastest if you have any drawing skills, is to improvise and draw the map using such notation as player knowledge expands. This works best for square or hex crawls.

My favorite procedural generators at the moment come from How to Host a Dungeon, Veins of the Earth (LotFP), Red & Pleasant Land (LotFP) and Praedor adventure cards. Let's go through these in order:

How to Host a Dungeon is a single player game that nets you a sideways cross-section of a dungeon network, complete with its history. You can then draw a top-down map of each room (as usual) and how they connect on per-floor basis. Put together, you have a three-dimensional map of a dungeon.

Veins of the Earth is sort of similar. It uses a special notation to create a two-dimensional map of a three dimensional network of caves, creating a sort of flowchart for how caves connect to each other. It then has plenty of content you can randomly place in your cave.

Red & Pleasant Land is more traditional has a lot of useful monster and room randomizers which you can use to create dungeon-like environments. It also has a system for randomly generating diplomatic relations between monster populations.

Praedor adventure cards have two sides. One side has a landmark, such as a ruined aquaduct, and the other has an encounter. There are more specific rules for how to use these in Praedor: Kirottu Kirja, but I'm not going to focus on those. The important part is that the randomly dealt landmarks create a loose map of a post-apocalyptic city, with associated dangers for each one.

So that's it for the map parts. Let's talk about death.

It is really important to accept that characters are going to die and become irrecovably lost. Both old-school games and most roguelikes share the trait of being games of incomplete information: you don't know what's ahead, you don't know if it's safe and, most importantly, you CAN'T know. This means, especially early on, that the only way ahead is trial and error, with error leading to character loss.

So have a lot of replacement characters at hand. Fast character creation is a plus, but having a lot of pre-made characters to throw at your game scenario is better. Save copies of character sheets in their start-up form and allow players to start from scratch with mechanically identical characters. Embrace conceits such as characters having large families, or coming from large organizations with identically trained individuals.

Related, create and embrace conceits allowing passing of information from dead characters to new ones. Maybe someone recovered diary of a deceased character, to excuse why a new character knows what the old one did. Even better, have players actually keep these diaries. If you have a cast of revolving players, not just characters, these in-character notes create an entire new layer of gameplay as they're passed from player to player.

Why do this? Because while you can scrap your map, asking players to scrap all they've learned of your game is a bit much. In fact, that's not how these games are normally played. Characters and maps change, but a player naturally retains knowledge of each playthrough. That is how they get better at these kinds of games and can eventually complete them. So you want to work the permission for them to do this into your game. You want to let characters know everything their players have come to know through legit gameplay. If players want characters to be ignorant of what's happened before, leave that choice up to them. Save complaints about metagaming for cases where the metagame is genuinely breaking the rules, such as reading game master's notes or something.

What else? Oh yes, challenge. Here's the deal: most obstacles in these games are NOT "equal challenges". Far from it. The challenge comes from getting through ALL of them in sequence. Remember rules of iterative probability: even if chance of failure for individual obstacles are small, the aggregate chance of failure keeps climbing up the more obstacles there are. A 5% fail chance per obstacle becomes 9.75% fail chance for two, 14.26% for three, 18.55% for four, so on and so forth. The corollary to that is: most monsters should be weak, most traps should be surmountable, a player paying attention should be able to avoid gambling on their character's life too often, especially for parts of a game they already know. If that doesn't sit well with you, if you want each obstacle to be dramatic, then the answer is to decrease total number of obstacles. Pay attention to where difficulty of your game is coming from. If it's just a consequence of the random number generator you're using (f.ex., dice), your game is just glorified Snakes & Ladders. Don't do that. Eliminating player skill isn't what you want to do, you want to engage player skills.

[Eldan]

I'd use a very stripped down rules system. Part of what makes good roguelikes palatable is that you can jump right back in and try again if you die. So character generation would have to be 10 minutes or less, preferably under 3 minutes.

If I have to do anything except choose two or three stats and write down a character class, I don't want to do roguelikes with it.

[Hytheter]

How roguelike are we talking? I'd say the most defining feature of a roguelike is frequent death that results in a total reset, erasing progress and creating a new set of starting conditions. If this is the kind of play you're looking for then the game needs to be fast, especially combat. Tabletop games are slow compared to electronic ones, which means the turnaround for each 'run' of the dungeon will take much longer at the table and risks becoming a slog even if your players are otherwise into this hardcore style of gaming. Choose a fast game system, have several map iterations and characters (if they're supposed to play different ones in each run) prepared before the session.

