Thinking About People: Designing Games for Social Simulation

This is a quick write-up of my notes from my GDC 2015 design-track talk, Thinking About People: Designing Games for Social Simulation. Essentially, the talk is a call to arms for thinking about the importance of social simulation in games, as well as some reflection on the value of social simulation, through a survey of existing games which have adopted various elements of such mechanics. Here it is:

First, I’d like to make clear what this talk isn’t, so you can make your escape early if you need to. So this isn't a talk about using AI methods to create believable characters. There is a track for that here at GDC. That’s not to say though that there won’t be some overlap, or things to help you if your stated aim is to create “believable” characters. Besides, one of the things I’d like to argue is that the ability to interact with believable characters isn’t strictly an AI challenge, it’s a design discipline with lots of different facets to it.

What I’d like to do is reflect briefly on what has impeded the evolution of social simulation, as opposed to other forms of simulation. After all, we have refined and iterated on things like platformers and shooters, but the extent to which we elevate social interaction mechanics in games is limited by comparison.

I also want to look at the elements that are essential to think about when you’re trying to design a good social simulation, and we’ll take a look how those things have worked n existing games, both AAA and indie.

We’ll also look at how modelling social systems in your games can maybe support and enhance the other structures in your game. So what can social simulation mechanics contribute to games which aren’t specifically about social interaction?

Lastly, this is really a call to arms, to get you thinking about this stuff, and to encourage you to think about why social simulation is valuable and worth pursuing yourself if you aren’t already.

[Note: this is the point of the talk at which I reflected on the weirdness of having delivered such a talk on the day that the closure of Maxis was announced. I made the audience 'pour one out'.]

So we’re talking about games about people, right? This is the goal of social simulation. And the term “games about people” was coined by none other than Chris Crawford.

Hopefully some, if not most, of you in the room will know what this is. It’s a shot from Chris Crawford’s infamous Dragon Speech, delivered at GDC – or CGDC as it was known then – in 1992. This was a fantastic talk (in every sense of the word) about pursuing his dream of games as meaningful expression, of games about the human condition. To him, this dream took the shape of a great and terrible dragon; one that he vowed one day to slay.

For Crawford, pursuing the realisation of that dream meant solving the challenges of interactive storytelling, telling truly emergent, yet thoughtful and engaging stories about human beings. He felt that the games industry was not going in that direction, and so, at the end of the talk he announced his departure from the industry by charging out of the conference room, holding his sword aloft in the name of truth, and beauty, and art.

He felt we would get there by making games about people, not things. And declared “people games” as a category for exploration – games about the way that people behave, their interactions with one another.

So let’s take a moment to define this fuzzy-sounding idea, “social simulation”. I think when we hear this phrase we sort of vaguely know we’re talking about human interaction, but how do we define this as a term of art?

I’m defining social simulation as anything that allows social interactions with or between NPCs to meaningfully affect the outcome of a situation. This is an inclusive definition of course, and that is on purpose. Remember, this isn’t just about AI, and social simulation can serve a far broader range of purposes than we may think.

Having started with that broad definition, we can start identifying the different ways in which we can approach the problem of social simulation, how we can approach making games about people. We can think of this a spectrum between emergent, autonomous behaviour, and authored, branching stories.

Each of these approaches has strengths and weaknesses, so it is not a value judgement. As designers, we may be particularly more interested in one part of the spectrum over another, but this is entirely a matter of our own artistic goals.

In my own work, I’m really interested in this end of the spectrum, but I think any game that intends to convey something about social interaction can be a valuable addition to the landscape.

Social mechanics do not have to be particularly complex to be able to express ideas, or end up with fun emergent results. We can look at GDC: The Game, a little text-based game by Jim Munroe made in honour of GDC 2009. You play as an attendee who can chat to other developers in the halls, and you can talk to them about ideas you’re interested in. Some of these ideas your character knows about already, and some of them you can learn about by going to sessions. Or you can try to pretend you know what you’re talking about, and hope no-one figures it out and judges you. And the point is you’re trying to get a development team together of people who share your interests and ideas. It goes to show that these simple mechanics can work well to convey the notion of ideas as currency and the virality of ideas.

