Chaos Engine
More games than ever run on randomness, and it’s transformed the industry. But for better or worse?
More games thaan than eever ever run on randomness, annd and it’’s it’s transformed the industry as we wew know kknow it
We’ve all experienced moments in games like this: wandering deep in a dungeon, you encounter an ogre. It’s beaten you nearly to death, but you’ve one hope left. In your pack is an unidentified potion that you picked up in a room a few minutes ago. With no other option, you drink it, because it could save the day if it’s a potion of life. Or it might be a potion of hallucination, which will mean that in your final moments you’ll see the dungeon walls pulse with imagined colour, and the ogre apparently turned into a tuft of grass as it finishes you off. Your fate in this game of Brogue is in the hands of randomness.
Here’s another: your veteran assault soldier is standing right next to a sectoid, ready to fire with her shotgun. The game says you have odds of 90 per cent to hit, and yet, when you hit Fire Weapon, she misses. Your turn is over, and the sectoid kills her. In this XCOM mission, randomness again rules your fate.
Since their inception, videogames have used randomness to add a dose of unpredictability to the cold sureness of computer logic. Will the enemy do this, or that? Will your final axe blow take down your assailant, or will they get the last hit? Will you get the special item, or the booby prize? It can make games interesting and mysterious; it can put your heart in your mouth and make you punch the air with relief. And today, randomness is being used more than ever.
It’s instrumental in Roguelikes, where it builds procedural worlds. An entire genre has exploded around the randomness of card draws, from Hearthstone to Slay The Spire. It’s also turning up in competitive games, which have traditionally resisted randomness. Fortnite couldn’t work without it; Dota 2 has introduced more randomness to its ruleset than ever before, while new competitive games such as AutoChess are built on it. “I love randomness,” says Zach Gage, developer of random-soaked solitaire games Fliptop Solitaire and Card Of
Darkness, as well as Really Bad Chess, in which he introduced randomness to one of the most sternly deterministic games of all time.
There are many reasons why, and they touch upon the vast shifts in how games are made and played. Randomness is intricately bound into the rise of indie developers, in the economics of in-game purchasing and methodology of games as a service, and it plays into the culture of game spectating on YouTube and Twitch. It speaks to the growing sophistication of game design as a craft, as it brings new ways to make randomness fun, engrossing and rewarding. But it also raises questions. Powerfully seductive and bound into the deepest layers of our psychology, randomness can be misused, a misbalance of the videogame away from mastery and towards gambling.
Zach Gage didn’t start out as a lover of randomness. In fact, he once resisted it, co-designing card games such as 2011’s Scoundrel, which is all about stripping out uncertainty from the deck. But then he took a trip to Atlantic City for a stag party and played a lot of gambling games, including threecard poker. “The level of depth and nuance and craft in gambling games was unbelievable, but three-card poker was just so fun to play.”
Three-card poker was designed by a player called Derek Webb in 1994, and today it’s widely played at casinos. In it, a gambler plays three-card hands against the dealer, choosing once whether to play their hand or not. It’s a simple setup, but one that’s given a sense of depth by the way they have three different bets going on against the outcome: one directly competes with the dealer’s hand, one is on the dealer getting dealt at least a queen, and another gambles on the player having a pair or better.
“The beauty of it is that you only do one thing in the game, one choice, and that choice feels informed, but really isn’t that informed, and I just couldn’t get over how much I loved it anyway,” says Gage. “That’s what turned me on to the
WHEN WE FACE UNPREDICTABLE SITUATIONS, WE FEEL THE COMPULSION TO UNDERSTAND THEM
positive qualities of randomness, and since then I’ve been deeply investigating it.”
Randomness works because it goes against our certaintycraving nature. When we face the unpredictable, we feel the compulsion to understand it, to turn future such events into the predictable and safe. In three-card poker, you’re playing the odds – the 17 per cent chance of getting a pair, the 3.2 per cent chance of getting a straight – but it offers different levels of success, which make a winning strategy seem graspable, yet it remains just out of reach. Your decision is clear and powerful, and the results lie with the gods.
