What “Roguelike” Meant

15 min readNov 6, 2020

Several recently released games have rekindled the old discussions on the meaning of the word “roguelike”. This post is an attempt to document how the meaning of this term has changed historically.

Roguelikes have evolved separately from mainstream video games: while mainstream games focused on features that sell well, roguelikes were made by skilled programmers who wanted to create games interesting for themselves, and share them with other people with similar tastes. The term was never defined well, with every player using it according to their own feelings; for this reason, I have decided to make this quite subjective. Also, much of the emphasis is put on the community — since the roguelikes tend to be ignored by mainstream media, the fans gathered in online communities, which have been crucial for the evolution of the genre.

The Beginnings (1993)

I have not yet been in any online communities when the term “roguelike” has been coined; luckily, Santiago Zapata has written a blogpost about this. The term was invented by the Usenet community to group several games which they thought they were similar; these games were (1) free (2) dungeon crawl games (3) using ‘character display’ (in-game entities represented by letters and other characters). Gameplay features such as (4) permadeath or (5) randomness were not considered. Roguelikes at that time included Rogue, Moria, NetHack, Angband, and several other minor games.

Early Years (1994)

Around that time, I have been playing Valhalla (more commonly known as Ragnarok). Was it a roguelike?

(1) free —partially: Valhalla was a commercial release of Ragnarok, which was shareware

(2) dungeon crawl — partially: the first major part of the game took part in a forest, although dungeons were prominent too

(3) character display — no: it had simple graphics (although you still can see some letter notation in the map above)

(4) permadeath — partially: while in a roguelike you typically you could not reload your older savefile when something bad happens, and start a new game if defeated, Valhalla featured an “expert mode” with very resticted saving, and a “beginner mode” where you could create a reloadable save every 200 turns. Still, the game was balanced for permadeath.

Yet, the Roguelike FAQ still considered it a roguelike. As it should — the actual gameplay was very similar to NetHack. It was

turn-based (played in turns, like Chess)
grid-based (played on squares, like Chess, positioning matters)
non-modal (battles took part during exploration, rather than in a separate screen)
the details of its world were randomly generated
and it had quite a lot of other minor similarities.

It had some weird conventions, like you could press the ‘!’ key to see all your potions (this was weird for me, but for roguelike fans it should be obvious — ‘!’ looks like a bottle, and was a typical potion symbol in ASCII roguelikes). I wanted to create my own game like Valhalla. Many minor roguelikes have been created around that time, such as Alphaman, a 1995 game set in the post-apocalyptic future when Donald Trump became the president. The Roguelike FAQ also includes Crossfire which probably would not be called a roguelike now, although I guess that the definition is loosened for multiplayer games.

Diablo (1997)

In 1997, Diablo was released. If you compare Diablo to Angband, the two games are obviously very similar. In fact, the developers originally wanted to create a turn-based roguelike; however, Real-Time Strategy was a hot new genre back then, and they have decided to make it a roguelike/RTS hybrid. Diablo featured great graphics. I do not know the exact reasons why the developers have decided to drop permadeath, but this was common in commercial roguelike-inspired games back then.

Reviewers did recognize it as a great game, and that it required a new classification — the light story, focus on combat, and randomness made it quite different from typical RPGs. If they recognized that these traits were taken directly from roguelikes, “action roguelike” would probably mean a game similar to Diablo nowadays; however, they did not, so the new genre was called “action RPG”.

More Roguelikes (1998–2001)

Somewhere around 1998 one of my school friends has found an Angband variant called GSNband. He has shared it with our group of friends, and quite a lot of us loved it too! We also started creating our own roguelike. Reading about GSNband, I have found that it is classified as a “roguelike”, and learned about the existence of other roguelikes. By then, a new major roguelike has emerged, named ADOM. It looked interesting, but its “simple character graphics” initially put it off. Eventually, I tried, and it was definitely worth it! I played it like an RPG as first (by copying the save files), but eventually switched to the correct, permadeath way.

Some roguelike definitions from around 1999–2001 could be found on the ADOM website and on Petri Kuittinen’s site.

On Darren Hebden’s site, a list of over 100 roguelikes could be found. There are some omissions that I consider very interesting:

Beneath Apple Manor, a 1978 game. It came out before Rogue, but it had all the roguelike gameplay elements, invented independently.
Castle of the Winds, a shareware game published in 1993. It was quite popular. if somebody remembered they played some roguelike in the 90’s, the answer was usually Castle of the Winds.
Mission Thunderbolt, developed in 1986 on mainframe computers, and released commercially for Windows and Mac in 1992. I have played it around 2010 and it was a really great roguelike, with quite good graphics for its time, great innovative ideas, and optional permadeath. It is sad how unknown this game is.