In fact, the beginnings of social simulation were also simple. One person wholeheartedly also at the ‘autonomous behaviour’ end of the spectrum is Chris Crawford. His game Gossip , which was released in 1983, was arguably the first game about people. And more than that, it was a game about how much people liked each other, and how that affected this tiny social network of sorts.

So as the player you’d choose one of seven characters to phone, and when they answer, you’d then choose who you’re going to gossip about, together with one of five different facial animations to reflect what you think of that person, from strongly positive, to strongly negative. Then, the NPC would reply with their opinion of that person too. Each NPC would like people who shared their opinions of others, were positively influenced by their friends’ opinions, and negatively influenced by their enemies’ opinions. Each character in the game was one node of this web of ‘springs’, trying to reduce the tension around them.

Gossip was a game intending to, obviously, simulate the dynamics of gossiping and how that works, so it was this little experiment which tried to tackle a subset of that much larger, complex behaviour, by reducing it down to what people said about how they feel about each other, and the knock-on effect that has. Sadly, as Chris Crawford has recounted in his books and talks, it was published by Atari in 1983, the year the American video game market crashed, so it seems few people ever had the chance to actually play. However, for Crawford, Gossip was primarily a design exercise, intended to demonstrate that social interactions could be modelled in a game, and also, by his own admission, it was obviously quite simple. In fact, he called it the Pong of people-games.

But at the same time, we might say that it works – the limited interaction in the game at least speaks to something about gossiping – these are shallow beings that only know how to talk about other people, and nothing else.

And more recently, under our inclusive definition, there are all kinds of games modelling social phenomena that we can think about, which, like Gossip, choose to simulate a subset of human behaviour rather than trying to do everything:

So let’s look at what has hindered the exploration of social simulation mechanics in games, particularly those on the autonomous behaviour end of the spectrum:

There’s the argument that simulating the ways in which people behave is much harder than simulating one object colliding with another, and that is certainly true, though I’ve just talked about the potential of even basic kinds of social simulation.

Part of that is the problem of feedback, and communicating to the player: it’s much easier to represent feedback for object collisions – your bullet hits someone, they fall over and die. Or, you jump from one platform to another. But, when what you’re communicating are thoughts and feelings, that presents a challenge.

Also, emergent simulations of any kind don’t always guarantee an interesting result. This is simply the nature of chaos. One of the advantages of telling a branching story is that you can ensure that all possible outcomes are interesting, because that’s how you’ve written them. This has the potential to detract from the characters’ autonomy, but we’ll see that these aren’t necessarily mutually exclusive.

Now, Chris Hecker’s Photoshop Challenge (which I was familiar with thanks to this Sims 3 talk by Richard Evans) is more of a tools problem than an implementation problem. This is the idea that even though AI is a design discipline, it’s much more difficult to create AI tools that are as intuitive for designers as Photoshop is for artists, because AI can’t be easily broken down into compositional elements the way textures or polygon models can.

It’s also important to recognize the cultural factors that have influenced game AI research. Because games have historically catered to a very specific audience, many of the biggest leaps in game AI have focused on pathfinding through complex 3D environments and detailed behaviour trees for combat.

What makes for good social simulation? Well, I’d like to propose 6 important considerations for when we’re designing social simulations, which hopefully helps with that.

I’m going to talk about why these are things to consider how you’re going to tackle them if you’re designing social simulations, and I’ll try to show these at work in a few different games.

Understanding and managing this tradeoff between autonomy and authored story is very important, and ultimately depends upon what you’re trying to achieve.

The systems-driven games at one end of the spectrum have all the strengths that emergence affords: the delight of the unexpected, that serendipitous feeling of order coming out of chaos, of something in a game having happened for you, and only for you – those “tiny moments of awe” as I think of them. Emergent systems, with their capacity to evolve and surprise, are also infinitely replayable. No situation will ever play out precisely the same way twice.

At the other end of the spectrum we find the emotional depth and nuance that only well-crafted writing can provide. A human being, with all their range of experience, can tell a story with layers of nuance and understanding that no autonomous agent or emergent system can match. That is a uniquely human strength.