Of course, if you relax on strict adherence to the genre you can breathe a little more easily. Consider letting players keep some form of progression (see: roguelites) and/or have periodic checkpoints so they aren't starting from zero every time they die. For example, if the dungeon has five floors let them skip floors they've already cleared once. You'll lose that hardcore experience but I think it will be more palatable for players given the pace of tabletop gaming.

You might also want to have some sort of in-universe conceit explaining the dungeon's shifting nature and the players' ability to retain information across runs. Maybe the dungeon takes place in some sort of dreamscape or virtual reality such that the characters don't really die. Maybe each new party are reincarnations of the old one, inheriting their memories but faced with a dungeon that has changed over time. Maybe the dungeon is for some bizarre life-or-death game show, constructed anew for each set of contestants who benefit from the footage of prior attempts.

[Mastikator]

I'd go whole hog on the roguelike aspects.

Randomly generated dungeon with random encounters and random loot.
No over world exploration, only dungeon. Only go down. Each level increases difficulty by 1 CR. Final level has final boss, you kill final boss = you win, game over.
Traders sporadically appear with limited (but potentially very powerful) trading options, you can also kill traders and steal their loot but they should be "deadly" level difficulty since they survived down here. Sentient NPCs can sometimes be talked down and often surrendered to (exchange stuff for life).
No revival spells, death is permanent AND you start at level 1 if you die. You level up through EXP. EXP is only acquired by defeating enemies. New PCs are encountered in next room. TPK results in new party starts at new dungeon. PVP is allowed (and gives EXP). Roleplaying may result in rocks falling, minmaxing/metagaming and powergaming may result in roleplay EXP :smallwink:
All equipment is breakable, even high level magic items can AND WILL be destroyed, AOE will damage equipment, environmental damage will damage equipment. Enemies will try to drink their potions rather than let you loot them of their corpses.
Food and water is scarce and must be tracked.
Resting incurs risk of encounter. And sometimes party will encounter sleeping monsters. But monsters do not respawn, if you clear a level it is safe.
Monsters will sometimes infight. Monsters will sometimes flee. Monsters will sometimes call for backup.
Skill challenges are brutal, unforgiving and will err on the side of unwinnable.
The DM is encouraged to be unforgiving, cruel but never to cheat. All rolls must be rolled openly, by everyone. Players knowing monster's AC/HP/etc is fine.

[Tvtyrant]

I have a casual thought experiment. If you HAD to implement roguelike features into your game, how would you do that?

https://en.wikipedia.org/wiki/Roguelike

So what I would do is make a city next to a dungeon entrance. The Dungeon is a deity that eats people as sacrifices, but also spits out treasure to get people to go in. It remakes the dungeon when you enter and leave, and has different floors that get harder as you go down (too deadly to start with and everyone would avoid it altogether.)

Think of it like a giant mimic.

[LibraryOgre]

One option I might steal from Dark Sun, to keep the game moving a bit, was the character tree.

Every player makes up 3 characters. You use one at a time. If your active character gets a level, then one of your other characters does, as well. If your active character dies, you sub in one of your other characters, and make a new one to fill out the tree... but the new one is a starting character.

[TexAvery]

In with the easy death needs to come RAPID leveling/ character advancement. A session should be able to raise a character through 8-12 levels, if they might die at any point and be expected to keep restarting. The typical D&D campaign is too slow even for its designed purpose.

[TexAvery]

So what I would do is make a city next to a dungeon entrance. The Dungeon is a deity that eats people as sacrifices, but also spits out treasure to get people to go in. It remakes the dungeon when you enter and leave, and has different floors that get harder as you go down (too deadly to start with and everyone would avoid it altogether.)

Think of it like a giant mimic.

Ooh... not a giant mimic. That is a giant Venus fly trap, or anglerfish. Or a cosmic horror, which needs to feed on powerful souls. That gives it a motivation to lure people in and slowly level them before killing them, instead of just putting a dragon or lich on D1 and killing everyone. Maybe make it addictive to the characters somehow, to avoid them grabbing some levels and loot and retiring; the only option is to keep delving in hopes of killing the core.

It's also now reminding me of Alistair's demiplane in Archive 81. Which is a podcast I can't recommend highly enough.

[MornShine]

One things that I would definitely consider is 'unlockable' options, probably achievement-based. Having no levels or gear carry over when you die is fine, but getting more options-- not stronger options, just different-- can still give players something to be proud of.
It can also scale the complexity with the player's experience; on the first runs, you play a fighter, but around the fifth you unlock the rogue, etc. which persists through all playthroughs.