Soon I have started participating a bit in the roguelike communities, for example here I have created a list of features which are typical to roguelikes, but are rare in other games. It was noticed that my list was slanted towards “hacklikes”; I am not a fan of Nethack itself, but a big fan of ADOM and Valhalla. People recognized that roguelikes are difficult to define. Some games started to appear which were a bit difficult to classify: Decker is a cyberpunk game that I did not consider a roguelike because it had completely different tactics.

DRL (2002)

In 2002 DRL (“Doom the Roguelike”) was originally released. It was quite clear from the name what it was: it was a roguelike (character display, turn-based, randomly generated) where your primary weapons were guns, and you fought demons from Doom.

RogueBasin (2005)

RogueBasin is a wiki initially created by Santiago Zapata. Created in 2005, it started collecting the knowledge related to roguelikes. RogueBasin is still active.

7DRL (2005)

The first 7DRL challenge has been held in 2005. Roguelikes at that time were often extremely complex games which took years to produce; the 7DRL challenge instead focused on games created in seven days. Many of these submission were very experimental in nature. 7DRL challenge is still active today, with hundreds of roguelikes (and similar games) created every year. In my impression, the 7DRL challenge really helped to extract the specific things which made a game a roguelike. While major roguelikes so far were extremely complex games, 7DRLs did show that you could have a simple game which still had this elusive roguelike feeling.

Dwarf Fortress (2006)

Originally released in 2006, Dwarf Fortress has two modes: the adventure mode and the fortress mode. The adventure mode was a typical roguelike, while the fortress mode shared some features, such as high complexity, random generation and ASCII display. Dwarf Fortress (and a bit of ADOM) later became a big inspiration for Minecraft, one of the most popular games today, and itself an inspiration to many other games such as Terraria.

Temple of the Roguelike (2007)

So far, the roguelike discussions were mainly held in Usenet, in the rec.games.roguelike.newsgroups. Temple of the Roguelike has been established by Santiago Zapata around May 2007, and its web forums quickly gained popularity, while the newsgroups gradually declined. I did not like the Usenet newsgroups too much, and became more active in the Temple forums. I have been also active in the IVAN forums (Iter Vehemens ad Necem is a graphical and very fun hacklike), where I started my own development contributions, with IVAN3D.

IRDC and Berlin Interpretation (2009)

In 2009 the roguelike developers have decided to meet in person, in the first International Roguelike Development Conference. They have decided that they need to actually define what the conference was actually about, and thus the Berlin Interpretation was born. In my opinion, also shared by other prominent roguelike people, they mostly failed, for two reasons:

(1) they only knew the games existing at that time, and some features were given too much or not enough weight (just like the original 1993 definition which cared about being free, portable and character display). Soon after, new innovations were created, bringing lots of confusion.

(2) the feeling of being a roguelike is quite elusive, it is more like I know it when I see it rather than some specific list of features.

Spelunky (2009)

Later in 2009, Spelunky was released. Spelunky is a platformer which took the compatible elements from roguelikes: random generation, as well as some other minor references, such as the non-modal shops from hacklikes. Moreover, the original version was free. While Derek Yu did not call his game a roguelike, he made it clear that the game takes a lot of inspiration of roguelikes. In turn, Spelunky has inspired lots of other games. Unfortunately, most game journalists and players did not know what a “roguelike” was, and they started referring to games inspired by Spelunky as “roguelikes”. If I recall correctly, we had discussions at the time whether Spelunky should be counted as a roguelike or not; ultimately, it turned out that most people who have actually played the original roguelikes have felt that these games do not capture that elusive roguelike feeling, and called such games “roguelike-likes”, or later “roguelites”.

Desktop Dungeons (2011)

Another interesting departure from the traditional roguelike definition. Similar to Decker, it had essentially all the basic properties from roguelike definitions (like the Berlin Interpretation), but I did not consider it a roguelike because tactically it was completely different. Still more similar to traditional roguelikes than games such as Spelunky.

IRLDb (2011)

In 2008–2011 I have been working on IRLDb, a database which took the data from RogueBasin and displayed it in a tabular way. It was also integrated with the RogueTemple forums, allowing the members to rate roguelikes, share their successes, and write reviews.

