The Roguelike – A Design Analysis

The Roguelike – A Design Analysis

The roguelike is a genre that has been gaining momentum and popularity, or so it appears to the bystander. The indie scene is being flooded with games that announce themselves as roguelikes, roguelike-likes or the newest word-salad: rogue-lites.

When talking about the roguelike genre I will be thinking of the classical view, the Berlin Interpretation. This is not to say that I am a purist and will try to force the Berlin Interpretation to the single last item on the list, however; roguelikes as a genre have a discoverable design-philosophy behind them, an irreducibly complex core, which makes them what they are for a reason.


When talking about the roguelike the first association to jump to is usually their random nature. Random levels, random enemies/placement and random loot is often seen as an indicator for a roguelike. More often than not, these elements are also held against them as being arbitrary or unfair, especially when coupled with another staple of the genre: Perma-Death (the permanent death of a character). Players often confuse these elements as the test itself when in reality they are just the elements that make up the test.

Historically, the roguelike comes from the game Rogue (1980) and hence as a genre description falls somewhere in line with “Doom Clone”, which we have since abandoned as a descriptor for First-Person Shooters.

The original Rogue was a loose adaptation of the Dungeons And Dragons license in which the player explored a multi-level dungeon with procedurally generated floors. The computing limitations at that time prevented high quality graphical output and so the game was entirely displayed in stylized ASCII format, with the ‘@’ signifying the player position. Other characters denoted either obstacles, rooms, items or enemies, often additionally color-coded with upper and lower case lettering.

For the longest time this method became a staple of roguelikes and the genre was often criticized for its arcane interface, and rightfully so. This overcomplicated display and control method earned the roguelike a reputation of being notoriously inaccessible and punishing towards players. Today’s traditional roguelikes have, more often than not, done little to alleviate this problem of accessibility.


Hard to read for newcomers.

Design Analysis

The perception of roguelikes is currently predicated on their random nature, which I consider to be a wrong interpretation of their mechanics. I even consider the usage of the word “random” in this context to be a misunderstanding. Roguelikes are not random. Their design uses random elements to simulate variance in the test-environment; in other words, the randomization is not where the game is at.

As an example, Poker (the card game) also features an element of chance in its design. The dealer (or player) draws from a deck of 52 randomized cards and players try to complete card sequences that have varying values. However, drawing the cards and successful completing a sequence is not the game itself. The game proper is a mixture of tactics and strategy with elements of psychology and mathematics. If summarized, it could be described as “playing the odds” in the sense of consistently making informed decisions that will most likely lead to success.

When looking at one round of Poker, the game can seem entirely arbitrary and random. However, over multiple games, a pattern will crystallize that separates the players with good decisions from the ones with bad decisions.

The typical roguelike behaves similarly in that the irreducibly complex center is a test of consistently making the best decision that lead to success, reinforced with permanent lasting consequences (perma-death). As an example, the decision to use a health-potion that refills 90% of the player’s health when they are at 50% is a strategically bad decision. In this instance, resources were managed badly when there existed the possibility of needing that extra 40% later.

Furthermore, roguelikes are not entirely random in their nature but rather present a limited variance in the mentioned test-environment. Procedurally generated dungeons/levels/floors often have specific limitations on enemies, items and hazards that can spawn, often limited by their level compared to the dungeon/floor level. It is highly unlikely or, depending on game impossible, to be pitted against completely impossible odds and fight creatures that are entirely unbeatable. (f.ex. encountering Elder Dragons on Floor 1)


See that dragon? You can ride it!

Traditionally, roguelikes are very challenging because they expect the player to always perform at the peak of their ability and hence they treat newcomers with the same harshness they would a seasoned veteran of the game. This is not a bad thing, as the design of roguelikes relies on the permanent consequences that befall the player as a teaching-tool. The sooner the player realizes that every action and every decision counts towards beating the game in the long term, the sooner they start to learn to try and perform smartly. It is a test for the ability to create a best-case scenario but plan for the worst-case scenario.