”Videogames and gambling games both harness our attraction to uncertain rewards, such as continued responding on variable-ratio schedules,” says Luke Clark, professor in the Department Of Psychology and Director of the Centre For Gambling Research at University of British Columbia. This principle lies at the heart of how we react to randomness: it says that when we’re rewarded unpredictably after performing an action, we instinctively attempt to perform it again and again, much more frequently than if we’re predictably rewarded. Think about a fruit machine: the very fact the payout is unpredictable is why we want to put more money in, because the next time might be the payout. If we know when the reward will come, the compulsion is far less.
But while randomness resists predictability, it isn’t necessarily unknowable. We can make educated guesses and have a feeling for the bounds within which the results will lie. And it’s on this knife-edge of knowing and the unknowable that games try to balance. “Chance and randomness are not universally appealing, so game designers need to approach this with care,” warns Clark. For example, the Ellsberg paradox says that people tend to avoid gambling when probabilities are hidden or unknown. We need to know our chances before we’ll take them. “And obviously, uncertain negative outcomes cause anxiety.”
One game that’s all the richer for walking this line is Slay The Spire, which mixes deck-building with Roguelike dungeon
delving. Randomness is everywhere: it defines the hand of cards you’ve drawn from your deck, the cards you’re offered after defeating enemies, the relics you win by defeating bosses, and what’s on offer in the shop. But all that uncertainty is balanced by the fact that you steadily edit your deck over the course of a run, taking out undesirable cards and choosing to add new ones, giving you an active role in raising the odds of drawing cards that will see you win.
“How you implement the randomness really matters,” says co-designer Anthony Giovanetti. “When you’re dealing with randomness, a lot of the system is determined by how the player feels about it and how they can reply to it.” Slay The Spire’s randomness is also mitigated by the Intents system, which gives you an idea of what the monster you’re facing will do next: if it’s an attack, you know what damage it’ll deliver. If it’s going to use an ability, you know roughly what kind. Without it, Giovanetti says, the game felt too random. “You couldn’t make good strategic decisions. It felt like a dice roll; it’s randomness you can’t interact with and can’t meaningfully know or understand, and it’s not bringing in variety.”
So the Intents system works with Slay The Spire’s randomness to give you, the player, a variety of
RANDOMNESS “FORCES THE PLAYER TO ADAPT TO NEW THINGS RATHER THAN RELYING JUST ON MEMORY”
choices. When you know that an enemy is going to hit you for nine damage, you can decide whether to spend valuable energy on playing defence cards, or on attacking. Your fate feels in your own hands, rather than purely in the hands of randomness. You are its master – or at least it feels as if you could be its master.
These choices, this variety, is one of randomness’ great strengths. It can provide constant novelty and surprise without a developer having to create it manually. You can play infinite numbers of dungeons in Brogue and Spelunky, their levelgeneration systems designed deftly enough that they can still surprise players who have thousands of hours of experience. These Roguelikes (as inspired by the dungeon crawler Rogue, which was released all the way back in 1980) have solved the problem of how solo or small-team indie developers can make huge games with enough long-term appeal that they can compete with studios which employ hundreds of developers and budgets of millions of dollars.
But randomness is not just about production efficiency. The variety it can conjure into existence can also change the player’s relationship with the game. ”It forces the player to adapt to new things rather than relying just on memory,” says Giovanetti. Think about Mario, which has pre-designed levels that are the same every time you play, and enemies that generally behave in pre-defined ways. Each attempt on a level is an exercise in memorisation, as you set and then perform a routine to beat it, down to where you hit the jump button on each platform. Now, imagine playing Spelunky. You can’t memorise any levels, since you’ll never play them again. Instead, each run is an exercise in learning Spelunky’s systems: understanding what a bat does, how to deal with an arrow trap, to look out for spiders lurking on ceilings.