And there’s a lot of room for variation, and all kinds of games along this spectrum:

So that's a good overview of possibilities for social simulation – barring a few outliers (hello, Dwarf Fortress!) – but let’s get a little more specific now.

If we think about the autonomous behaviour end of the spectrum, and something like The Sims, the beauty of playing the game comes directly from the emergent interactions between your Sims. They may surprise and delight you with things you didn’t expect them to do. There’s the fun of getting to know your Sims and their individual personalities, to work out what they’re like. It’s fun to oversee their myriad little stories play out while we watch, exerting just enough control, but not necessarily too much.

So in terms of the overarching model of how the game and the interactions within it works, the world is made up of objects – that includes chairs, tables, TVs, beds, books, showers, espresso machines. And each of these objects have specific affordances for a Sim, expressed to the player through a contextual menu. So a TV is for watching, a book is for reading, a bed is for napping, or sleeping, and as of the later games, a Sim can do multiple of these interactions at once. All the Sims themselves are objects with affordances too, so Sims can interact with other Sims.

In Sims 4, Sims’ feelings toward each other have separate values for friendship and romance, and different combinations of these will produce different interactions. Sims also have emotions that modify their interactions, and their emotions can also be affected by memories of past interactions. All of these systems eventually layer on top of each other to give each Sim a unique personality.

Really, the fundamental unit of the Sims’ behaviour is the ways in which people and objects interact. Obviously, this does not encompass all of human nature, but this is an example of expression through limitation. Governing behaviour this way is completely appropriate for a series which essentially centres on the acquisition of objects. But the beauty in this design is in all the different kinds of play that this gives rise to – the types of emergent interactions that occur between the Sim-objects and the world-objects, or the Sims with each other.

Sometimes, it can arise in things like this, and there is a glorious story to be told when it does. This is a photo originally posted on the Sims Gone Wrong tumblr, which amongst various admittedly hilarious graphical glitches, also posts some wonderful emergent moments from the game.

Also, the agency granted to the player through these sims with all these possible different interactions means that lots of people often play subversively – I know the The Sims has often been likened back to Victorian dolls house play, in which girls could express their autonomy by messing with dolls and having them break from norms where they themselves could not. Mary Flanagan talks about this in her book Critical Play.

So, we can see how there’s this playful possibility space where NPCs behave autonomously, like dolls come to life. Or, as plenty of people have described the Sims, like an ant farm.

That’s one end of that spectrum - that is, autonomous behaviour. What about if we think about the other end, authored branching?

Consider something like the excellent Walking Dead series, where your interactions with other characters follow pre-authored paths, which fan out and then come back together. And these choices are about much more nuanced personal interactions than there are between Sims. Of course, there’s a lot to be said for the setting itself, but much of the dramatic tone is imposed through purposefully limited interaction. The developers have stated that it was designed around a philosophy of “what would you do with the time you have?”

Of course, given the post-apocalyptic setting, it’s a game crafted to create a sense of relentless bleakness, of making choices which may be morally dubious, and about doubting our dealings with other people. And the game doesn’t make any judgement about these dealings, there’s no morality meter, nothing to tell you right from wrong – nothing that is, except the player’s subjective feelings.

At one end of this spectrum we find the delight that comes from watching and interacting with emergent systems. And at the other end is our emotional attachment to well-written characters, the people we care about because a writer meant for us to care about them.

I don’t think we need to claim that any point along this continuum is more valuable than another – it depends so much on the kinds of truths we’re trying to communicate to the player.

And communication is a key ingredient of what makes a social simulation work. Even a system magically imbued with emergent nuance wouldn’t have much of an impact if it didn’t try to help the player understand it.

Blood and Laurels is a game which handles communication really well. So, it’s built on the Versu engine by Richard Evans and Emily Short, which I think must be one of the premiere examples that I’m aware of really balancing emergence with pre-authored events.