In a way, the mechanism is there to discourage the player early on from treating the challenges with indifference or relying on his characters statistics to beat them. Contrary to public perception, roguelikes do not substitute the player’s prowess with stats and levels (as f.ex. RPGs tend to do) of their chosen avatar, but rather treat the doll as a tool to test the player’s strategic and tactical thinking. Roguelikes are far from unfair or arbitrary in this respect; losing is generally predicated by the ability of the player to make the optimal decision at any given time rather than the game presenting a faulty challenge.

As an example, in the game FTL: Faster Than Light there are multiple ships at your disposal. Some might think that running the Federation Cruiser is easier than running the Stealth Cruiser. After all, the Stealth Cruiser has no shields and a lower crew count in comparison to the Federation Cruiser. In practice however, the Stealth Cruiser is the superior ship; it simply requires different decisions than running the Federation Cruiser.


If I may elaborate, The Federation Cruiser can soak up a good amount of damage by way of shields, but the Stealth Cruiser can entirely negate the attack with timing. Additionally, the Stealth Cruiser has one slot less to upgrade (shields) which mean other systems will be stronger for it, as the player has the same amount of resources available but less slots to upgrade. Furthermore, there are situations where shields are disabled and a ship that relies on its shields is at a strong disadvantage, while the Stealth Cruiser is no worse off than before. There is a consistent way to win with both ships; they just require different decisions to be made. It all depends on the mastery of the mechanics a player should have ideally acquired during his failed attempts.

This is not to say that all roguelikes are perfect representations of this design philosophy of course. Some games are designed to be unfair tests in a zeal to make them challenging, and those are just bad games.


Given that the roguelike is a test of informed decision-making, there are certain design elements that do not necessarily mesh well with it. For example, the turn-based nature of traditional roguelikes is a calculated decision as to not present the “wrong test” to the player. Even more modern iterations like FTL are not real-time; they feature an action-pause system purposely to not test the player’s reaction time and Actions-Per-Minute limits.

The roguelikes design-philosophy sometimes simply stands in direct opposition to other genres philosophies.

Platformers for example, at their core, test the players input precision, mastery, reaction time and their memorization. These stand in direct opposition to the roguelike elements as variance in either of them would make the tests arbitrary, denying the player mastery over them. If Mario jumped slightly differently every time the game resets the test would become unfair and mastering the core mechanics impossible. Alternatively, if the levels were procedurally generated, the player would be denied the chance to learn from his failures and the difficulty of the challenge would be entirely random, leading to a bad test environment.

This is true for other genres like RPGs; persistent characters which delegate challenge to the avatar rather than the player invalidate the test of letting the player creating a best-case scenario in a limited environment. A player can simply negate the test by making his avatar more powerful by continuous grinding and brute-forcing his way through the game.


The Present And Future

Recently many games started mixing and matching roguelike elements without taking these limitations into consideration, creating hybrid games that more often than not just don’t work very well. A recently released “rogue-lite” called Rogue Legacy introduced metroidvania, platforming, bullet-hell and RPG-like persistent character upgrades into the Rogue formula, essentially invalidating the philosophy altogether. An upgradeable persistent character by itself removes permanent consequence from the roguelike philosophy, as death becomes a thinly veiled tolerance test (e.g. grinding) instead of the tool it is intended to be.

Another example is the game Routine, which is being described as a First Person Sci-Fi Horror Roguelike on the grounds of having randomized hazards and levels. If that’s the case, I guess we can add Diablo 2, Torchlight and a slew of other games to the list of roguelikes.

I don’t want to start another genre definition discussion (for now), but just slapping elements that are vaguely reminiscent of a roguelike on a game and calling it that isn’t what we should be expecting from designers. Don’t get me wrong, I think all these concepts, like inheriting traits in Rogue Legacy or random room-layouts in Routine, are fantastic ideas. Just not for their respective games and in the context of their designs, considering what the core challenges of these games are.

