What is your current favorite Roguelike? [Or all things roguelike]


Can you sell me on it? I’ve been procrastinating playing Zangband because I wanted to read Zelazny’s books at some point, but maybe I should give up on that dream. I’ve never played a -band because they sounded sorta grindy. Is that a fair characterization?


I’m reading the Spelunky book by Derek Yu (it just came out last night) and I read a passage early on when he’s talking about the roguelike aspect of the game I felt may be relevant to the debate about what makes a rogue-like game.

[QUOTE=Derek Yu;Spelunky]At the 2008 Roguelike Developers Conference, a group of roguelike developers attempted to clarify the definition of the genre by creating a list of high and low value factors called the “Berlin Interpretation.” The Berlin Interpretation’s high value factors define a roguelike as a turn-based, grid-based, dungeon crawl that features randomized levels and permanent death. The less important low value factors include ASCII graphics, one player character as opposed to a party system, and monsters that share rules and behaviors with players. When I was working on Spelunky, I focused on just three attributes of roguelikes that to me held the essence of the genre:

  1. Randomized level generation.
  2. Permanent death (also known as “permadeath”), whereby the player has one life and cannot reload their game to take back mistakes.
  3. A ruleset for physical interactions that is shared by the player, non-player characters (NPCs), and items.

The third attribute is my version of “monsters share rules and behaviors with players” and it extends not only to monsters but to items as well. In many games, monsters and items are coded separately and handled separately. You might be able to pick up an item and kick a monster, but you couldn’t kick an item and pick up a monster. In a roguelike, you should be able to pick up and kick items and monsters, with results that are based on consistent rules. This suggests that the player, the monsters, and the items are treated more like variations on a single type of object than three completely different objects.

Yu, Derek (2016-03-29). Spelunky (Kindle Locations 77-84). Boss Fight Books. Kindle Edition. [/QUOTE]

I thought this was pretty cool. I also looked for more on this “Berlin Interpretation” and found this handy link - http://www.roguebasin.com/index.php?title=Berlin_Interpretation



I was referencing the Berlin interpretation for my description earlier, but should have made that explicit. The Berlin Interpretation is a rough guide for capturing the essence of classic, strict, or pure roguelikes.

This further complicates the discussion but outside of game mechanics there is also a philosophical position that many in the core roguelike community have an affinity with. That is that pure roguelikes are open source, frequently free (as in beer), and non-commercial. In recent years roguelikes have become more commercialized (which I support) but developers usually offer a free and open source code branch (several versions behind) alongside their purchasable game. Just another wrinkle in the discussion.



Yes, I was also referencing the Berlin interpretation as being overly hidebound and myopic, focusing too much on the incidental details borrowed from RPGs. I like Derek Yu’s idea of a universal system of physical interactions, and it is great in Spelunky and some other games, but there are enough counterexamples that I’m not totally convinced it needs to be the third pillar of the genre. I would think that it is one effective way, though not the only one, of designing interlocking systems that interesting situations can emerge out of, which is the attribute that’s actually critical when you’re going to be playing through reshuffled mixes of the same basic elements. Definitely want to read that book though.

Yes, that’s a very fair characterization. At this point I’d have a hard time recommending a *band out of anything other than historical curiosity. Their fatal flaw is that you can generate new dungeon levels ad infinitum by going up and down stairs. This means that you can hang out at a relatively safe depth, building up treasure, experience, stat boosts, etc., with no mechanic pushing you forward. And once you CAN do that, if you choose not to because it’s boring, then when you get in over your head and die, you know that you probably would have survived if you had just put in more time grinding.


Fair enough but I find it a good starting point and not the end of the conversation. :-)

It is actually a relatively flexible interpretation since it lists a bunch of “high-value” and “low-value” factors and goes on to say:

This list can be used to determine how roguelike a game is. Missing some points does not mean the game is not a roguelike. Likewise, possessing some points does not mean the game is a roguelike.
The purpose of the definition is for the roguelike community to better understand what the community is studying. It is not to place constraints on developers or games.

So it isn’t as dogmatic or restrictive as it first appears. It is just a handy yardstick.