In Blood & Laurels you play as a poet named Marcus in ancient Rome, whose scheming patron Artus wants to become the Emperor, and sends you off to ask the oracles if it would be possible – and from there you get embroiled in conspiracies and all sorts. All of the characters are autonomous agents, which act depending on the context. So if it’s a specifically authored event that all the characters are at a dinner, there’s a number of social things they can do that would be things that could happen at a dinner, – like eating food, leaving the table, chatting, and so on. Some of these things are socially appropriate and some are inappropriate. Meanwhile, they also all have judgements of each other and evaluations of each other’s actions. They are also able to do multiple things at once depending on what the context is. They gossip, show off and flirt, they are autonomous agents with defined beliefs and desires which are what let them weigh up the consequences of many possible actions before deciding how to behave. Also, even though there are a limited number of outcomes in terms of the shape of the narrative, your moods and relationships with the characters, and between the characters, is variable, and its this, and the decisions you make about how to deal with characters, which really makes the story.

Playing Blood and Laurels feels very much about the journey, rather than the destination. It’s about the interrelationships between the characters, and brilliantly weaves together authored narrative with autonomously behaving characters. It brilliantly addresses managing the trade-off. And it communicates those interrelationships well too. I think there’s an inclination when we’re thinking about designing autonomous, believable characters for our interactions with them to feel natural and unmediated – but Blood and Laurels recognises the value of being able to click on a character to explicitly see their current state of mind.

It might seem counterintuitive, but more mediation is good, communication is good, otherwise things can be frustrating or confusing. It’s also tempting to think that this might apply only to games with abstract interfaces, like text-based games or those without reasonable animation budgets...

...but, The Walking Dead games do this too, even though there’s the implied effect of your interactions through facial animations and such, It also communicates to player when their actions have an implied significance, like telling them a character will remember a decision that they’ve made. It makes clear to them where they supposedly have some agency.

Communication is massively important, it removes the ambiguity of otherwise unmediated interactions, it helps clarify to the player where they have agency, and that in turn is hugely important for them to understand the particular social world you’re communicating.

Also worth noting though that despite all that, characters will still always be subject to the way we read them, to the attitudes we bring to them. We relate to autonomous and authored and actual people alike. At both ends of the NPC autonomy to authored spectrum, and no matter how you communicate to the player, we bring our own meaning.

Tomodachi Life is a game which is very much just about people and their interactions – there is pretty much nothing else in the game. In some ways it is a true ant farm game. You can use your own Mii and import those of your friends or of some of your favourite celebrities as I did (pictured above is one of my islanders, Biiyonce, who has, of her own accord, fallen for Kanyiizy, which is very much outside the canon, of course).

In this game, your Miis become Tomodachis, meaning friend in Japanese. You level up each of these characters whenever their happiness bars are maxed out. You eat or play your way to the next level of apparent status, and your happiness is highest until the moment you reach it, then you become empty again. So it’s systemically its kind of about the unattainable pursuit of happiness and existential emptiness.

But it’s okay! Because you have friends to help you through it. It would have been easy for Nintendo to design the game such that you could directly manipulate the ways in which the Tomodachis interact with each other, and have them be directly instrumental in raising each other’s happiness. But it doesn’t do that. Instead their interactions are a bit weird and unpredictable. You watch on in amusement as their behaviors play out. This is the kind of chaotic emergence that it’s lovely to watch unfold, sometimes. The game operates in real-time, so it lends itself to very short play sessions throughout the day, checking in on how your Tomodachis are getting on with each other. They don’t let you direct them, but they will ask you for your opinion sometimes on what they should think about another one of the characters – they treat you as a confidant. I suppose you’re a friend to them, too.

And they can get into romantic relationships, get engaged or married – though it’s sadly relevant to note that the system in place here is exclusively heterosexual – the game expresses a heteronormative outlook through that limitation in the characters’ behavior.

Of course, that limitation also limits the kinds of outcomes possible. A nice counter-point to this is Dragon Age: Inquisition, whose characters aren’t all exclusively hetereosexual, and with who, in most cases, you can romance via the game’s ‘Approval’ system, where character’s agreeing or disagreeing with your actions will raise your approval with them, potentially to romantic levels.

