The Role of Randomness in Rogue-likes

I JUST BEAT FTL ON NORMAL yesterday and that (heart-pounding, jazz-handing) experience got me thinking about rogue-likes. The devaluing of the term rogue-like is in full effect. If a game has perma-death: it’s probably a rogue-like. If a game has randomly generated levels: it’s probably a rogue-like. If a game is punishingly difficult: it’s probably a rogue-like. Though these basic game design similarities imply a similar gameplay experience, the truth is that the variations within this “genre” result in wildly different experiences. One of the central differentiators in modern rogue-like design is the role of randomness. Each rogue-like incorporates some degree of randomness, but how and where that randomness is employed can be the difference between a game feeling tough, but fair, and feeling like a Sisyphus simulator. I’m going to take a look at a few of the prominent rogue-likes that have come out the last few years to point out how those games have incorporated randomness in their design, for better and for worse.

The first game I would like to look at is The Binding of Isaac, a dual-joystick shooter with all of the trappings of a rogue-like. The Binding of Isaac is full of potentially frustrating room layouts and luck-based difficulty spikes. If you go a floor without seeing any health hearts: bad luck. If you get three fart pills in a row: bad luck. However, the game itself does have an element of skill to it. Skillful dodging and shooting can be enough to progress deep into The Binding of Isaac, even without any great item drops. This skill elements helps alleviate the wild luck swings. You always feel like there was an element of personal culpability when playing since there is no randomness in how you are damaged. If you get hit, you get hurt; if you don’t want to get hurt, don’t get hit. On top of the sheer skill component, The Binding of Isaac does have some consistent upgrades built into its levels. The first floor of every game always includes at least one upgrade room. Every floor always ends with a boss that always provides a few hearts and an upgrade. Thus, the game builds some known elements that the player can rely on into its random levels. These constants help to alleviate the crushing feeling of losing a run, since the next run could immediately be promising if you get the right items on the first floor. Now elements like the slot machines, beggars, shops, and secret rooms do introduce frustration-inducing variance, but good luck is not necessary for playing a long game. Good luck is helpful, but not mandatory.

FTL: Faster than Light errs further towards randomness in its design. Other than which starting ship you pick, there is no single constant in FTL to rely on. While there is certainly skill involved in how you approach battles and events, there are very few ways to plan out what encounters you engage in or run from. Sometimes you run into a ship and are offered the option to try to avoid conflict, but your choice is not always honored since the game decides what is going to happen based on dice rolls. Furthermore, most decisions are only made with partial knowledge. For example, you could try to stop a colony from getting attacked, but you never know what ship you will be up against. As a result, FTL requires both flawless execution and star-aligning-luck to beat on any difficulty, but especially Normal and Hard. FTL constantly hides information from the player and forces the player to adjust to the shitty situations it concocts. One regularly occurring event asks a simple question: do you want to attempt to fight space spiders? If you win the dice roll, you are rewarded with free scrap and resources. If you fail, a random crew member is permanently killed. These two wildly different outcomes are the random result of the same simple choice and it is that high variance that makes FTL so frustrating. The reliance on high variance outcomes for choices that are themselves determined by random chance can lead to a feeling of powerlessness or frustration if you get too invested in one run. This randomness also leads to a brutal game that offers “this could all turn to crap at any moment” thrill and catharsis that few games can muster.

Spelunky lands firmly in the skill camp, but it also adds an element of persistent progress that eases the journey to the end game for less skilled players. Winning a game of Spelunky requires intense skill and focus, but it is entirely possible to beat the game without having a “lucky run.” Though some exploits and tricks are luck based, actually beating the game doesn’t require a forgiving item drop or a nice level layout. On top of the game being largely skill based, the game offers you the chance to build shortcut tunnels that bypass entire sections of the game. These shortcuts allow less skilled players to see the end of the game without navigating an entire game worth of precarious jumps and traps. As a result, Spelunky is less daunting or frustrating to finish than FTL or The Binding of Isaac… if finishing the game is your only goal. Spelunky also introduces a bunch of leader-boards: perfect for inspiring multiple playthroughs. Between the time based speed runs and the max money runs and the eggplant runs, Spelunky gives players a ton of avenues of play to keep the game interesting without relying on randomly generated roadblocks.

Rogue Legacy takes the constant progression found in Spelunky and adds permanent character growth to the equation. In Rogue Legacy you are able to unlock new character classes, character buffs, and skills that carry over from one run to another. Furthermore, the bosses in Rogue Legacy stay dead when you beat them. This leads to a game that can be brute-forced if necessary, since you can out level enemies and bosses over time if you are so inclined. While the game is reflex and strategy intensive, it does not require its players to truly master its systems thanks to its permanent progress. Every run in Rogue Legacy has its luck elements: your hero and caste configuration. However, even these random elements can be mitigated since you can pick between three heroes every generation and you can lock in configurations of the caste to keep the castle constant. The end result is a rogue-like that utilizes short term randomness, but on the whole is governed by attrition. Almost anyone can see the end of Rogue Legacy if given enough time. Whether that low barrier to victory is a positive or a negative is up for interpretation.