[Pauly]

Warhammer Quest is pretty roguelike, and had a very good way of generating random maps. The main difference between WQ and a true roguelike game being after each map you returned to town rather than keep going deeper into the dungeon. Iirc, treasure was awarded by drawing cards from a treasure deck, which is a fast easy way to generate treasure. With a bit of tweaking you probably could turn it into a more proper roguelike experience.
From the top of my head you would want to make some changes:
- add traders
- add secret doors that lead to a more extensive part of the dungeon.
- remove end of level bosses, but make the average encounter more challenging.
- Have a random roll for each room to see if it has the ladder leading down.

[False God]

I have a deck of 3X5 cards that I've drawn straight, corner (left and right), 3-way intersection, 4-way intersection and dead-end. As we go we draw out the map on a piece of graph paper (unless I run the resetting "The Cube" style where it seals off every X number of rooms).

Each round I draw a new card and that's the next ~30ft of the dungeon. Then I roll percentile to generate what in that room (percentile chart for treasure, monsters, nothing, so on...), and I have a couple variations of this chart for more loot, lots of monsters, high-magic depending on what my players and I are in the mood for.

[Pauly]

Other things to include in a rogue like game.
- unidentified magic items,
- some magic items being cursed.
- finding out what an item is by trial and error.
- randomized level ups, what skills/abilities/fears you get will not be the same so players can’t pre-plan characters but have to adapt to what happens.
- scrolls of enchant weapon being more common than finding already magical weapons.
- Monsters that steal your stuff such as Nymphs and Leprechauns.

[Psyren]

In a board game it's difficult - the length of time it takes to set those up can make permadeath mechanics offputting, and the limited combination space of the pieces can hurt the procedural generation / replayability factors. For TTRPGs it can be an even harder sell depending on how involved character creation is and how much the challenges need to be tailored to the players.

Lovecraftian games like Arkham / Eldritch Horror, Elder Sign and their ilk incorporate some roguelike elements - randomizing the challenges, items, rooms etc., and facilitating "permadeath" by having your dead investigator get removed from play with all their items, only to be replaced by a new investigator who has to start over from scratch. I'd consider taking a look at one of those for inspration.

[Lacco]

...by wanton application of rogues! :smallsmile:

Now, a bit more serious. My experience in roguelikes is based on ADOM, so I may be biased a lot.

Mechanically:
Permadeath
Fast character generation or semi-random character generation
Relatively fast combat (even on higher levels)
Item management is a big thing
Starvation is a thing
Summons, pets and followers are allowed, but usually limited

On GM side:
Overland travel
- hexcrawl
- random encounters on timer
- areas with different threat level (needs to be informed choice!)
- non-random map with random dungeons
- fixed locales (towns, inns, "safe" points of light)

Dungeons
- three types of dungeons: random, semi-random (quest-related) and fixed
- random = loot & randomly generated quests; where they are is fixed but what they contain is different
- semi-random = quests are predetermined, but the locale changes; utilize mainly dungeon geomorphs

Quests
- overall story = overall fixed quests
- random quests (e.g. given by local NPCs)
- few hidden quests for crazy rewards, available for most characters
- treasure maps, dungeon maps, etc.

Items
- mainly randomly generated, with quest items being fixed
- rule: items are not placed based on their usefullness (so you are playing an orc brawler? here's this magic book as a loot!)
- cursed items are FUN! (use Dwarf Fortress definition of FUN).
- potions need to be identified or tried & tested
- no shopping during chargen, you just get the basic items
- mostly totally random loot...

[kyoryu]

I'd probably just run an old-school open table megadungeon.


Town sits on/near the dungeon.
Whoever comes, comes.
Players have multiple characters.
Each session has the people that showed up pick a character to go in the dungeon, or roll one up.
XP=GP for sure. Roguelikes generally work best if you don't kill everything (you typically will run out of resources). The goal is to maximize the treasure you can get in a single session, not to murderdeathkill the world.
A single session is one trip into and out of the dungeon.
Roll everything in the open.
GM should be tough, but fair.
Death can happen. Deal with it. At higher levels, it might be reversible, but at a cost.
The dungeon "evolves" over time. Maybe not layout so much, but what critters are where.
Randomized treasure. You figure out how to make the most of what you get.
Magic items generally not available for purchase, unless someone else brings them out.