Meanwhile, the traditional roguelike lives on but only in partially obscure indie titles; while FTL was a great Kickstarter success, its popularity was rather limited. Its future is uncertain, as challenging games, especially ones with failure as a teaching tool, are wholly unpopular in the industry.

The roguelike is significantly different in its test-philosophy and players might easily be put off by their initial experience with the game. However, there is a deep satisfaction associated with learning and mastering these types of games that transcends other, similar, statistics-based combat simulations.

Maybe a fresh coat of paint and improved accessibility could popularize this traditional genre again. Who knows?

  1. Torquey says:

    Honestly, rougelike titles aren’t in a state if decline. FTL and Binding of Isaac have both been major successes.

    I never thought that Rouge Legacy would have a problem meshing all of those platformer/RPG/rouge elements together. I guess it’s not that good?

  2. Taylor says:

    Roguelikes aren’t in decline, and more casual roguelikes have larger audiences than you seem to give them credit for. They’re gaining popularity as more people either learn to get past the text-graphics, or find roguelikes with a higher level of accessibility. As far as I’m concerned, they aren’t dying until Toady needs a real job.

    I do think that the ridiculous amount of games that have terrible “roguelike” mechanics of crazy amounts of randomness and absurd difficulty is troubling, but the mainstream’s misunderstanding of where a good roguelike’s difficulty comes from is nothing new.

  3. I need to know, where have I said that the Roguelike is in decline?

  4. Hardcore Bro says:

    You mention Platformers as being incompatible with roguelike, but what about Spelunky? That was pretty much the first “with roguelike elements” game that wasn’t an actual roguelike, way before Binding of Isaac for instance.

  5. AngryFrenchKanadian says:

    I don’t think I’ve ever head the Routine dev refer to their own game as a roguelike. This is eurogamer’s fault, not the dev’s.

  6. Johanz says:

    Good analysis Aleksander! You really hit all the big points on why I love roguelikes. And games journalists should read this article before throwing the word roguelike to every game that has a random dungeon or perma-death. Random generated dungeons is what it is and so is permanent death. You don’t call Myst a FPS just because it is in first person. So why slap on the name of a genre just because you borrowed an element (that isn’t wholly unique to that genre to begin with).

    But yes, good article. And I do think that roguelikes would be more popular if the graphics or overall presentation was better. I mean, look at Grimrock, it was a pretty basic dungeon crawler but sold well and I bet that was due a bit to it having a nice presentation and decent graphics, especially for such a small team. Same with Dredmor, I’m sure it won over fans with its humor and pixelstyle graphics (I’m not a super fan of it all but it was well executed).

    I think that graphical presentation of a game can go a long way to rope one into the experience. If you just play the game for that extra two or four hours just because you like the graphics and how neat it all looks, that might be the extra time you need to learn the game better and understand why it is fun to play. But yes, if one is interested in a game, one will learn it (or try to) regardless of the graphics. Anyway, I don’t know what point I was trying to make but I do think that many roguelikes would benefit from having nice interfaces and nice graphics.

    • I concur, currently presentation sells a game. Presentation and trailers.

      It’s hard to make a trailer on the base of ASCII graphics.

    • Gig says:

      You pretty much stole my line. :) Calling Routine (or any of these other games) a roguelike for having procedurally generated environment, permadeath or a grid-based movement a roguelike, is like calling Fallout 2 a First-Person Shooter because your character has a gun, or Fallout NV a Sandbox game because it has open environments. Those are but elements of a whole and sticking one into a game does not a roguelike make. This false classification really irks me because it stems, in my opinion, from the dev’s desire to have roguelikes’ positive characteristics (like difficulty or replayability) “rub-off” on their game by association.