”The whole appeal of Rogue, and frankly, of all its successors, is learning over many hours of play how to optimally tackle a level you’ve never seen before,” says
Brian Walker, who is the developer of Brogue, a philosophically faithful but far more sophisticated adaptation of Rogue, with intricate dungeon generation and a physics
system in which grass burns and poisonous gasses spread. “And that experience gets more satisfying the more hours you’ve invested into doing it.”
“I love randomness because I care about making videogames that are really hard and that you have to learn, and I care about figuring out how to teach players how to come up with strategies and explore the depths of them,” agrees Gage.
If you follow Star Wars Galaxies designer Raph Koster’s theory of fun, it’s deeply rooted in learning. “Fun from games arises out of mastery,” he wrote in his 2005 book, A Theory Of Fun. “It arises out of comprehension. It is the act of solving puzzles that makes games fun. In other words, with games, learning is the drug.” It’s absolutely fun to figure out and then memorise the best way to get through a pre-designed level or area, whether in Mario or Half-Life or even Skyrim, but, as Koster went on to write, “You’ll only play [a game] until you master the pattern.” Games based on memorisation tend to have shorter lives than those with systems to master, since you’ll sooner run out of new things to learn.
“But these kind of games run the risk of being very dry,” admits Gage. He, however, sees the potential for randomness to buff the corners off hard games. “For players who aren’t
comfortable with games and go into them without expecting to know what they’re doing, randomness puts a curtain down over the complexity,” he says. “It makes the game feel friendlier; sometimes the challenges are easier, sometimes harder. When you don’t really know if you did that, or if the game did, it allows them to flourish and find comfort, even when the game is really hard.” And so, you can play a level of his actually rather difficult Apple Arcade game, Card Of Darkness, and fail, the cards seeming to go against you at every turn, and happily try again.
On the other hand, Gage acknowledges, the same game can feel very different for an experienced player. “If you come with an expectation of having mastery over it because you have literacy in similar games, then randomness can feel like the game is cheating you out of the skills you’ve learned before.”
“I think theming is really critical here,” says Greg Borenstein, a designer at Riot Games, on its auto-battler, Teamfight Tactics. “In Teamfight Tactics, the thing that works really well for us is that the theme is having this team of champions who fight for you; you want them to do well and you root for them, or you criticise them when they don’t succeed. The classic example is that every character has a base 25 per cent crit chance to do double damage. It creates these dramatic moments, where you have one unit left and it’s about to die. But if you get a crit off, you’ll kill their character before their spell comes up, and I’m sitting there saying, ‘Crit, come on, crit!’”
Teamfight Tactics is heavily designed around randomness, since the slate of champions you get to choose from is random, and their actions during a battle are out of your control. But by placing the focus on these autonomous champions, the game feels less capricious: the champions become totems for the randomness, there to absorb the blame for an unexpected bad result and to explain a win that comes from out of the blue.
We’re an emotional and superstitious species, after all, always anxious to create meaning, often in terms of a story, where there is none. “People get extremely complicated emotional relationships with their dice,” says Borenstein, remembering a D&D-playing friend who’d put their set in the freezer to punish them for bad rolls. Since they lack a physical dimension, a tangible thing to blame for bad luck, videogames themselves tend to be blamed for a bad roll: that missed shot in XCOM. Players will put their perception of the outcome down to unfairness, rather than the roll not going their way.
Putting the blame on champions is one way to deal with player reactions to bad outcomes. Another is pure design. Sid Meier spoke at Game Developers Conference in 2010 about trouble he and his team at Firaxis faced in Civilization Revolution’s battle system. The game previews the powers of each unit, so if a veteran Warrior has 1.5, compared to a Barbarian’s 0.5, the Barbarians have a 25 per cent chance of winning. “But that’s not the way our players thought,” he said. They complained when they didn’t win a battle in which they had a 75 per cent chance of success, despite being quite content to win a battle when they had a 75 per cent chance of losing, which is the same situation in reverse. “We found that around three- or four-to-one odds, people expect to win every time,” Meier said. And on top of that, while testers were OK about sometimes losing 50/50 battles, they didn’t like losing two in a row.
