Back to Small Clever Rooms

Ten Thousand Lifetimes with Roguelikes


When I was a kid, a few factors combined to give me what I suspect will be a lifelong neurosis about video games: my basic respect for my parents’ authority and opinions, my mother’s extreme disdain (it seemed to me, at least, at the time) for any and all electronic gaming, and my borderline obsession with same. My brother and I weren’t allowed a game system until, after much parental deliberation, I gather, we received a Super Nintendo for the Christmas of what I think was one of my middle school years.

Unfortunately, Mom’s yielding on this issue didn’t mean her suspicion of video games had abated, and I can still remember pretty vividly the discomfort I felt whenever I was playing and she was around, radiating a high-intensity aura of disapproval that made me feel somewhat ashamed to be wasting my time like this. It wasn’t enough to stop me, of course. But it was always there.

I don’t mean to make my mom out to be some kind of harridan, much as it might seem so in the context of today. Now, video games are moving inexorably into the mainstream, and all the kids like me—kids who would happily spend six hours straight playing Final Fantasy II whenever we could get away with it—have become perfectly healthy and normal adults and sometimes even had kids of our own, many continuing all the while to play video games. Back then, I think my mom had every right to be suspicious of them, and it’s probably good that I feel a little uneasy, a little guilty even, when some new game captivates me and I spend just a little more time than I probably should playing it.

But anyway. My dad was an engineer, and as a result we had computers growing up, certainly before we got that SNES. I don’t remember computer games having quite the maternal stigma that consoles did. I was, as can be gleaned from the preceding paragraphs, kind of a nerdy-ass kid, into Tolkien more because I sensed somehow that he was a Thing than because I actually liked the books. And somehow the combination of these factors led me across this very weird game, Angband, based somehow on Tolkien: a game that surely seemed incredibly dated even back then, not even having any graphics really, just a @ moving jerkily around in a land of .s and bumping into #s, occasionally running into rs and ls and ps.

But I certainly recognized the trappings of a game, nonetheless. The character selection, for sure: elves! Dwarves! Half-orcs! Hell yeah! That canonical column of stats: Str, Int, Dex, Wis, Con, Cha. Gold, natch. There was definitely a game in here somewhere. That I, with my preadolescent levels of patience, actually put in the effort to find it may have been more a function of desperation than anything else, but it was certainly a revelation to me when it turned out to be an incredibly deep dungeon crawler: literally, in its having 100 increasingly dangerous dungeon levels to descend through before meeting the final boss, and figuratively in its D&D-esque preponderance of monsters, items and stats to keep track of and optimize.

When was this? It must have been after the era of passwords and GAME OVER, because I remember my brother and I being mystified when we died and couldn’t restart from where we saved. The game was deleting our save files! Was that supposed to happen? And did my brother and I figure out that we could copy the save files somewhere else to protect them from this weird and bloodthirsty throwback of a game, or did my dad give us a protip? In the era of save points, did this even feel like the cheating it was, or just a workaround for what must surely have been an inexplicably buggy game?

Whatever the circumstances of our discovering it, “savescumming,” as I later learned was the term for this most vile of practices, became the cornerstone of my Angband technique. It may have seemed analogous to normal save point practices at the time, but there was a very significant difference. Thanks to the large amount of randomness that Angband employed in everything from level generation to shopkeepers’ wares to damage calculation in combat, playing it ended up being something like playing a game of poker where you could bail and start the hand over for free if you didn’t like your cards: more a test of patience than any kind of skill. I’m not proud of this, but one of my most common techniques was to get a wand of polymorph (cheap and easy to find due to their being generally quite dangerous to the player) and turn a hapless and harmless townsperson into a random monster. Usually this would result in my character’s grisly death and another copy of the save file, but on the rare occasions when it turned the monster into e.g. something slower than me (though probably quite dangerous if you let it catch or corner you), I could pick it off easily if tediously and reap a preposterous bounty of experience points and treasure. Save, copy, and repeat until the wand is drained.

Best of all, because Angband didn’t look (to me) much like a “normal” video game, I felt—however inaccurately—sure that I could play it while escaping the larger part of my mom’s disapproval. I’m sure I wasn’t fooling anyone. But my dad found it genuinely cool, up to a point, that I would play a game with no graphics or sound.

