[In this Gamasutra analysis piece, UK-based game researcher Mitu Khandaker examines video games and chance, examining why "our brain loves to seek patterns", and how games should treat randomness.
Though this is only the second installment of this series it seems promising that this is turning out to be a set of meditations on being human
After all, there are many ways in which games can be socially beneficial, some in more obvious, practical sorts of ways than others, but the most important of these, in my estimation, is the potential that games have to teach us about ourselves; about the wonder and the complication of what it is to be human.
Humans are pretty amazing. To think of our fantastic, collective achievements as a species is humbling, and awe-inspiring. The fact that we have a space-probe
16 hours of light-travel away from Earth.
That we can create virtual galaxies populated by thousands of remotely connected human intelligences. These are extraordinary feats in engineering, and scientific understanding. We humans are pretty special.
Despite all this, though, we humans are actually quite bad
at thinking about things clearly and logically, which is a tendency that results in all kinds of problems. This fantastic list of cognitive biases
serves to highlight some of these embarrassing, very human tendencies we have to think about the world incorrectly.
There are a number of biases to do specifically with how we estimate probability and chance. We're not very good at it. Our brain loves to seek patterns, even where none actually exist.
This was, no doubt, developed for evolutionary reasons, because completing certain patterns would keeping us away from potential sources of danger, or bring us food time after time. But it also means that we really don't understand - or trust
- true randomness (or, stochasticity, if we're being fancy. Which we are
It is well-known that people attempting to either identify - or even mimic - random data (such as the outcome of a set of coin flips, or dice rolls) will almost always get it wrong. Where they see 'clusters' of patterns, they will tend to discard this as intentional, not realizing that true randomness actually tends to contain lots of clusters.
Indeed, one symptom of our mistrust of true randomness can be seen in terms of the iTunes 'shuffle' feature, which has lead to a myriad of common conspiracy theories about how random it actually is; with listeners seeing 'patterns' (e.g. some artists playing more than others, repeats, etc) in a way that seems to make us want to mistrust iTunes. Or, when someone exclaims "What are the chances!" when you are wearing the same obscure t-shirt; actually, the chances tend to be quite good.
To understand what all this has to do with games, though, let's go back to the conception of our theories about randomness. There was a 16th century Italian mathematician, named Girolamo Cardano, who is best known for being the father of algebra. Cardano was, however, also a notorious gambler. Though it may seem trivial to us now, Cardano was the first to provide a systematic analysis of probability - and worked out that there is a one in six chance of rolling a certain number on a six-sided die, etc. He even published a book called Liber de Ludo Aleae
("Book on Games of Chance"), which, interestingly also contained a section on methods of effective cheating.
Essentially, then, probability theory itself was borne out of games, and the link between the two remains important. Martin Hollis, lead designer of GoldenEye 007
and director of Perfect Dark
, suggests that "the brain is engineered to spot patterns, and find them even in noise; that's how casinos manage to exploit people so successfully, via your cognitive biases."
However, in contrast to traditional games, randomness in video games is generally considered to be problematic, and even to be avoided. Sid Meier, in his 2009 GDC keynote on player psychology, noted
: "Any kind of randomness needs to be treated with a lot of care. Whenever something random happens to the player, paranoia sets in. The player feels like the computer rolled that random number just to be difficult." He asserts that "small doses of randomness, however can be helpful -- just make sure that it seems fair to the player so he doesn't feel cheated."
Perhaps, once again, this difficulty with true randomness in video games is down to our cognitive biases; in board games and tabletop games alike, it is we
who roll our own dice. Therefore, we perhaps feel some sense of control over the outcome; as though, by rolling the dice, we are the masters of cold, impartial randomness. However, with video games, the computer rolls our dice on our behalves. We are reminded that randomness is out of our hands; a randomness which we had no effect on in the first place.
Hollis reminds us that what we think of as 'random' in video games, often is not. He observes: "Tetris
doesn't deliver the bricks in a completely random order; it shuffles ahead a set of bricks. If I recall rightly, about four sets of seven. What that does is it evens out the distribution, so it means you won't get a load of S or Z bricks. You can't possibly get more than say, seven in a row. These kind of tricks don't make games worse, they put a lot of work into that and they believe that it makes the game better. I'm inclined to think that it does. But, it does isolate you from the brutality of true randomness."
To a human mind that's precisely what randomness is: brutal
. Our flashy pattern completion engines are always looking for ways to soften its impact, which has ramifications for how we create and understand video games, even if we aren't consciously aware of it. For games to be fun, we often want them to reduce randomness.
Hollis goes on to say, "I'll bet a lot of people do believe that Tetris
does deliver random bricks; I did too for a long time. It's very hard to distinguish if it does make any difference, it's a kind of subliminal difference, I think."
Subliminal differences could, in fact, make all the difference in how we feel about video games. Luke Dicken, a PhD researcher, is aiming to enable AI to react to dynamic environments "in a way that makes sense probabilistically." He, like Hollis, suggested that "No matter what you do, the player is going to think it's more complicated than it is - it's sort of the same for randomness."