Meier’s solution was for the game to fiddle the outcomes, taking into account the results of previous battles. “We don’t only do this to make players happy. It’s really when something happens in a battle like this that feels wrong, we start to lose that suspension of disbelief, the player comes out of the game and pays attention to other things that are going on.”
Many, if not most, other games tweak their randomness for the player. “Randomness has to be wrestled to the ground to prevent it from coming up with inappropriate pacing in Brogue,” says Walker, laughing. One example relates to an item that’s critical to the player’s advancement in the early game; if it were left to generate like any other item, some dungeons would feature many of them and be too easy to play, and some would have few and they’d be too hard. “So I came up with this system of rubber banding, where every time one of those items is generated, the odds of generating another go down. And every time a different item is generated, the odds go slightly up. That gives a consistent pace of generation without ever putting you in a position that you know what the next item is going to be.”
So randomness is sometimes, and invisibly, slightly less than random. And here we stray into one of its problems, one to which game designer and writer Keith Burgun, creator
of puzzle-strategy game Auro, has based a philosophy of opposition. He believes strongly that there are sweeping good and bad uses of randomness, and he’s embarked on a project to design a game at least as deep and lastingly involving as a Slay The Spire or Brogue, and which uses none.
To understand his issue, we must first understand two of the key ways games use randomness. Many of the examples we’ve discussed so far are of ’input randomness’, which is the flipside to what’s known as ‘output randomness’, terms which were coined by board-game designer and writer Geoff Engelstein. Input randomness presents a set of new information that’s given to a player, to which they get to react by making a decision. It’s the new hand of cards you’ve been dealt, or a procedurally generated dungeon. You couldn’t predict what you were going to get, but now you have a chance to act on what you now know.
Output randomness is generated after you’ve made a decision. It’s the D&D skill-check, the XCOM hit result, the Civilization battle outcome. It’s randomness inflicted upon you. It can be a subtle distinction, since it’s often invoked by a prior decision in which you were given an idea of the odds – that low-chance shot in XCOM – but it’s about getting dealt results which you can’t have predicted.
Lots of games feature it, but in general, designers don’t like featuring output randomness in their work. You probably don’t enjoy encountering it, either. It can feel pernicious, and Burgun thinks it’s ‘bad’ randomness because output randomness can mask the relationship between cause and effect, and so it can impair our ability to learn. “You can set up a big play in, say, XCOM, and there’s all this meaning around how you moved your characters and what they’re equipped with, and then you shoot and everything’s a miss,” says Burgun. ”It creates this feeling that maybe some of what you did isn’t being expressed in the final outcome. Some people will say it’s unfair, and that’s one way to look at it. But I care that it’s an inaccurate representation of what’s been happening. Why did my solider die? Because the dice rolled a certain way.”
So, Burgun argues, while the odds might be connected to your prior decisions (where you moved your soldier to, how you trained them, the gun they have), the roll itself was not, and that means it’s harder to figure out whether your soldier dying was down to your bad decision-making or simple happenstance. “The more random something is, the less efficient the system will be at giving you reliable feedback,” he says. That means you have to put hours into playing a game before you can understand it, artificially stringing out the value you get from a game that perhaps has less depth to it than it first appeared, and is therefore disrespectful of players’ time. And, what’s more, it can mask the true quality of a game. “Randomness is used as a crutch,” he says, to hide weak game design, to give the illusion of complexity.
But Walker, who uses output randomness in Brogue’s combat, counters that randomness can lead to rich trade-offs as you juggle one option over another. It’s interesting to choose between making a high-damage but low-hit-chance attack and a definite hit with low damage. The danger of a purely deterministic game, where there’s no random roll and you know what the result will be before you make your decision, is that it can make your decisions obvious and bear no cognitive challenge. That’s certainly true for simple deterministic games like noughts and crosses.