Probably more, but that's off of the top of my head.

[Telok]

I'd just run Paranoia. Already have the requsite sector, room, & corridor generator in a spreadsheet, other randomizers are in the book. I prefer the anniversary edition.

Hardcore mode would be just one clone instead of six. Thats fine as coming up with a punny name is the longest & hardest part of char-gen.

[Vahnavoi]

As a comment on what kyoryu said, the reason you don't kill everything in a roguelike is typically because you can't - new monsters keep arriving to replace old ones and genociding even a single species so this doesn't happen is very hard. This isn't hard truth of the genre, though. Plenty of roguelikes more or less expect you to murder and loot all the monsters on a level, while still allowing pacifist runs to be viable. You can vary this significantly.

[kyoryu]

As a comment on what kyoryu said, the reason you don't kill everything in a roguelike is typically because you can't - new monsters keep arriving to replace old ones and genociding even a single species so this doesn't happen is very hard. This isn't hard truth of the genre, though. Plenty of roguelikes more or less expect you to murder and loot all the monsters on a level, while still allowing pacifist runs to be viable. You can vary this significantly.

Even in early levels of like Nethack it's often a bad strategy - you have limited resources, and wasting them on extraneous exploration and kills is a good way to an early death.

I'd pretty much argue that that's a big step in effectiveness - getting out of "explore and kill everything" mode and into "what am I trying to accomplish?" mode.

[Tvtyrant]

Ooh... not a giant mimic. That is a giant Venus fly trap, or anglerfish. Or a cosmic horror, which needs to feed on powerful souls. That gives it a motivation to lure people in and slowly level them before killing them, instead of just putting a dragon or lich on D1 and killing everyone. Maybe make it addictive to the characters somehow, to avoid them grabbing some levels and loot and retiring; the only option is to keep delving in hopes of killing the core.

It's also now reminding me of Alistair's demiplane in Archive 81. Which is a podcast I can't recommend highly enough.

It's even symbiotic in a sense, it wants to make bigger heroes and heroes want to get bigger. You could die, or become a wealthy and unstoppable mage/warrior.

If I was adapting 5E I would replace rests with rest potions you buy in town as well. An actual rest would take weeks, the potions take an hour or a night but make the party constantly go back for more potions. Now that's grinding.

[kyoryu]

In with the easy death needs to come RAPID leveling/ character advancement. A session should be able to raise a character through 8-12 levels, if they might die at any point and be expected to keep restarting. The typical D&D campaign is too slow even for its designed purpose.

I think easy death and "One True Party" don't work together. You need a structure where people have multiple characters - think more like XCom or something.

[Curbludgeon]

I imagine I'd cleave closer to a roguelite than -like. I'd prefer a three entry stat line (Muscle/Mysticality/Moxie) to four, have simple classes with 2-3 abilities, and have a scaling collection of treasure which comes available as a result of something akin to achievements. Discovering synergies between the items (and to a lesser degree class abilities) would constitute the majority of system mastery.

[LibraryOgre]

I imagine I'd cleave closer to a roguelite than -like. I'd prefer a three entry stat line (Muscle/Mysticality/Moxie) to four, have simple classes with 2-3 abilities, and have a scaling collection of treasure which comes available as a result of something akin to achievements. Discovering synergies between the items (and to a lesser degree class abilities) would constitute the majority of system mastery.

Sounds like it might work well with PbtA system... 3 stats, a couple class abilities, defining treasure as the new "levels".

[Grek]

Have them play out (and plan for) the Orb Run.

For those unfamiliar, in most Nethack-inspired roguelikes, the central premise is that there is a Dungeon which goes down however many levels, and at the bottom is an amulet, an orb or some other artifact. Your job is to get down there, get the artifact and then return to the surface. Dungeon Crawl: Stone Soup has a fun twist on this gimmick though: once you get the Orb, monsters start spawning around you. Suddenly you're not just looking at a cake walk through already cleared levels - you're fighting in those same old environments with new enemies spawning in constantly.

Then it's just a question of including enough set pieces and puzzles that going through the dungeon in reverse is an interesting third act to your story. If they defeat enemies on floor three by cutting a rope bridge, that rope bridge is still gone when they go to leave. If they disarmed a trap, that trap is disarmed still - but maybe it could be re-armed if they pull the right lever. If you made them get keys for locks, they can now lock those locks as they go up, to seal off enemies they don't want chasing them. If they descended down a shaft to get in, they now have to climb it going back out, and that's much harder. That sort of thing.