Another counterpoint to that is Kitty Powers’ Matchmaker, which is primarily a dating simulator and puzzle game, it encompasses elements of social simulation – you play as someone running a dating agency, with the responsibility of pairing up not only compatible clients, but also guiding your clients through their dates. You help them through what to talk about, or in some situations, leave them to talk through hypothetical situations and how they’d handle them, depending on their personality types. It’s a game to which inclusiveness is essential, as you navigate the dating lives of clients based on their interests, their various sexual orientations, and their personality types. The simulation becomes one which communicates various truths, both about the universality of love and of the inevitability of awkward first dates.

I’ll talk briefly about a game I made, called Redshirt, because that was one in which diversity and inclusiveness was very much central to the game, by design too, but so was expression through limitation. I’ve unticked some of the other boxes, because I don’t think Redshirt was particularly successful along the first two axes for a number of reasons, though I’ve left the ‘treating people as ends rather than means’ box unchecked by design. And that’s precisely because Redshirt was a simulation of this parody social network, populated with a cast of autonomous NPCs that the player could interact with in order to climb the social ladder. By satirising social networks and science fiction tropes through systemic commentary, the game intends to confront the player with their own behaviour, even when that behaviour sees people end up being basically self-serving and horrible. I rather like the IGN review which summed it up like this: “Because it reduces human behavior into only what’s documented online, Redshirt forces you to treat relationships and hobbies as manipulable, disposable, and only useful as far as they can improve your character's position.” 

Redshirt explicitly limits the system – it’s a game about ambition, whether that’s related to money, your career, or social ambition. And to reflect that world, the NPCs behaviour was made up of traits which reflect that fairly cynical world.

NPCs are formed of these 5 public axes which the player can mess around with to generate different personalities at character creation – these remain static - and 5 other hidden axes of ‘moods’ or current personality-related states – so things like confidence, happiness, sociability, and so on. These combinations hopefully lead to some different kinds of personalities.

The possibility space of action  in the game involves deciding between various Spacebook actions, which are limiting - all behaviour is directed through that lens. And amidst all the dystopian social climbing, I saw dealing with diversity as central to the game.

The first kind is simply diversity of types of personalities, but, also, a diverse range of species, some of whom for which the play experience if different – for example, playing as the race of green alien space babes yields an uncomfortable experience where characters with the ‘bigoted’ attribute will constantly send leering messages, raising the strength of your relationship as they see it (since relationships in the game are bidirectional), but also making the player character unhappy in the process. But other than that, identity is fluid in Redshirt – the player defines, through their actions, who they are, since who you are is what is represented on social media.

Redshirt was an attempt, albeit perhaps only a partially successful one, to pair social commentary with emergent gameplay, though the balance of achieving consistently interesting emergent outcomes wasn’t quite there, and it suffered some problems with communicating to the player what was going on.

What about games which aren’t explicitly about people, but still offer some element of social simulation amongst the rest of the games’ features? After all, games can feature people and the social world in interesting ways without needing to centralize those themes.

For example, social simulation can help support the construction of your world, offering an additional way of engaging with it, which a larger part of the audience will immediately ‘get’. All people have some kind of existing mental model of how the social world works as they see it, and that’s something they can bring to the game – it means that interacting with the game world doesn’t exclusively require an existing game literacy.

These next games are presented in ascending order of what I find personally interesting in terms of implementation (because this is my talk).

Persona 4 Golden is a JRPG, which incorporates social interactions and ties them directly to the combat ability of both the player and the NPCs in your team. You, the player character are a teenage boy, newly moved from the city to live and live with your police detective uncle and your little cousin in a town that suddenly gets besieged with murder and disappearances as soon as you arrive. As you investigate what’s going on with the help of your new school friends, there’s a whole host of characters around you to meet and interact with at various stages, and through the various scripted dialogues you learn about and help these people through their own various personal struggles.