My squandered time did not turn me into a master manipulator of any gameable system, unfortunately; after all, I’m fairly certain that America and capitalism are both pretty gameable, especially for someone as privileged as myself, but I’m not a multi-billionaire or a particularly powerful person. I suppose it gave me an early visceral and intuitive understanding of pseudo-random number generators, as I observed that despite Angband’s surface randomness, if I loaded up a save game and did exactly the same thing as last time, the game would react exactly the same way in response. But I think that what I most appreciate now about my hours playing Angband is that they gave me a concrete point of reference when I grew up, learned a thing or two about game design, and realized what a fucking shitty game it was.


Speaking of shitty games.

But I’m getting ahead of myself.

You can’t get as engrossed in Angband as I did and remain unaware of NetHack for long. It’s probably the most well-known example of a Roguelike, that genre name loose enough to fit just about any game with permanent death and randomly generated terrain.

It’s a somewhat better game than Angband. As an example, like almost all roguelikes, NetHack has what’s known as a “food clock”: your player starves to death if she doesn’t eat. Angband’s food clock didn’t work. You could always return to the town and buy more food that would never spoil. Why include a feature that just introduces busywork for the player? In NetHack, the most common way to wind back the food clock is to eat the corpses of slain monsters, which eventually go bad. “Perma-food” like that in Angband is typically harder to come by, forcing the player to keep killing monsters and, often, go deeper into the dungeon to do so.

But even as I became an expert at (cheating at) Angband, NetHack stymied me every time I tried to play it. Superficially it’s similar. But once you’ve mastered the basic mechanics and aesthetic, or total lack thereof, that they share, the thematic and mechanical differences rapidly become stark and baffling. Kick a sink and you might get a ring, or you might get an angry water demon and yet another abrupt and gruesome death. Drink from a fountain and you might get, I dunno, a stat boost or something, or a water nymph might show up, steal your stuff and make you chase her down. You can play a “tourist” and start with a Hawaiian shirt and credit card (useful only for picking locked doors, if I recall correctly). You start with a pet and can get it to steal from shops; if you try it yourself, the shopkeeper comes after you. If you kill a townsperson, the Keystone Kops show up and chase you with pies.

I was just a kid and had no idea what the Keystone Kops were or why they had pies or why they killed me anyway. It’s true that NetHack offers far more opportunities for deep interactivity than Angband, which is more in the “kill monster with axe; kill monster with sword; kill monster with fireball” vein. Here, monsters can use items on you. You can shout and write on the ground with wands and dip swords in potions. Sometimes, things that happen as a result make sense; sometimes they really don’t. But I grew up in the age of Internet spoilers, and NetHack couldn’t keep its secrets from me for long. I learned that you could increase your stats by “exercising” them: carrying slightly more than your weight limit, for example, exercised strength. I started to do this scrupulously, because increasing stats was a big deal in Angband. But this was small potatoes compared to the extent to which you can — and are, in many ways, expected to — game the system in NetHack.

Here’s an illustrative mechanic that exists in NetHack: wishes. I remember getting a wish at some point, perhaps by having a stroke of luck and finding a wand of wishing, perhaps by rubbing a lamp and getting a benevolent genie (the other kind probably would have spelled death for that character). NetHack threw up a prompt that said something like “What do you want to wish for?” and I thought, “What, I can just type anything?”

Well, yes and no. You can wish for most any item that you can get through other means in NetHack if you know how to ask for it. I remember at some point I wished for “Sunsword”, that being a special unique sword that I’d lucked into in some other game, and was pleased when it worked. Of course, I’d have been better off wishing for a blessed Sunsword. It would have been a better idea still to wish for a blessed rustproof Sunsword, and better yet to wish for a blessed rustproof +7 Sunsword. (+8 or up doesn’t work. Why? Because.) How do you find out about these modifiers? Get blessed and rustproof items the normal way, I guess, or read spoilers. But if I’d read more spoilers, I’d have known that what I actually should probably have wished for was a blessed rustproof +7 grey dragon scale mail so that I could start putting together what’s known as an ascension kit. According to the NetHack wiki I just linked, an ascension kit is “a set of items that are virtually required for a successful ascension.” It’s also part of what I was talking about when I made the bold claim that NetHack is a terrible game, and another quote from that wiki sums up why:

One of the main design goals of [NetHack variant] Sporkhack was to do away with the notion of a clear-cut ascension kit, instead forcing the player to make hard and interesting choices.