While Spelunky shares a lot of roguelike qualities I find that the real-time nature of it creates a different circle in the Venn diagram. Reaction- and reflex-based rogueish games are their own beast since the emphasis is on player’s quick and instinctive reactions and not based on careful thought and analysis of the situation (though some of that is required too). Real-time action-based rogue games live comfortably alongside turn-based roguelikes but demand a whole different player approach and skillset.

Maybe what I am trying to say is that if I value one rogue-like design factor over all others it is the turn-based requirement.

I love Spelunky and games like that but I find myself getting more frustrated at real-time ones because of stupid things like not reacting fast enough or timing a jump poorly. I barely have time to think or consider my situation. And all of that is fine and I am ecstatic that they exist. They offer a newer evolutionary branch in the history of rogue-* design.



But that’s entirely customizable behavior, and has been for ages. One can run with “preserved” levels, or tweak level gen parameters in a variety of ways, to taste. “infinite level regen” wasn’t really a core property of *bands. Although this conversation [I]was[/I] a fun way to stir up flame wars in the usenet group for rogue likes back in the day.

porousnapkin, it’s next to impossible to pick up anything meaningful about the Chronicles of Amber from Zangband. It’s mostly just light flavor. All the baddies from Angband (Tolkien), Cthulhuband (you know), and IForgetBand (uh, stuff) were basically “displaced” downward on the “holy shit this is scary scale” by various characters from the Amber books. Well, Cthulhu was probably roughly as difficult as the friends you make on 99 and 100. Also, Corwin could show up as a friendly and he wrecked everything. Which was not inappropriate thematically.

Zangband’s “what makes it, it” comes from:

  1. Large overworld with many towns. Again, you can restrict it to 1 town or even 0 with all the world gen options if memory servces. But towns make for a different sort of RL experience and this was something of a core feature (Angband had a single town, but again a lot of people turned it off). Certain things that are precious in a “dungeon” game become more readily available with towns (especially many towns. . . e.g. various scrolls).

  2. If one stuck to the default towns, there’s lots of quests they give. Some of these are horrid, horrid death traps of death (“investigate the island that appeared off shore south of here” yeah no don’t ever do that). Some were easy. they sometimes had rewards that were critical for dungeon progression. E.g. a ring of see invisible. Which you could find anyway, but it was simply a “guaranteed” source. Which is not as big a deal with all the scrolls and shops available. Some of the resistance ring rewards in later quests were harder to come by, granted. But not impossible by any stretch.

  3. Pretty much having a zany mix of things from a variety of other *band variants. Kangband’s guild quests (where the overworld questing mechanics came from IIRC), Oangband’s massive weapon damage overhaul (before it, daggers were always the best weapons in *bands because speed was king and weapon damage dice were a tiny drop in your damage pool at high levels) as well as several other mechanics therein, many others I am forgetting. About the only “noteworthy” as of 15+ years ago flavors it didn’t borrow from were Pernband (which was more heavily Tolkien with some Adom influences; it’s what became Pernband became TOME) and Eyeband, which was one guys variant that had some interesting flavor (a crazy alchemy system that could leave you with artifact creation potions if you got REALLY REALLY lucky early. Or blown up. Or anything in between). Zangband took from a lot of other stuff in the family. Oh, Cthulhuband went off on it’s own and developed a very deep but very frustrating learn-by-doing skill system that got ignored by everybody else. It was probably a little too ambitious. And it’s frustrating to discover you can’t raise skills [I]at all[/I] without very aggressive diving (which was extra dangerous in Cthulhuband).

  4. Expansive, even for a rogue like, game options.

  5. This may not really be a *Band thing but I always saw it as one: part of the end game involves farming unique bad guys on level 99. Because the level 100 bad guy can summon alive uniques, and you just didn’t need any additional surprises at that point (even from wimpier uniques). Even with fixed levels, IIRC the mechanics were such that you could spawn any unique around the next corner as soon as you lose line of site on it, assuming they had not already been killed.

Zangband’s difficulty easily accounted for the fact that you could have ready access to all the uncursing, enchanting, teleport, blink, level change, satiation scrolls and healing/stat boosting potions you wanted. In effect, you were expected to be at all stat maximums and stuffed to the gills with all sorts of usables, and have all resists covered, by the time you passed '2500/50. It was a brutally difficult game. In fact the bottom 50 floors can be rightfully termed a slog (I mean, they’re not all a slog but many could be). It was an interesting game in it’s time, but the genre has gone in other directions and this “type” of rogue like (such as it was) seems dormant.