Randomness is one of the foundational concepts that rogue-likes are built upon. Too much randomness and a game can feel unfair or cheap. Not enough randomness and the game loses some of the trill and challenge that people love. Now it seems like the real question here is whether high levels of randomness are a bug or a feature. Is it poor design that it took me around a hundred runs to beat FTL on normal or is that toughness essential? Alternatively, is it good design that I was able to beat Rogue Legacy pretty quickly just by leveling or is that a devolution? Most rogue-likes present a pretty stark rejection of the player-control based game design philosophies that seem to be dominating the industry right now, but then maybe that’s the point. I personally hate feeling out of control in a game. I feel cheated when I lose due to something completely out of my control, but then again learning to accept that things can be out of my control is certainly worthwhile. Ultimately, the thrill of beating a game that seems actively malicious is a rush that few games outside of rogue-likes can offer (I BEAT FTL ON NORMAL!), and that is why I continue to play them.

Tennmuerti will be in here white knighting FTL's lack of randomness any second now. Roguelikes are neat, doesn't mean every indie game should be one and doesn't mean we need more than a couple per genre other than actual roguelikes (Castle of the Winds, Dungeons of Dredmor, and so on)

@fredchuckdave: I can buy that FTL is not completely reliant on randomness, but certain aspects of the game are completely beyond your control. Like running into a ship with a beam drone and a beam weapon as your first opponent when you have a stealth ship with no shields (has happened to me). I think indie games are leaning into it so hard because random generation is a real time saver if you can manage it. Creating a handful of level templates and enemy types and then letting an algorithm lay them out can be less labor intensive than making levels by hand. Plus you get a ton of replay-ability that is flat impossible without some randomness. On the whole, making a rogue-like is some of the best bang-for-your-buck you can get as an indie dev.

I didn't mind the randomness in FTL, because it felt somehow appropriate to the game's setting and story that every run was bound to end in faliure. It's almost not a question of "if", it's "when".

I guess that's what I like about the few roguelikes I've played. Lots of games try and make you feel that "the odds are stacked against you", but roguelikes are the only games where the odds really are stacked against you - with the results you'd expect. So it feels like it makes narrative sense that some totally unpreventable tragedy might befall Captain Butters and his crew, or that Lady Sonia VII could get killed by a trap her traits gave her no chance against, because there was only one way their ill-fated missions were going to end anyway.

@immortal_guy: My main issue with rouge-likes is that the odds can be stacked against you, or they could be wildly in your favor and you have no control over which occurs. I never know whether I am doing well or the game is just giving me handouts. I like testing my skill and expertise, and some rogue-likes really don't care about my skill beyond my ability to not screw myself over. Also it seems really ambiguous as to when I succeed or fail sometimes. But, then isn't that how life works sometimes?

See this is my central struggle: fighting to claim that I succeed due to merit, when I know that it could easily be luck.

I tend to agree with you that I generally want to feel in control, as it feels like a game is disrespecting my time when it applies random effects to my experience to a point where my input feels greatly diminished. Most of these games have a skill component, as you pointed out, but I value my time too much to feel good about spending it on a game I could play for an hour, only to have something out of my control end it on the spot. I can accept when I make a mistake; that I don't consider wasted time, as I will learn and improve from that. But there's diminishing returns when combating pure randomness (you can only improve your odds so much with skill), and I rarely feel like it's worth my time.

On the other hand, I know people who love randomness because it makes every time they play the game feel "unique", and that makes it more replayable. In some ways, I agree with wanting each playthrough of a game to feel unique, even if I don't care a ton about "replayability" (I'm usually a "play through a game once" kind of person). But reducing the player's control seems like a lazy, disrespectful way to do it to me.

Certainly it's a problem if random procedural generation creates an unwinnable situation. However, the best roguelikes are capable of generating scenarios that look unwinnable to an inexperienced player, but as you learn the game's more complex systems, you start to realize there are methods of mitigation -- strategies to survive and escape even the most dire situation the game is capable of producing. In other words, it's the player's skill at understanding the systems that really determines their ability to succeed. This is doubly true in traditional turn-based roguelikes. Of course for this to work, the game's systems have to be both complex and carefully crafted, which is a tall order.

In Spelunky and Issac, you theoretically can "win" (meaning you can complete the basic path through the game) without picking up any items at all. I'm pretty sure the designers purposefully made those games that way. If you're lucky enough to find great items, that's frosting on the cake, but you should never need them. I haven't played FTL, so I'm not sure if that game can generate runs that are just predestined to be unwinnable.