It seems to me that forcing the player to make hard and interesting choices is pretty much the only way a turn-based game can be successful. But NetHack’s design, with respect to something as fundamental as what you need to do to win, instead severely limits those choices and funnels you into one of relatively few equipment setups. Back when I was trying to play it in grade school, NetHack seemed like sort of a joke to me. Because of decisions like this, it still does, though now for somewhat more profound reasons than the existence of misspelled police who throw pies. I never really got into it and I don’t think I ever will, but I do enjoy reading spoilers and seeing all the minutiae that this decision is keeping me from having to attend to.


I associate ADOM (Ancient Dungeons of Mystery) with one particular summer at home during my college years, when I had access to a Windows computer, and with George Harrison’s post-Beatles solo album All Things Must Pass. ADOM was Windows-only; I always had a Mac, and so I think my brother was the one who discovered it. But I got into it and All Things Must Pass that summer in a big way: so much so that I think of one anymore, the other is invariably there too, clinging like a barnacle. This weird nostalgic synaesthesia is not really germane to the main themes of this essay, but it’s so strong a part of my ADOM-related memories that I couldn’t help but mention it.

Anyway: ADOM is in many ways an old-school roguelike, with its ASCII graphics, turn-based gameplay and high fantasy setting, but it also deviates in some significant ways. With the possible exception of some Angband variants in which I may have dabbled, ADOM was my first Roguelike with an “overworld”: a fixed (in this case) aboveground world with forests, mountains and the like that serves mostly as an obstacle that you traverse in between bouts of good old dungeon crawling. It even has a story of sorts beyond Angband’s “kill the bad guy” or NetHack’s “get the MacGuffin,” or at least there’s a sort of narrative thread running through some of the quests you’re asked to carry out. Unlike Angband, which started to feel a bit geographically monochromatic after the 30th or so dungeon level laid out using the same algorithm, ADOM has many diverse and flavorful locales. There’s a flaming tower where the heat is so intense you take damage every turn you’re inside it unless you have some fire resistance; an underwater cavern you need air to navigate; an infinitely deep dungeon, as though someone had taken Angband, removed the boss, and dropped it into the ADOM world; an ancient tomb with a piranha-infested moat; and a forge that’s home to a dwarven artificer who might be a bit crazy due to the powers of chaos slowly creeping over the land. Oh, did I mention that the entire game has a background radiation level, a.k.a. “corruption,” that will gradually turn you into a freakish mutant if you take too long? Did I mention that you can die of old age?

ADOM even sidesteps NetHack’s crazy proliferation of spoilers by the simple expedient of being closed-source, preventing smartasses from spelunking into the source code to determine that a goblin with a club will hit you for 1d3 + 2 damage or whatever. As one of the more popular roguelikes, it still can’t escape having many of its secrets laid bare in various places on the Internet, but I believe some of them took quite a while to discover. As a result, ADOM has retained a bit more mystique than the thoroughly vivisected NetHack; its world also feels far more fleshed out than Angband even though the latter is loosely based on over a thousand pages of actual published source material. Angband’s dungeon is a huge skeleton of a location; ADOM’s Drakalor Chain lives and breathes. Its cohesion may be due to being the vision of one guy, the same guy who steadfastly refused to release the source. But that refusal had other consequences too. ADOM was buggier than Angband or NetHack, plagued by occasional crashes and gameplay glitches that stood little chance of getting fixed because its creator no longer had time to work on it.