Around 2011 the roguelike community has started shifting from the web forums (such as RogueTemple) to the roguelike subeddit. While the RogueBasin is still updated, the IRLDb ratings mostly show the state around 2011. Most roguelike fans have tried Angband and NetHack, but they are not considered to be that good by modern roguelike players. They prefer newer major roguelikes, such as ADOM (very popular and rated quite highly) or DCSS (even more popular and more highly rated). Spelunky is also rated very highly. Cogmind and Caves of Qud have already existed by then, but they are (almost) not rated yet. They are extremely popular today.

Faster Than Light (2012)

Another innovative game which takes lots of inspiration from roguelikes. While a real-time game, it lets the player pause at any time, which makes it appealing to the part of the roguelike community who do not like action games. Another difference is that, while traditional roguelikes are non-modal (combat and exploration are the same thing), in FTL you explore the galaxy and get “events”, which may affect your run in some way, or pull you into a combat, which is a separate mode. It calls itself a “spaceship simulation roguelike-like”.

Rogue Legacy (2013)

One of the platformers inspired by Spelunky. It described itself as a “Rogue-lite. Your character dies, but with each passing your lineage grows and becomes stronger.” Today, there are lots of people who consider themselves experts on roguelikes, and consider this particular thing to be what separates roguelikes from roguelites. For them, roguelike is a run-based game where the strength of your characters remains roughly constant in subsequent runs, while a roguelite is a run-based game where your characters get stronger and stronger. This feature is also caled “meta-progression” (progression which happens between runs, as opposed to inside a run).

It does not make too much sense from the point of view of traditional roguelike communities. Most roguelike players seem to see the lack of meta-progression as an advantage — winning DCSS for the first time after playing it for half a year was a great feeling, that could not be achieved in a meta-progression game, where it is never sure whether you win because you have mastered the game, or because the game let you. However, NetHack (and other hacklikes) had the “bones” feature, where you could find the remains of a previous adventurer, with all their equiment, which also could make the further games easier. Sproggiwood, ToME, or One Way Heroics have meta-progression, and they still do feel like roguelikes to us.

Risk of Rain (2013)

An action platformer. As usual with games from that era, it does not call itself a roguelike, but rather a game “with roguelike elements” or “rogue-lite”. Despite of that, players call it a “classic roguelike” nowadays. While these players give “procedural generation” and “permadeath” as the requirements for this classification, Risk of Rain is not really procedurally generated: the level layout is mostly the same in every run (actually, there is one of two choices).

Darkest Dungeon (2016)

According to the Steam store page, “Darkest Dungeon is a challenging gothic roguelike turn-based RPG about the psychological stresses of adventuring”. While previously the developers of games inspired by roguelikes clearly said that their games are not roguelikes themselves (only inspired by them), it appears that the more recent games just call themselves roguelikes.

I have not played Darkest Dungeon myself, but I have seen its roguelike, or even roguelite, nature challenged because it does not really have permadeath in roguelike sense. Your characters die, but they are more of expendable resources, your run does not end, you will win your first run. Furthermore, according to the complaints, the game is very repetitive, which the randomness of roguelikes aims to avoid.

Slay The Spire (2017–2019)

Slay The Spire has popularized the genre known as “roguelike deckbuilder” games. It combines the ideas of deckbuilders such as Dominion, and the general event-based structure of FTL.

From my point of view, the name “roguelike deckbuilder” does not make much sense. The event-based structure is an innovation from FTL, which goes against the traditional roguelike style of being non-modal. The combat system feels more like a jRPG; when playing jRPGs, I felt that the combat system (and its modality) was the major thing which separated them from roguelikes. Slay the Spire is run-based, but this is not an innovation, as all deckbuilders are! At least, every game of Dominion you play will be completely different. Something like “dungeon crawl deckbuilder” (Dominion is not a dungeon crawl) would make more sense.

The War is Over? (2020)

A look at a thread of roguelike recommendations shows what people consider to be “roguelikes” today. Most people recommended run-based games which are not roguelikes in the traditional sense, some did mention traditional roguelikes. Almost no one recommended games of both kinds. Some roguelike celebrities say that the war is over, we should accept that “roguelike” means something different now.

While it is hard to fight the language change, I will exhibit some problems with the popular conventions.