Sure, I agree that they left themselves some wiggle room in application, but I don’t think that makes the definition any more useful, especially as in practice it usually gets used to shut down discussion with “that’s not a [I]real[/I] roguelike”. All it measures is how similar to the canon of classic roguelikes a game is. That’s not a bad thing, but it is a missed opportunity to achieve their stated goal to “better understand what the community is studying” because the critical elements of procedural generation and permafailure are reduced to just two factors among a bunch of others. Those two things are the core of what makes the genre special, and lumping in a bunch of surface/implementation details borrowed from RPGs doesn’t help understand them, and in fact tends to exclude a bunch of games doing interesting things to explore them.

Are you familiar with the Brandish series? They are turn-based, grid-based, mostly non-modal, complex hack & slash games with significant emphasis on resource management and exploration and discovery, in which a single character with numerical stats explores dungeons, battling tactically challenging monsters. But they have a fixed dungeon, and you can save your game. In other words, they hit 7/9 high-value factors and 4/6 low-value factors of the Berlin interpretation. Or imagine a version of Rogue that always used the same random seed for the dungeon, and let you save and reload your progress at will. Are they roguelikes? I would argue no – that procedural generation and permafailure are absolutely necessary and their absence overrides every single other factor.

That’s not to say that they are sufficient criteria, though, as that would allow something like Minesweeper or Flappy Bird. I don’t have a handy encapsulation of what else is minimally required, but I’d say that the Resource Management and Complexity features from the Berlin interpretation, or Derek Yu’s idea of a universal system of physical interactions are in the right ballpark.

In an alternate universe where we had some commonly accepted term for procedural generation + permafailure besides “roguelike”, I’d be fine with using that broad term and leaving “roguelike” exclusively to the subset of what I would now call classical or traditional roguelikes. But that ship has sailed and usage has evolved. Any attempted definition that would tend to exclude Spelunky, Isaac, FTL, Desktop Dungeons, Galak-Z, etc. is counterproductive.


Interesting. I don’t know if I ever tried that, as it’s been a while since I seriously tried any of those, and I don’t think I did much in the way of trying to tweak individual parameters.

But what does the process of getting all those stats/consumables/resists look like, if you’re not re-generating levels? Do you just find them all in the course of a single normal progression down through the dungeon?


Well, the GUI was a bit involved but it was there. I think for big time *band lovers that level of customization was a big deal. For anyone who just picked it up at random? It was more customization than normal and it was tucked away. I don’t even think you could access all of it during character creation. But it was there.

It was charming - if completely suicidal - to do things like spawn all small open levels (“arena” levels I think they were called). You’d never do an entire playthrough that way but it was cool you had the option. You could take certain types of levels out that way as well.

But what does the process of getting all those stats/consumables/resists look like, if you’re not re-generating levels? Do you just find them all in the course of a single normal progression down through the dungeon?

Depended on how vigorously you shop scummed the towns. ;)

More specifically, potion farming at ~28 or so was to make the next 20-30 levels easier, but it was not mandatory. By the time you hit 70-80 - I never did it without wizard mode - you’ve probably seen all the stat potions you could ever want a couple of times over. Or you were a few tenths away from a max stat that wasn’t really that important. You also had protection from draining (or else you’d be dead). There was a crazy difficulty spike some time in the 40s. I don’t recall offhand what it was. I think rotting Q’s was one of the reasons. They start summoning stupidly powerful undead. Also some of the lovecraftian stuff starts to show up that’s just… . yikes. I mean there’s a swarm polyp thingy that’s like everything just shy of ethereal (don’t recall why but see invisible doesn’t work on it either), breeds 10x faster than anything else you’ve seen, and is nearly impossible to deal with (I tried with wizard enhanced characters, just absurd). And it hurts worse than any swarm you’ve dealt with since you were level 1. By many orders of magnitude. You catch that + anything that requires your attention and it’s “fuck teleport out and find a stair ASAP”.