@prestige: The problem for me in FTL is that I can;t really tell if a situation was truly unwinnable because so much information is hidden from the player. If I choose a bad point to jump to then I could easily die on my first encounter. However, simply choosing a different point could have resulted in the easiest run I've ever seen and I would never know. Spelunky and Isaac punish impatience more than hammer you when you have bad luck. FTL does both.

@majormitch: Yeah I have long been a one to two playthrough person for most games and I love story in games. As a result, I did not really like rogue-likes at first. But, it turns out I'm more competitive than I thought so the idea of taking on a challenging game is really enticing.

@majormitch:

I could see you liking Spelunky though.

Out of all these, it is the only one imo that I don't feel like I get occasionally cheated by randomness. It never feels unfair like most roguelikes do, like Dark Souls in a way, Spelunky forces you to get better at it. Its' mechanics are very good at teaching you the technique it takes to win.

@slag: I feel like Rogue Legacy doesn't so much cheat you as much as you just can't beat certain things until you unlock certain abilities. Now some of that stuff is luck-based, but most of the time it is simply whether you can double jump and dash or not.

@fredchuckdave: Naaaaaah. I feel like I said my piece already in other threads. And I only really white knight it when the randomness it is seen as some major flaw rather then a challenge that needs to be prepared for or conquered, being much of the point of the game in the first place.

@thatpinguino Generally only at the very start can one encounter really unwinnable situations in FTL. And they do indeed exist. But after a few sectors it's 99% up to the player to succeed.

For example that spider situation you mentioned, an "experienced" FTL player will simply not risk the dice roll in the first place if they are low on crew, may risk it if they have disposable crew members, and will choose to engage when they have the 100% success options like an anti personnel/boarding drone. Skill/experience in FTL can and does compensate for randomness, it can be easily observed with streamers who will consistently do successful runs.

@fredchuckdave Ok ok maybe a little bit :P

@slag: Of all the recent "rogue-likes", Spelunky is easily the one I've played and enjoyed the most. It does feel more skill-based than the others, which I appreciate, and when I die it usually does feel fair. But there's still always some amount of "luck"- different item and level layouts are going to help or hinder your cause to varying degrees. Sometimes I can bumble and/or brute force my way through a level unscathed, and sometimes I can play well and still have a super hard time. In other words, in my time with the game I don't feel like how well I'm playing is directly analogous with my results, which is I guess the main idea. I'd like to feel that if I'm playing better, I'll have better results, and vice versa. While that's more true in Spelunky than these other games, I don't think it's universal, and that can bum me out at times.

@thatpinguino:

I love Rogue Legacy, just love It was my number 4 GotY last year

http://www.giantbomb.com/profile/slag/lists/goty-2013/88492/

that being said you can definitely get real bad draws in the game, especially related to teleporters and distances to the different zones/bosses. Especially if the heroes available have classes you don't level up or have traits you don't like playing.

I do agree the semi-permanence of the game makes it very tolerable, but if that wasn't there I don't like I would have liked the game nearly as much.

@majormitch:

It does have randomized levels that's true. I imagine that's a practical concern, as a Spelunky with set level designs would probably be shorter than even Gunpoint.

I guess I don't feel the item/level draw is as determining of my play as you do. It feels like a game where you can have good luck and rarely truly bad. The level design rarely ever causes my deaths. Because none of the obstacles ever feel "unfair" to me, thus it feels like my skill is rewarded.

But yeah I feel ya on real roguelikes like Dunegons of Dredmor. That game is fun but I can't play that for long because the randomness of the outcomes annoys me.

Perma-death and randomly generated level design is key combination to rogue-like-like-esque-ish games. You need the randomness to maintain replayability in conditions where you will die frequently and be forced to start at the very beginning of the game again. If it was the same levels every time, it would be less compelling.

I love Issac, Spelunky and Rogue Legacy, though I am not good at any of them. For some reason I get a huge kick out of the random element in each and I love the feel of a new game every time. The only game I feel like I have gotten better at is Spelunky. That game makes you learn from your mistakes.

I've played a little FTL, but I need more time with it.

This is exactly it, if it was always the same you wouldn't want to play through however far you got to get to where you were.
It's also cheaper than actually designing levels.

Debatable! Though having infinite levels is a really nice perk to have in your game.

@tennmuerti: Yeah it is interesting how the choices in FTL are usually to either expose yourself to a high risk/reward situation or avoid the situation all together. The blue dialog choices are a great way to provide good options to terrible situations though.

Yeah, making a good level generation algorithm can be just as daunting as hard coding the levels. Not to mention making all of the component parts in such a way that they can be randomized. The reward of mostly infinite replayability is worth it though.