At the other end of the scale, complex deterministic games such as chess or go have vast and rich possibility spaces that make it very hard to weigh up decisions. But with all those interesting choices comes the cost of the fact that they’re incredibly demanding to play. “They’re games of optimisation, deeply analytical,” says Borenstein. He sees output randomness as a way of giving players the chance to make decisions within a range of outcomes, rather than having to decide on just one; to feel their way through the game. “That’s powerful, you relieve the player of the burden of having to do all that computation, right? A lot of players can be paralysed by chess, and also, you don’t have to think systemically. You think through a series of specific states and compare them to find which is best, rather than think of the whole system.”
For Burgun, though, at the worst end of output randomness lie the, as he describes them, “Farmville Skinner box” games, which are specifically designed to string players along with unpredictable reward schedules to eke from them as much playtime – and profit – as possible. He entirely believes that it’s possible to design a game with lifelong appeal that is entirely deterministic, where you can see how every result is determined by your prior decisions. “People say they like that randomness creates surprise, but that’s not the only way we can create surprise and unpredictability,” he says. That’s the project is what he’s working on today, and he sees it as the next evolution in game design.
But while he’s certain that game design dogma will shift in his direction, he also acknowledges that the game business has become heavily entrenched in randomness. The industry has altered course, away from producing single-release linear games and towards games as a service, where randomness drives high engagement while also controlling players’ progress. Destiny 2, which has a business model based on players purchasing seasons of content, couldn’t exist without its random drops to keep them playing, day after day, in hopes of scoring the guns and gear they want. The industry’s desire to keep players drip-feeding money into their games to pay for new content directly challenges Burgun’s dream of producing a self-contained, systemically rich, and evergreen game. “It’s a really hard problem,” he admits. “Difficult to get over.”
It’s true, too, that randomness is also behind the explosive popular growth of games such as Fortnite, where it fuels some of the reason why they’re so good to watch: you don’t know what’s going to happen next. The drama of skin-of-the-teeth escapes, lucky breaks, and playing the odds drives exciting emergent storytelling that really works on YouTube and Twitch. Ninja would be nothing without it. For more evidence of our love of watching randomness play out, look at the view counts on lootbox-opening videos, and Giovanetti says that it’s helped Slay The Spire sell, too. “A YouTuber can put up a video and say, ‘You wouldn’t believe this crazy-rare coincidence of events that happened,’ and it gets a bunch of views as a result, and that gets a good feedback loop that drives interest in the game,” he acknowledges. “Those effects are a big deal.”
Randomness is having its time now because it’s fuelling so much of what’s happening in and around videogames today, from the economics of indie right up to those of the biggest publishers. It’s governing game production, and making streamers’ fortunes. And it’s also having its time now because game designers have learned fascinating, engaging new ways to use it. “I think the state of the art in game design is advancing,” says Walker. ”I think game designers are getting better at harnessing randomness in a way we just didn’t know existed 20 years ago.”
“Game designers have lots of things in their toolkit, but there are few as powerful as randomness,” says Gage. But he sounds a note of caution. Randomness is, after all, deeply involved in the gambling industry. “It’s weird that we have a tool that literally destroys people’s lives. If we ignore that, then the only people who will are those who will use it for the bad stuff. So we have some responsibility for that.”
Designers like Burgun, who question its use in games, are therefore providing a useful counterpoint to randomness’ pervasiveness. But its strengths are already well understood by designers like Gage. “With my games, I really want to give people the experience of having to figure out a system and come up with ideas and test them; to find the joy of critical thinking,” he says. “When I use randomness, it’s in service to making that more fun, more inclusive, more interesting.”
THE DRAMA OF SKIN-OF-THE-TEETH ESCAPES AND PLAYING THE ODDS DRIVES EMERGENT STORYTELLING