One of the projects Dicken is currently supervising deals with Ms. Pacman
; and he mentions that while Pacman
is deterministic, Ms. Pacman
is not. In the former, the ghosts follow a determined path, so the player may memorize the optimal way to play Pacman
. The latter, however, features randomly moving ghosts, and so the player's inherent reaction is tested. It's a significant difference for how the game must be understood by players. Ms. Pacman
is the harsher game experience.
How we interpret that harshness when we find it in games is important, and has two major implications: one about fate, and one about skill. Our tendency to find patterns means that we can be quick to jump to superstitious conclusions, and ascribe things to 'fate', or 'the will of God'. As Greg Costikyan noted in his 2009 GDC Austin talk on randomness
, in early cultures, or, indeed, for many people now, a random test is seen as "divinatory." The Romans would play games to test their favor with the gods, and even now, it is difficult not to think of "luck" when one is doing either particularly badly, or particularly well, at rolling dice. The idea might have no basis in science, but it has huge claim over the way we think about everyday life.
That said, the Will Of The Gods doesn't generally play too much role in how we understand ourselves today. Ostensibly we live today in a largely meritocratic sort of culture, wherein how far we rise is related to how skilled we are, how hard we work, how smart we are (the Dunning Kruger effect
notwithstanding). The writer Alain de Botton points out
that in England, in the Middle Ages, a very poor person was described as "an unfortunate."
Literally, de Botton says, "somebody who had not been blessed by fortune." By contrast, these days, someone who is not very successful may be unkindly referred to as "a loser." De Botton states: "There is a real difference between an unfortunate and a loser. And that shows 400 years of evolution in society, and our belief in who is responsible for our lives. It's no longer the gods, it's us. We're in the driving seat."
Of course, each of us have our own mental model of how the world works; this could be informed by religious (or secular) doctrine, be drawn from experience, or from the media, though most likely bits of all of these. Our mental models of the world also govern whether one believes in fate, or destiny, or whether we exist in a meritocracy within a cold, indifferent universe. How we understand chance, luck, or stochasticity is going to be defined by that view of the world. For many of us cognitive biases will dominate, and will therefore give us a skewed view of the influence of chance on our lives.
Like Cardano with his dice games, video games are potentially superb tools for exploring randomness; and, even for dispelling our cognitive biases. Could they serve to model to teach us about the true nature of randomness? After all, video games, as simulations, embody particular models of how the (game) world works; they represent a particular rule set.
This, combined with player agency, means that the meaning-making potential of games is particularly salient to understanding randomness. In other words: games have a chance to model randomness, or fortune in a way that will then affect our own model of how randomness should be understood. Games have the potential to persuade the player through interacting with their systems; by allowing them to explore the possibility space.
Hollis argues that game designers express their understanding of the world in the models they create for their games, saying that "I believe that, in a strong sense, all of the game design comes from the personality of the creators." How randomness is articulated in a game is therefore going to be related to how the designer actually perceives randomness, and how they want it to be perceived.
Could video games, then, bring about a sort of scientific, secular awe for the way in which when coincidences do occur, explaining they are not for some magical purpose, but rational, and make sense; but, that they are still wonderful? If a designer wanted to improve human thinking, and to counteract the cognitive biases highlighted at the start of this article, then it might be possible to create games where all the probability was "on the surface". To expose stochasticity; to leverage randomness as a theme itself. These might teach us, for example, that there are no coincidences, but that the mechanistic explanation was just as thrilling or extraordinary as any notion of fate or meaning.
Of course, video games are systems that can manipulate both randomness and deterministic meaning in a single experience. As such they preserve something about the human mind that isn't available to computational analysis or pure superstition, which is the ability to weave between purely logical understanding and creative imagination.
As Hollis observes: "Magical thinking is important for creativity, and the mind is an incredible parallel computer that's able to see links between things before you even realise it. So there's definitely a connection with creativity there, and that's important and exciting, but you have to keep your feet on the ground and be a scientist as well. A good balance is important in life as always."
A good balance will accept that there is some value in our human biases. Value such as the delight of serendipity, where, again, thanks to our pattern-seeking brains, we find something favourable in amongst all the randomness. On the topic of serendipity, Hollis remarks: "there's a wonderful feeling; the magical feeling of synchronicity, and how wonderful and marvellous the world is, and surely there's some force, some agency out there that helped to make this happen."
In the real world, there probably isn't. But in the game world? Well, that's up for the game designer to decide.
[Mitu Khandaker is a games PhD researcher, and wearer of several other hats, all of which you can keep track of at Mitu.nu (where you'll also find her other writing). She has a few nascent game development projects currently in the works, and is planning a regular podcast to accompany Gambrian Explosion. You can contact her at mitu AT mitu.nu or follow her on Twitter.]