Stat Potion spawn chance appears at maybe floor 20 but peaks at 28 but stays at that point until like 60. All other consumables stay about the same in terms of spawn I think (they’re unusual in that respect). Except the BETTER variants (like Identifiy, Enchant Weapon, Uncurse) [I]start [/I]spawning past the early floors and don’t spawn frequently (but the chance might rise). Playing Zangband sans towns? Yep, it’s just harder. Like wise, no level regen? Harder. I don’t think it was impossible (hell, this one guy used to love playing halfling mages that just threw spell books. And he got one to like 91 or something). But practically speaking it probably is impossible for the majority of genre fans, let alone other gamers.

So mind you I do think the towns and the like are part of the *band experience. But so was customization, and a lot of people were fervent about playing “classically” without silly towns or level regen. Like I said, I think it’s sort of a dead-end in the family tree and there are valid reasons for that. I mean, if nothing else just having 100 “dungeon levels” isn’t as exciting as “this is a themed segment of the main dungeon. That’s a themed side branch, optional” or “there are 15 different themed dungeons to explore, to varying degrees, based on [factors]”. But it’s still fun to look back on. The Z was for “zany!”.


But has that ship sailed altogether?

It seems widely accepted that [I]rogue[/I] = procedural generation + permafailure
[I]like[/I]=turn based
[I]lite[/I]=action or real time

Though to be fair I am mostly speaking from a player preference standpoint. I think Spelunky and TOME should be discussed as peers and equals and this Roguelike thread should be filled with all sorts of real-time and action based rogue-ish games. I know that some people use [I]roguelike[/I] as a term of exclusion or discussion-ender, as you pointed out, but that is not my intention.

However, I do feel that a distinction should be made between turn-based implementations and real-time ones and I suspect that a lot of people share that preference. The distinction is important since they require a different mindset and skillset and I like to know beforehand what the game will demand of me. Will I be able to take my time and sip a drink while I play or am I playing something that has the same attention and skill demands of Mario?

So intellectually speaking or abstract-wise Spelunky, Nethack, Angband, Rogue Legacy, and Unreal World all belong in the same conversation, but labeling for a player perspective distinguishing between the how time is controlled matters quite a bit to a large number of the community. In the same way that the RPG community or strategy community go to great pains to distinguish that difference as well.



That doesn’t quite match up with the usage I’ve seen, though. I’ve generally thought of “roguelite” as implying a de-fanging of permadeath by adding more persistence between runs, not specifically about action elements. I’d call modern Mystery Dungeon games roguelites, though they predate the term. This articlecalls Dungeons of Dredmor a roguelite, though that doesn’t fit by either of our definitions; maybe intended to mean “simpler and more accessible than true roguelikes”.

Glancing at a random sample of game descriptions on Steam, I see a lot more “roguelike ______”, “action roguelike”, “roguelike-like”, “roguelike elements”, and “rogue-inspired” than “roguelite”. The only game I noticed to really embrace the term is Rogue Legacy, which has this bullet point implying that it’s about persistence:

Rogue-lite. Your character dies, but with each passing your lineage grows and becomes stronger.

I agree that a distinction between turn-based and real-time is useful. But given the way usage has evolved, I just don’t think you can rely on “roguelike” to draw that distinction for you without other words attached to it (i.e. “classical roguelike” or “tactics roguelike” vs. “platformer roguelike” or “action roguelike”).

And of course, you also have games that embrace procedural generation + permafailure + turn-based play, but go pretty far from the classical approach in other ways (Desktop Dungeons, Out There, Massive Chalice, Battle Brothers). I’d consider these to be their own weird and wonderful subgenres within the roguelike umbrella.

Your example of other genres handling this sort of taxonomy question is on point. I distinctly remember comments in CGW and PC Gamer that Dune 2 or Warcraft weren’t REAL strategy games because they didn’t give you time to think. Eventually people gave up and started specifying “turn-based” or “real-time” strategy, and the issue was solved (except for the confused people who call Civ an RTS because they don’t know what the letters stand for). Or how within RPGs, the dungeon crawler subgenre has both real-time (Dungeon Keeper, Legend of Grimrock) and turn-based (Might & Magic, Wizardry) branches, but they’re all still dungeon crawlers.