The game features a system called Social Links though, in which you maintain a certain ‘rank’ with another character, or even with a group of characters, and you basically earn points towards attaining these ranks, which you get when you choose to spend time with them, and also depending on the dialogue options you choose. This rank, in turn, affects the level, skills, and abilities of the player character and or your teammates, if they are the ones with whom you’re increasing your rank. Or, affects the bonuses you receive when you’re creating new Persona’s, which are the magical avatars for yourself that you use in combat. There’s a strong incentive to spend time with other characters, and to build up ever-closer friendships with them, with the aim of increasing your Social Link rank with them, and therefore enhance your combat abilities for crawling those dungeons which will help you through the story. So there's a nice coupling and balance of social interaction to combat abilities, both of them working to support your progress through the game. The downside of course, is that it doesn’t quite tick the box of treating people as ends rather than means, in that case since your relationships are a way to raise your combat stats – but, there are enough charming branching interactions that it works.

Skyrim is an example of a game that uses interactions between NPCs to create a sense of persistent conflict in the world. NPCs can belong to one or more factions, or guilds, and what faction they’re a member of defines their combat reaction, both towards the player and towards each other. But NPCs individual attributes contribute towards this as well – they have various attributes along the axes of Aggression, Assistance, Confidence, and Morality, and each of these have only 2-4 different values. Aggression and assistance values contribute to their tendency to engage in combat with each other or with the player, and so the ways in these and the other attributes which they overlap gives the impression of a lot of variety in terms of attitudes towards conflict.

This leads to emergent situations that communicate what the game is about. Wandering through the wilderness, you may come across a fight between Imperial soldiers and Stormcloaks, and you’ll know that it happened because of the way those factions feel about each other, and not because of a scripted event that a designer placed there for your benefit. This creates a convincing sense of a world in conflict with itself, which is in keeping with the story of the game.

Shadow of Mordor is one of my favourite implementations of secondary social simulation of recent times, precisely because it’s a relatively simple system which communicates its purpose without needing to be massively complex. You play as a Dúnedain whose family has been slaughtered by Uruks, and finds himself unable to move on into death since he’s basically possessed by an elven ghost – but it’s basically a game about exacting your revenge on these Uruks. Move about the missions in this open world, there’s this Nemesis system, through which you can view the ranks of the orcs you’ve met or have information on, generally through brutally interrogating another orc.

Now, this may be slightly cheating on my own definition of social interaction, because pretty much the only means of interaction between the player and the orcs, and the orcs between themselves is one of brutal slaughter. Because the orcs vie for position amongst each other – that is, through killing each other to move up the ranks. That is basically the extent of the social interactions between them.

While there are pre-authored orcs around pertaining to specific missions, if any random orc underling kills you, a name and attributes get generated for them, and they join the ranks of the nemesis system.

Problematic Tolkienesque racial essentialism aside, because they’re orcs, it’s kind of believable that their motivations are one-track. If they were humans (unless you were framing it as a commentary on social climbing), they’d be much less believable. So there’s a nice synchronicity between the design, and the narrative wrapper, and the code. Everything is complementary. Your interactions with them mirror their bloodthirsty cutthroat world, in which there is seemingly no hope for forgiveness or mercy.

The result: Amid a franchise which massively 'others' orcs, I found it strangely humanizing. You care about orc drama.

Crusader Kings II is a big and complex deep historical strategy game which also uses elements of social simulation to create an emergent narrative layer. In terms of the balance between autonomous and authored content, Paradox have talked about really embracing “Chaos as powerful creative force” – but it’s a chaos which is underpinned well with all the drama of the strategy game, and interesting things happening from just the sheer number of autonomous agents working at the same time.

You play as a King or other member of the nobility with your various attributes. And at the same time, there are tonnes and tonnes of autonomous agents with distinct personalities and opinions of each other working at any given time. And it’s the emergent interactions between all these characters that give rise to interesting things happening. Since the point of the game is to have your bloodline thrive – which means that it really nicely interweaves a game about people with a game about historical warfare. Characters grow old, have children, and die, and their children carry on, with their own distinct personalities and traits. And they also all have their own motivational drives, their own individual goals and reasons for doing things. NPCs will strive towards goals, and they’ll compete with you and the other characters for power.

There are 5 hidden underlying personality parameters of rationality, greed, zeal, honor and ambition, and these add up in different ways to drive the character’s behaviour. So a zealous character will be very religious and prone to starting religious wars, and so on. These attributes support the particular model of the world.