I found ADOM compelling enough that when I returned to school, I found a DOS emulator for my Mac and used it to continue playing. I’d stopped savescumming at this point, but I still managed to get pretty far with a few characters before they died and I had to start from the beginning. What started to sour me on ADOM was, perhaps strangely in this era of hugely story-driven games, that it had too much of a plot. As a result, the beginning of the game saw me doing the same few quests over and over and over: rescue the puppy from the ant cave, save the town healer from the dungeon, kill the roving pack of bandits near the town. It’s in the nature of roguelikes that you will repeat the beginning of the game many times; the more alike each iteration is, the more quickly it gets tedious. The levels were still generated randomly, but too much else wasn’t, and so when I accidentally stepped on an evil altar and a pissant goblin sacrificed me to its chaotic god, I was weary enough to put ADOM down and walk away from it.

I remember this very vividly.

Part of the appeal of roguelikes is their crazy difficulty, but they walk a knife edge between fun difficult and infuriating difficult. Permadeath doesn’t sound fun, but when you combine it with randomness, it’s fun because every game is new. Introduce what’s known as “instadeath” to the equation, though, and in my opinion it tips back away from fun again. NetHack has instadeath out the wazoo. A gnome with a wand of death, touching a cockatrice corpse with bare hands, eating the wrong monster’s corpse: these things will kill you (sometimes literally) stone dead in a single turn no matter how well you were doing on the previous one. Having it happen to a character you’ve put a few hours into can evaporate your goodwill toward a game pretty quickly.

That was probably the better part of a decade ago now, and because we live in an age of minor miracles, ADOM has been resurrected and is once again under active development. There’s even a followup project, ADOM II: a sequel in the same world? A completely different game capitalizing on the lucrative ADOM brand? I don’t know, but I still remember ADOM fondly enough that I’m willing to forgive its Pious Goblin Death Strike and maybe see what’s happened in the last eight years or so.


After ADOM, I believe I went on a bit of a Roguelike hiatus, but by this point I’d been playing them for over a decade. There was no chance of the hiatus being permanent. It didn’t end, though, until I came across the peculiarly named Dungeon Crawl Stone Soup, “DCSS” or “Crawl” for short.

DCSS is actually a continuation of another roguelike abandoned by its creator: Linley’s Dungeon Crawl, the product of an Australian named Linley Henzell. Thanks to this heritage, it contains some uniquely flavorful monsters and items; in early levels you will often see your character fighting quokkas and eating chayote squash (“chokos”) for nutrition. But Crawl’s most salient feature is not its idiosyncratic Australianisms but the clarity of its design principles. It seems like a weird and egghead-y thing to single out in analyzing a game that most people probably use to unwind after work, but I think those principles, and the hard work done by Crawl’s lead designers to ensure that gameplay changes adhere to them, are the reason that Crawl captivated me for years even in my adulthood and made a whole swath of other games seem like uninspiring tedium in comparison.

How can you get a sense of what a game is like to play? In our age of minor miracles, the way that requires the least personal investment is to fire up YouTube and watch a “Let’s Play,” a video where someone plays a game while recording their screen and narrating the in-game events. If a game is a meal, watching a Let’s Play is smelling its aromas wafting in from the next room (and, depending on the quality of the player, possibly listening to them chew loudly with their mouth open). Playing is eating the meal yourself. But there’s a third way to acquaint yourself with Crawl: read the document explaining its design philosophy. You might call it the recipe.

As a particularly illuminating example of one piece of Crawl’s design philosophy (“anti-grinding”), its dungeon has randomly generated shops, but you can only buy from them. All those magical swords, spellbooks, and suits of armour are of no interest to Crawl’s shopkeepers and you can’t sell them. As a result, with few exceptions, the only money you get in Crawl is what the dungeon decides to give you. To players of many other role-playing and dungeon crawling games, this may be startling. All the games I was used to before Crawl allowed you to sell the spoils of battle back to shops, even if only at a drastically discounted price. But when this was forbidden, it made it clear just how much time in such games was spent shuttling back and forth between dungeon and shop with armloads of swag, facing only the challenge of trying to guess what was worth the most. That’s the “grinding” that Crawl is adamantly anti-, and once you’ve played a game where you don’t have to do it, it can feel like a bit of a chore when you do.