      I might not even call Grimrock a roguelike, because very often it DOES depend on quick reaction and instincts; in the same fashion, even a game of Dungeons of Dredmor where the player took the opportunity to disable permadeath is no longer a roguelike – if you are never afraid of making a bad move, because you can brute force your way out of any situation by repeatedly reloading, you’ve eliminated one of the basic elements that shape how (and why) we play a roguelike (this is also why savescumming is so frowned upon in the roguelike community).

      For instance, I’m currently in the middle of a Nethack game where my character has just begun sliming (slowly turning to green slime, a quick and fatal process), and have only 3 options: I can take off all my rings, take out of my bag two new ones that will transform me to something immune to sliming, and wear them one by one; or I can take out a blank scroll and magic marker and write and then read a scroll that would set me on fire; or I can drop everything I have, quaff a potion of speed and sprint for the stairs to the previous level where I can safely pray to my god for help; but I can only choose one, and I haven’t played it in two weeks because I’m afraid of making the wrong choice and lose my character. If I were able to save, if the game wasn’t turn-based or if the dungeon layout was planned ahead with this was a scripted scenario than this wouldn’t have been an issue.

      However I disagree on the point made about graphics and accessibility. It may have started as a computational limitation, but the simplicity and immediacy of the top-down (or isometric) ASCII based visuals serve the game best – they’re merely abstract symbols on a board, representations of a unit or object with fixed characteristics for the players to take into account when calculating their next move, much like in chess. Immersion in roguelikes, at least in my experience, has nothing to do with the way the game SHOWS me that fire-breathing red dragon, and everything to do with how the game PLAYS with it; I couldn’t care less what it looks like, all I care about is that I’m terribly flammable and the door behind me is locked.

      Great article on a subject close to heart. Really enjoyed reading it.

  7. RevDoktorV says:

    Tales of Maj’Eyal is very good, and manages to remove a lot of the frustrations that novices have with traditional roguelikes without sacrificing the foundations of the style either. The graphics, while very basic, are enough so the player doesn’t have to memorize huge numbers of abstract glyphs, and there are no potions to manage. Instead, the character accumulates what I think are supposed to be magical tattoos that have unlimited uses but with cooldowns, and using too many at once reduces their effectiveness. With this system in place, the long-term aspect of resource management is de-emphasized significantly, without reducing tactical complexity.

    Admittedly, I’m not a roguelike veteran; my only frame of reference for traditional ones is Dungeon Crawl Stone Soup. ToME seems to eliminate a lot of the the things that frustrated me about Stone Soup, but the only thing that’s explicitly easier about ToME is the lack of food tracking.

    So, there you have it, the genre certainly doesn’t appear to be dying, and it seems like it can be made accessible to newcomers without stripping out the elements that make the games unique.

  8. Dave Kurgen says:

    I’m surprised that you didn’t mention Sword of the Stars : The Pit (currently on sale on Steam right now), . I would say it ranks with ToME, Soup and FTL for the most recent spiffy rogue-likes. But SOTS : TP does retain a nice crafting system that I’ve only really seen in say Adom or Dreadmor. Also a good chunk of success in that game comes from understanding how the non cobatitive elements work (e.g deciphering alien computer messages and devices, engineering, bio medical skills,and others). And the most recent expansion for it does up the difficulty as well.

  9. Virtual Academic Resource Center provide you free access to hundreds of books, scientific journals and articles from the field of Psychology, Philosophy, Sociology and Educational Sciences.

  10. Mark says:

    I hate reading articles when they get some basic facts wrong. You wrongly categorize Rogue Legacy as a game that doesn’t have permadeath. I doubt you have ever played it. When you die – you are dead. There’s no coming back. Any upgrades you earned are passed on via blacksmith to your children. When you enter the dungeon again, you are presented with a randomized castle, unless you pay to have the castle non-randomized for one attempt. You really should try playing the games you talk about.

You can use basic HTML in your post. Gather Your Party will never share your email address with anyone, ever.