Fair enough. :-)

Just wanted to say that I really appreciated your comments and perspective Thraeg!



Likewise! This was a fun conversation!


I get so happy every time I see a civilized debate where both people see each other points and come out of it having food for thought, and grateful for the opportunity. It gives me hope in mankind.

So thank you both. ;)


For what it’s worth, I apologise to the thread for my role in helping kick off this well worn topic… and now again for replying. From memory, I think there’s even two threads on QT3 already dedicated to this very topic? Infact it looks like you guys have agreed to stop, but I can’t help myself.

Who’s to say that the “critical elements of procedural generation and permafailure” are the two most important ones? Why shouldn’t they simply be two amongst many others? Because you like them when they’re present in other games? :)

Permafailure was the de-facto game mode when Rogue was released (Citation: Every arcade game ever), so in a way it wasn’t really that novel, so I don’t see why it should be the pillar of what people consider to be Rogue. To me, Rogue is square, black and white dungeons and continually starving to death. But these days I’m alone in that regard!

In an alternate universe where we had some commonly accepted term for procedural generation + permafailure besides “roguelike”, I’d be fine with using that broad term and leaving “roguelike” exclusively to the subset of what I would now call classical or traditional roguelikes.

There was a push a while back for something silly like “procedural death labyrinth”. I don’t really see what’s wrong with people saying “this game has permadeth and procedurally generated content”, and instead relying on the term “roguelike”. I guess it’s shorter?

But that ship has sailed and usage has evolved. Any attempted definition that would tend to exclude Spelunky, Isaac, FTL, Desktop Dungeons, Galak-Z, etc. is counterproductive.

I question the need to even group together the stated games, as they’re so wildly different. Things get even more wild and different if we include that 80s black-and-white smash hit, Rogue. At this point we should simply stick to the most common term that accurately describes all of the listed games: Games.

Which brings me back to this part of your post:

I’m not convinced that the community that plays Rogue, Zangband, Nethack, DCSS – the same community that spends hours messing with libtcod and entering 7DRL competitions – is the same one playing Isaac and Galak-Z. If it is, it’s because those people are gamers, and by the same measure they’re playing Call of Duty.

Even in this very thread you see people playing one sort of game and poo-pooing the other. Arguably, that’s what this entire discussion is about :P

I’m not convinced those 5 games do even belong in the same conversation. I can’t see what useful conversation there is there, other than “not having lives or being able to save is sometimes a fun way to play a game”. Except even Rogue Legacy doesn’t fit in there because in effect you are saving?

For me, that’s an interesting example. I used to consider Dungeon Master, Eye of the Beholder, Might & Magic, Dungeon Hack, Menzoberranzan, Ishar, Dragon Wars, and Nethack to be the exact same “sort” of game – just different riffs on the same thing. Even though some would now be considered Roguelikes and others wouldn’t? I guess they’re all Dungeon Crawls of some sort?

But then I guess I played enough RPGs that M&M kind of got split out from those… and now I’m not sure how I’d group them. Dungeon Hack is basically first-person Rogue, but it looks practically identical to Eye of The Beholder… Thinking about it, way back when, I might have even considered the goldbox games to be lumped into this series, mainly based on the fact that they were all Forgotten Realms games :P

I didn’t learn about the term Roguelike until 2005 or so. Back then I at least could see what it meant. “Oh, these games are all like that game Rogue!”. And yet now the term is so warped I have no idea what someone means when they say it. I feel like we should just abandon the term, but doing so what upset the Roguelike community. (And by Roguelike community I mean the ones on Roguebasin, not the ones competing Spelunky in ever-faster times and crazier ways)


I’m not convinced those 5 games do even belong in the same conversation. I can’t see what useful conversation there is there, other than “not having lives or being able to save is sometimes a fun way to play a game”. Except even Rogue Legacy doesn’t fit in there because in effect you are saving?

Well this thread is an example of a useful conversation. It is a Roguelike thread but people post about modern non-classic or non-traditional games with rogueish elements all the time and the conversation is better for it. I’m all for trying to keep the purity of the roguelike term but this thread has introduced me to titles like the excellent Vagante which is more of a Spelunky-like but it is fun to see what aspects of classic roguelikes it draws inspiration from.