Crawl is open source, and it is as thoroughly spoiled as NetHack is. There is a chat room, ##crawl, often frequented by people who are probably among the best players in the world, with a bot (a program that will reply to messages sent to it) hooked up to a tremendous database of information about the inner workings of Crawl and strategies for playing it. With liberal use of this bot and the experts in ##crawl, I have achieved a win rate of about 1%, which I think puts me well above the median. This extreme difficulty persists, despite the wealth of available information, due to exquisite game balance. The low win rates cause some to say that it tilts too far towards being unfair; that its randomness is too unforgiving; that too often one ends up in inescapable situations through no fault of one’s own. But the best Crawl players have a win rate of over fifty percent, and even they make mistakes. Truly unwinnable situations can happen in Crawl: they’re recognized as an unavoidable consequence of its design decisions, but they’re far rarer than you might suspect.

Like NetHack and Angband, Crawl has only a single dungeon, but Crawl’s dungeon has almost as much flavor as ADOM’s world. It has 27 levels with a final 5-level region, “Zot”, at the bottom, and at the bottom of that lies the Orb of Zot, the MacGuffin you need to retrieve and escape with to win. But in order to get into Zot, you need to collect three runes which rest in the bottom of variously flavored branches of the dungeon: the Snake Pit, for example, full not only of snakes but also nagas, human/snake hybrids who are a little more sophisticated in their attempts to kill you. There’s a Swamp, and a Crypt, and The Vaults with wide open levels (anathema to the player who doesn’t want to face down eight monsters at once) and a variety of giants, centaurs and other tough guys. Each branch has its own algorithm for generating levels, ensuring that the Snake Pit doesn’t just feel like the Spider Pit with the monster set swapped out. There are sub-branches and sub-sub-branches, and though you only need 3, you can collect up to 15 different runes. Even among players who’ve won, the ones who’ve ventured into the incredibly dangerous Hells, Pandemonium, the Tomb, and the Abyss to collect all 15 are probably in the minority.

The heterogeneity of Crawl’s dungeon is further enhanced by a vast and continually growing set of pre-built “vaults”, many of which are contributed by players. Though Crawl is open source, the likelihood of any given player contributing code is pretty low: even if you meet the standard prerequisite of enough programming proficiency to write the code you want to add to Crawl, the strict adherence to the overarching philosophy means that code’s acceptance by the lead designers is far from assured. Vaults, on the other hand, have both these bars to entry lowered considerably. They’re mostly described using a reasonably simple markup syntax, with the code needed to implement any sophisticated interaction (e.g. gates triggered by floor switches or countdown timers) written in Lua, which is far simpler than the C++ that makes up the bulk of Crawl’s code. Maybe more importantly, vaults are much easier to add without breaking Crawl. As long as the loot isn’t too bounteous, the monsters too outrageous or the traps too unfairly nefarious—and they can be pretty damn nefarious—a randomly appearing vault just adds another dash of flavor to the dungeon. The end result is a place that often feels as if it was malevolently designed despite being mostly algorithmically generated.

Back to Crawl’s design philosophy for a second. Here’s another thing that many games make you spend a lot of time doing: running around from place to place. With the depth and breadth of Crawl’s dungeon, traversing it could take a great deal of time if not for a brace of features the likes of which don’t exist in any other game I’ve ever played. In Crawl, you can do the following:

The end result of these autotravel features is that traveling to anywhere in the dungeon, with remarkable precision, takes a matter of seconds plus however long it takes you to deal with any monsters between you and your destination. I can’t overstate how much tedium autotravel removes from the game.

If I haven’t made it obvious already, my discovery of Crawl was an epochal moment in video games: in short, it ruined a lot of them for me. Autotravel alone made me loath to go back to any Roguelike that doesn’t have it, and decreased my tolerance for spending time just running around in any other genre of game by about 90%. RPGs that make you grind levels to advance are pretty much out of the question now that I’ve played the hell out of a game that doesn’t make you do so. Crawl took games whose tropes I enjoyed and trimmed huge amounts of fat that I didn’t even realize was there. The result is a game that, for the most part, just throws up one interesting, tense challenge after another with very little dead space between.

Despite all this, Crawl is not the only game I play anymore. I still enjoy a well-put-together indie platformer and, though I’ve mostly given up on AAA games, I played and enjoyed the hell out of Portal and its sequel like every other gamer with a pulse. I just don’t play most of them a second time after I beat them once. I’ve beaten Crawl 40 times. I’m sure I’ve spent more time in my adult life playing it than any other game. Once you’re used to the quirks of the genre, it has a crazily low bar to entry: starting it up takes a matter of a few seconds, getting yourself into an interesting or dangerous situation generally about a minute, and you can stop instantly at literally any time without losing any progress.