So while at the same time you can read Crusader Kings II as this serious strategy game about historical warfare, you can also see it as a game about all this interpersonal drama between the nobility, and their heirs. Interestingly Paradox have said that they’ve noticed far more women playing Crusader Kings II than their previous strategy titles, which is awesome.

It's a game designed around philosophy of: “The story of a nation can never be as engaging on a personal level as that of characters.”

And thanks to all of this interesting, emergent stories about characters, players love to talk about their experiences on the forums, writing up particular campaigns as novellas or short stories.

So, we've looked at all these factors for consideration in designing social simulation.

And we come to the final factor: what is the extent to which we make it the focus of our game, use it as a form of commentary on social interaction and the human condition. We can use it to enable subversive play, to surprise the player and perhaps offer them opportunities to surprise themselves.

Or, anything else! We can use it to support a completely different set of core mechanics, to show the impact of human relationships on other parts of life. Most often we do this to make conflict and combat more interesting, but what other possibilities are there?

On a purely practical level, games about social interaction are so much more replayable, whether they offer new emergent situations with each playthrough or allow players to try different branches of an authored story.

We’ve looked at how social simulation can help support the construction of your world, offering this additional model of engaging with it, which a larger part of the audience will immediately ‘get’, because it’s not something that requires game literacy per se, and they can bring their existing perspectives and mental model of social interaction to it. Human interaction is a system that most people are already familiar with, and we should take advantage of that.

I think it’s a really salient concern for games that we need to foster wider public understanding of systems literacy in general, of what it is to interact with a system, and unpack the ways it works, and make sense of what it’s expressing. Perhaps modelling interactions after social simulations can go some way towards fostering that kind of understanding.

And to support all of that, if there is one crucial thing that I could get you to take away today, I think of the 5 factors to think about, it’s communicating to the player what is going on – even when it feels counterintuitive. It’s so tempting to want to make an experience feel more natural or unmediated – but really, mediation is really important to not only create an experience that is engaging and not frustrating, so that a player understands the systems within.

And really, that is the virtue of social simulation – the ability to present us with systems that bring the way we think about people into sharp relief, whether by challenging those ideas or just by playing with them.

That’s how simulation conveys meaning in general - the things we leave out are often more important than the things we include, we think through the gaps in the simulation. Especially if we invite players to bring with them their mental model of how people work.

So how do we slay the dragon? How do we, through making games about people, achieve a vision of games as artistic endeavour?

I think we began our fight with the dragon a long time ago – and we will forever be approaching our distant, asymptotic victory. The dragon will continue to weaken, but it will never completely succumb to us, whether or not we choose games as our weapon. That is part of the human endeavour, that is what creating art is all about.

One way, one of so many different ways to continue fighting the dragon is to use our limitations to our advantage. We can’t make endlessly interesting, nuanced characters who behave and interact in limitless ways, maybe not without solving Hard AI – but maybe that’s not the desirable goal. Instead, how can we best focus our implementations of social simulation, to say something about the world?

More open-ended systems would definitely give rise to much more possibility, but at the same time, carefully constructed limited systems, still have so much to say about people. Both types of things, and everything in between is important.

As a final note, thinking about people and designing social simulations results in games with a wider demographic appeal – just like how Paradox noticed more women playing Crusader Kings II than their previous titles, giving more people a chance to play with their existing ideas about the social world beyond the confines of usual games literacy is compelling.

And, if we’re interested in more nuanced emergent outcomes, more diverse types of characters with diverse personalities are crucial. There’s also the potential for games highlighting more types of lived experience for players to explore, potential for us as developers to think about the kinds of meaningful stories that can give rise to. There is so much about our complex social world to explore through games –all the way from the nuances of romantic relationships to the deep, multilayered systems of social privilege in all its many forms and with all its intersecting axes.

But ultimately, I think the pursuit of social simulation in games is not only beneficial to players. I think an industry that thinks more closely about people and their interpersonal interactions is also an industry that is more invested in empathy. We need more games that think about people.