For some people, any game with permadeath and procedural generation is a “roguelike”. But it is not clear to me why these people even need this term. It is not a cool word: it was originally chosen as “the least of all available evils”. The main thing that all games called “roguelike” share is that they are run-based: the game is designed so that you play it many times. Randomness is a good way to make this interesting, but it is not necessary —in Risk of Rain, randomness is limited, and it is still fun. For some time, we had a “doom clone” genre, which was eventually renamed to the more descriptive “first-person shooter”, and I think we should similarly refer to run-based games as “run-based”. This would include games like Tetris, but it does not seem to be a problem, because these games are fun for the same reason as roguelikes are. Roguelike-like and roguelite did not catch, because they are too vague, and too similar to “roguelike”. Especially that there are like five significantly different meanings of “roguelike” commonly used, so why not just say exactly what you mean if you can.
On the other hand, we need a word for a “roguelike” in the traditional meaning. There are people who want to play a game in this particular style, and game developers who create such games and want to quickly tell what it is. When “Doom the Roguelike” was released, the name made it perfectly clear what it was (as long as you have played any roguelike and Doom). HyperRogue is a roguelike taking place in non-Euclidean geometry; since roguelikes share their classic chess-like gameplay, they also share some basic strategies, and the change of geometry brings a new twist on them. It is annoying that such things would not be clear anymore. The original word was already bad, the least of all available evils, and run-based people have stolen it, even though they did not need it.
Also, it seems that being run-based, while important, is not the most important quality of traditional roguelikes — commercial traditional roguelikes usually dropped it, others are still fun if you were cheating permadeath.
It is not clear whether so-called “roguelikes” are actually a genre. Slay the Spire, Darkest Dungeon, Spelunky, FTL are very different games. Why should they be a genre while, say, “games where you can pet a dog” are not?
It is suggested that “Traditional roguelike” should be used for the original meaning. There are some issues with this idea. Some people started calling Spelunky a “traditional roguelike” even before this term was coined. Furthermore, “traditional” suggests lack of innovation; I would not call HyperRogue, or mostly any other puzzle roguelike, “traditional”. Some people do not consider puzzle roguelikes to be roguelikes; I am fine with this, but it appears that the majority of the community does. The alternative for “traditional roguelike” is “action roguelike”, which would not include turn-based games which are not traditional roguelikes, and also does not include Diablo for some reason, even though it is more similar to the tradition.
It seems that most roguelike fans accept using “roguelike“ as an adjective, for example “roguelike platformer” or “roguelike first-person shooter”. While slightly confusing, this makes sense: a game is a combination of two genres, it has all the properties of the genre appearing as a noun, and the compatible properties of the genre appearing as an adjective. We can also have “first-person shooter roguelike” (DRL) or “platformer roguelike” (a grid-based turn-based game where the structure of the level is similar to that of a platformer, e.g. Fuel or Bump). Although some people do this incorrectly (e.g. “deckbuilding roguelike” is wrong, “roguelike deckbuilder” is better).
It appears that the people who argue that roguelike means something different now have never (or almost never) played a roguelike in traditional sense. They think it is a tiny thing, or its rules are too specific to be a genre (there is about a thousand of games satisfying the criteria, and there are several subgenres, like hacklikes, puzzle roguelikes and broughlikes), very niche one (I know lots of people who play them — they are just not represented in more mainstream game media; while the part of the gaming audience who pays the most money would not be likely to enjoy them, you don’t need to be some kind of ultra-nerd either), less influential than classics such as Spelunky (roguelikes inspired three major genres, Spelunky inspired just one). There is no rational reason not to try a roguelike, especially if you want to be an expert on game design! Roguelikes would not inspire that many great games if they were not great games themselves. The best ones are free, they respect the player’s time, and players who try them are often amazed. DRL and DCSS are quite approachable. Since roguelikes are not based on state-of-the-art graphics, older roguelikes are still great, although the design of NetHack or Angband may feel too old today.
It is argued that traditional roguelike fans do not recognize that the language changes. We do: the 90’s definition, Berlin Interpretation and the current meaning are very different. If you went to the roguelike subeddit and said “Haque is not a roguelike because it is not free and has no character display”, you would probably be downvoted, just like if you said that Binding of Isaac is a roguelike.
On the other hand, while it is good that the current definitions does not include things like “free” and “character display”, these things are an important part of the roguelike culture. Roguelikes were not made for money, they were often made because the dev was not satisfied by any games “on the market”, and created a game for themselves; no graphics were made, because they were irrelevant. If your tastes were similar, you would also find these games much better than anything on the market. While there is nothing wrong with wanting to be paid for your work, this approach has made things possible that were not possible commercially: traditional roguelikes were extremely complex games, which would not be possible if the devs had to create graphics for everything. While the original Spelunky is also free, almost no further run-based games are. It is sad to see the term originally created for free games evolve into an essentially meaningless marketing term.