At this point, I consider Crawl the culmination of my lifelong interest in Roguelikes. I spend less time on it these days: my self-imposed Crawl hiatuses longer, my binges shorter and less intense. But I still find it compelling, and of course it fundamentally changed my outlook on what makes a video game good. Nevertheless, there is another game that I mention as a coda, a sort of Platonic ideal of Roguelikes that reveals how even Crawl has a fair amount of extraneous stuff that adds flavor and character but isn’t needed for a concentrated shot of pure tactical gameplay. Of course, if you removed that stuff, Crawl wouldn’t be Crawl. It would be something more like


Brogue is quite a recent Roguelike that seems a bit like it took Crawl’s after-the-fact design philosophy and built a game around it from the ground up, with none of Crawl’s Australian heritage or somewhat more loosely designed foundations to preserve. Crawl has about 27 each of races and classes, combinations of which make for a bewildering array of options for the kind of character you can build. Add its religion system, which lets you choose from among a dozen or so different gods, and you have enough different types of characters to last a lifetime (or, rescaled to the average longevity of a Crawl character, seventy or eighty thousand lifetimes).

Brogue has either one or zero classes, depending on how you define the generic @ with fixed equipment who starts the game, and no religion system. A recent change, occurring in the 1.7 version of Brogue, also removes the entire experience and leveling system, which is a mainstay of almost every other RPG—including traditional Roguelikes—in existence. The stated reason for this is that gaining experience for killing monsters disproportionately rewards a playstyle based around doing so when sneaking around them or relocating them out of your way should be viable as well. Even after having my mind gradually reconfigured by Crawl’s forward-thinking design, I was taken aback by what might seem like removal of the skeletal system of a game built around a character’s facing progressively more dangerous challenges. But, in fact, I had to double check the release notes to learn this, after switching from 1.6 to 1.7 and playing several games before noticing something missing.

So how does character progression happen in Brogue? How do you go from being killed by jackals to being able to slay a dragon? The answer to that is the same as the answer to how each game of Brogue can be completely different despite the starting character having exactly the same attributes every time: all progression and differentiation happens through items, which are randomly generated like everything else in Brogue’s dungeon. Here’s how you get a Brogue character ready for the dangerous depths of the dungeon:

A comprehensive list like this for Crawl, though it would end similarly, would have probably about ten times as many entries in it. Crawl has about fifty times as many monsters, too; for example: swamp, fire, ice, storm, shadow, quicksilver, iron and golden dragons where Brogue has exactly one variety and doesn’t seem to want any more than that.

The end result is that Brogue is a game I admire immensely for its restraint and focus, but one that will probably never replace Crawl entirely in my affections. The thing is, it doesn’t seem to be vying for them. The eight species of dragons, proliferation of dungeon branches, hundreds of different character combinations and items: all of them give Crawl a lot of flavor and mystique, a sort of implied narrative even though its philosophy is decidedly anti-story. In pure design terms, Brogue is almost certainly objectively better than Crawl, but it feels like a game. Crawl feels more like a world. I fear these two things may be somewhat in opposition to each other, but it means there will always be a place for both Brogues and Crawls.


How much of my attachment to Roguelikes has been due to my exposure to them as a kid? Plenty of people try them and, understandably, can’t get past their rough edges: the lack of graphics, the unintuitive interface, the unforgiving permadeath. If I hadn’t persisted with Angband and learned the idiosyncrasies of Roguelikes back when I had fewer options, would I have the patience to stick with Crawl if I dove into it, blind, now? I sort of doubt it.

Having made my way past those initial hurdles, though, I find that the best Roguelikes offer more reasons to revisit them repeatedly even than games with big development teams, big budgets and price tags to match. Their turn-based nature almost gives them sort of the feel of puzzle games, but whereas most puzzle games have the problem that finishing them once suffices to consider them solved, the randomness of Roguelikes ensures that they have a nearly inexhaustible supply of different situations you can find yourself in.