I think it is great that we have this contemporary resurgence of games drawing ideas from the rich pool of roguelikes. This new crop of roguelites are an evolutionary branch off of the main roguelike concept and enrich the conversation and may even feed back ideas into the traditional roguelike branch of game design. Yes many of the denizens of Roguebasin get grumpy when someone brings up Spelunky but I enjoy a dialogue that is more inclusive and broad. If this thread only allowed posting about pure classic roguelikes we wouldn’t see much activity except for the occasional TOME expansion, DCSS update, or Steam release of Unreal World.



Just when I thought I was out…


Ahem. I obviously have no inherent authority beyond however persuasive you consider my arguments (and definitely much less than the developers behind the Berlin interpretation). But I do think my approach captures both the essence of the genre and the way that modern usage has evolved, and thus makes a more useful communication tool than the hardline traditionalist stance. I think it’s more useful to consider it as a fundamental decision about the structure of the game than the about the individual mechanics found in it.

As for why those two are so important, go back to my example of Brandish or a hypothetical modified version of Rogue that removes them. There you would still have “square, black and white dungeons and continually starving to death”. Is it a roguelike? If you agree that the answer is no, then we have established that they are necessary conditions – that without them, a game categorically cannot be a roguelike regardless how closely its other attributes stick to genre conventions.

There may be OTHER necessary conditions to avoid casting too wide a definition net (and in fact there certainly are, if only to exclude games that no one is arguing are roguelikes such as Tetris or Temple Run). I lean toward some mix of complexity, resource management, and maybe a concrete win condition. But if you hang your hat on most of the other factors in the Berlin definition, then you wind up disavowing most of the recent explosion of games that cite Rogue as a chief inspiration. That’s not an irrational position, but it impedes communication by staking out a definition that’s incompatible with the way a large number of people use the term.

And you’re right that permafailure wasn’t a novel innovation of Rogue, but combining it with procedural generation was, because that created the structure where each subsequent attempt had to involve consideration and reaction to new situations rather than rote memorization of obstacles and the ways to overcome them.

Yeah, as I mentioned, if some other term had caught on it would be fine, but that didn’t happen. And yes, like any genre descriptor, “roguelike” concisely gets across a package of related concepts.

Those games (including Rogue but excluding CoD), share a fundamental structure where:

[li]Play consists of some number of “runs”, each of which can end in success or failure.[/li][li]Regardless of previous results, each run starts mostly fresh. Meaningful amounts of progress toward the goal are frequently lost.[/li][li]Each run has a complex and novel arrangement of game elements, from which gameplay decisions emerge systemically. [/li][li]The player only gets one attempt at any given arrangement, and cannot try again or erase the consequences of mistakes.[/li][li]Other game design decisions are made with the above in mind.[/li][/ul]

You say that dismissively, but as a fundamental design principle it stands in sharp contrast to most of the game industry over the last 30 years, which has emphasized constant progress and is afraid to take anything away from the player or force them to live with consequences of mistakes.

And I actually agree that Rogue Legacy is an edge case because it puts so much emphasis on the meta-progression over and above the results of any individual run, and quickly becomes about grinding.

I haven’t played all of those, so am not sure exactly how to classify them all. I’d call Dungeon Hack a dungeon crawl roguelike.

I really don’t think it’s warped as much as all that. Pretty much everyone using the term, whether they favor Berlin or modern interpretations, could agree on the minimal common ground I’ve outlined above. And that common ground stands in contrast to the ubiquitous game design paradigms in the industry at large (so your facile suggestion of just calling them all “games” doesn’t help). It’s a meaningful categorization within which some design lessons and principles can be shared and adapted, even in games with superficially different mechanics. More precision is always available when needed by attaching other descriptors to “roguelike”, and no specificity is lost by simply referring to the original longstanding canon and the games that follow their template as “classical roguelikes”, just as there was once a time when classical music was just called “music”.


Anyone played this [I]Vanilla Bagel[/I] thing?


Nope, but it looks great. Might bite on it during the sale.


Wait … so this is [I]not[/I] early access? Intriguing!