Roguelikes aren’t for everyone. But I’ve been playing them for something like two decades, some intermittently and almost exclusively for years at a stretch, so I think I can safely say they’re for me. If this essay has a thesis, it’s that you should give one a spin.

  1. Nerd pedantry pre-emptive counterstrike: that’s the (somewhat dumbed-down) American release of what’s now known as Final Fantasy IV. America didn’t get the original releases of Final Fantasy II, III or V, so IV became II and VI became III. ↩︎

  2. “True” roguelikes are almost always also turn-based, rather than realtime: the world waits for you to act, giving you time to make the right decision (if you haven’t already foreclosed on it with poor previous ones). If you relax some of these criteria, you might get a game like Blizzard’s Diablo (real-time, no permadeath except in “hardcore mode”) or Derek Yu’s Spelunky (real-time 2D platformer), but these are often known as “roguelike-likes.” All this terminology is kind of crappy, but unfortunately it’s also fairly entrenched.↩︎

  3. This is the kind of phrase that Keith Burgun would use. Keith is a game designer with very strong opinions about what constitutes a “game” and what makes a game good. He would probably say that games such as Halo, Portal, and all Super Mario games are trash, and is on record as saying so of Final Fantasy VII. I am almost sure I could find a quote from him saying, in effect, that the quality of a game is inversely proportional to the amount of story it has. Because of such controversial opinions he tends to start shitstorms whenever he discusses game design, which he does a lot. Keith has designed a roguelike, 100 Rogues, but he now says the genre has fundamental flaws. His true love seems to be board games, and his latest game Auro looks to have a lot in common with them.

    Personally, I think Keith Burgun is an insufferably opinionated, caustic, narrow-minded blowhard with a lot of fantastic ideas about game design that he seems to be actively working to make as unappealing as possible to as many people as he can infuriate. I’m really looking forward to Auro. ↩︎

  4. [sic]. Australia, remember? ↩︎

  5. This is true in the conventional sense that vaults are less likely to introduce bugs into Crawl’s code that might cause it to crash or behave erratically, but “breaking” a game also means to drastically unbalance it: to introduce a technique that enables the player to defeat most or all dangerous situations with little effort. To use an example from a previous footnote, Final Fantasy VI had a spell combination, “Vanish/Doom,” that would enable the player to defeat almost anything, including a boss, in two turns. The term can be used two ways: Vanish/Doom was broken. It broke FFVI.

    Crawl has had broken features; I don’t think any were broken that badly, and most or all of them have been either weakened (“nerfed”) into mediocrity or removed outright. I feel like a significant proportion of my 40 Crawl wins came through abusing broken things in pre-release versions of the game, and even in those cases, there were few wins that I would consider to have been “easy.” ↩︎

  6. I have seen indications that Brogue makes use of a randomization technique known as “rubber banding.” As the evocative term suggests, rubber banding adjusts the likelihood of something happening based on how many times it has happened previously: in Brogue, this is used to prevent your finding too many or too few of some very critical types of items. So if you’ve found more of such an item than normal, they’ll be less common, and vice versa. These kinds of shenanigans aren’t necessary in Crawl because there are so many possibilities that even if you don’t find any of something important, there are typically ways to get around it. In the more ascetic Brogue, a paucity of one kind of item can torpedo a whole game with little hope of recovery. ↩︎

  7. As evidence of this claim, take my experience with Dwarf Fortress. It’s hailed as an unparalleled world simulation/Roguelike hybrid, and there seems to be no other game like it. Its motto, “Losing is fun,” sounds right up the alley of a dedicated Roguelike player such as myself. But, Reader, I tell you that when I tried to play it, I took one look at the number of skills that a single simulated dwarf could learn, said “I don’t have time for this shit,” and unceremoniously deleted the whole thing from my hard drive. ↩︎

  8. I’m certainly not the first or only one to see this connection. See for example Desktop Dungeons and Dungelot, which sort of split the difference between dungeon crawls and Minesweeper. Or, as Dungelot’s web page charmingly puts it, “almost like a strategic Advent Calendar.” ↩︎

Contact me? My email is my first initial and last name, which you can deduce from this website's URL, at I'm also on Mastodon as