How Do You Put the Sim in SimCity?

Ocean Quigley has been at Maxis since 1995. In that time, he's been an artist and art director, and now serves as creative director on the company's newest SimCity title -- which is due to be released next year, simply under the title SimCity.

This game marks a new horizon for the series: For the first time, computing power has increased to the point where the simulation can be built from the ground up, with individual simulation elements (Sim citizens, buildings, vehicles) able to react to one another to produce a full city simulation. Previous games in the series faked granular detail.

As an artist, Quigley wants the game to visually portray information to the player as much as possible -- to instantly represent decisions in the world itself, rather than forcing the player to look at charts, graphs, and menus.

Quigley explains the work he and others at Maxis have done to achieve this, and how the team makes creative decisions that impact not only the simulation, but also bring forth satisfying decisions for players and also accurately reflect the ways in which these systems interact in the real world.

You've talked about tilt-shift as a visual trademark for the new game, and how aesthetics can convey information. How important is it that the aesthetic choices you make in the game also convey information to the player?

Ocean Quigley: Let me answer that a couple different ways. One answer is that in previous games like SimCity 4, for example, we had an aesthetic of piled-on detail, without necessarily giving it meaning or giving it context. So, for this SimCity, basically, the simulation is sophisticated enough and complex enough that if we don't take every moment, if we don't take every opportunity to tell the player what's going on in their city, then, basically, the art's not doing its job. The simulation would be difficult to parse.

So, I've got a big slide, aesthetic commandment for all the visuals in the game: that if we can assign meaning to something, then we assign meaning to it. We really try not to put anything in the game at all that doesn't serve as a UI function to tell the player what's going on in their world.

So, you look at it and you think, "This is a city." You're looking at a city, you're looking at architecture, you're looking at buildings, and all that is true but what you're really doing is you're looking at UI that's telling you the state of the simulation. The UI is just aesthetically stylized to look like the city.

To quickly touch on this idea of meaning, is all the information you convey via the game just about the state of the world? In other words, is all the meaning that comes -- other shades of meaning -- does that all come from the player?

OQ: So, of course the player is going to be projecting their own story and their own imagination on it. I don't know enough about what the player is thinking to anticipate that. All I can do with integrity, or all I can do with legitimacy, is faithfully represent the state of the world at any given moment.

So, for example, if you see a car parked in front of a building, it's because there's somebody inside that building. And if there wasn't anybody inside that building, the car wouldn't be there. Or if the lights are on in the building, it means that the power is on, and that there's somebody inside that building. Or if you see green terrain, it's because that terrain is watered. Or if you see a house with graffiti on it, it's because a crime has occurred there, and so forth. Right?

So, what I've got to drive the meaning in the game is the actions, the behavior, the state of the simulation. That's what has to come out in the surface.

Of course, there are a lot of choices about what the simulator is doing that are happening at a deeper level, and there are aesthetics to that as well. But assuming you're primarily talking about the visual aesthetics, they're there to communicate to a player the state of the world that they're creating.

Is there such a thing as a simulation aesthetic, or an aesthetic of simulation?

OQ: Yeah, yeah. Of course. It's really about the cause-and-effect relationships between things and how you want to bind it together, those causes and effects.

So, for example, we could have people in a house get sick for no reason. They go to the hospital and get cured, but that's kind of unsatisfying. You'd rather bind the fact that they got sick to something that is in principle, and hopefully in practice, understandable and parse-able by the player, right?

If they got sick, maybe they got sick because a patient zero came into the city and carried a disease with them. Or maybe they got sick because they drank polluted water. Or maybe they got sick because they lived downwind from the toxic chemicals of an industrial plant.

And so the simulation aesthetic is about drawing the cause-and-effect relationships between things and giving them integrity the player can understand, and not just doing things as smoke-and-mirrors, but having it be like a watch, where all the gears visibly move the other gears forward and the whole thing has got an integrity -- almost a mechanical integrity of all the pieces stacking into each other and moving each other in a way that the player can understand.

And then, of course, it's up to you as the designer of the game to figure out how you want to make those relationships work. You want to have crime be a function of education, or lack of education, or crime be a function of employment, or do you want crime to be a function of pollution of the environment, do you want crime to be intrinsic to people, do you want tie crime to class? Those are all the sorts of simulation aesthetic decisions that you make as a designer of a simulation.

And then you try to decide if it makes sense, if it has coherence and integrity, and is it something I can express to the player? And so that's kind of what simulation aesthetics is. It's the art's job to express that to the player. Of course, there's a feedback loop there, because, if you want to express something that's so abstract that you can't surface to the player, then there's really not a whole lot of point to do it. So the aesthetics that you make in the simulation are constrained, let's say, and not entirely bound, but constrained, by what you can transparently present to the player.

That opens up a few more questions. First of all, I want to ask: When you come up with these ideas like you talked about, whether you want to tie crime to class, or to education, or whatever, do you do sociological research and then determine this?

OQ: Yes, we read up on it. I mean, we don't do exhaustive academic-level research, but we do due diligence to get a sense of what the commonly accepted reasons -- for something which the causes are as nebulous as that, we'll do proper diligence. We'll read up on it.

And then it also comes down to [the fact that] we have to make decisions... When we make decisions about what affects things, we have to make them in such a way that the player can do something about it, that the player needs to have agency into the simulation.

It's not enough for us to make this clockwork that ticks away out of the player's control. The whole point is that the player can push it, and pull it, and pry it, and transform the state of the simulation. So, another constraint or a big input into the decisions we make is, how is the player going to see it, and what is the player going to get to do about it?

So, for example, something we would not want to bind is crime to social class, because that's outside of the player's control. Sims of different wealth move into your city. So, ignoring the sociological statement we're making with that, but just looking at is as a game mechanic, that'd be a sucky thing to do, because what can the player do about it?

Or, by contrast, if we make the assumption that crime is a consequence of education, then the player can elect to deal with their crime problem either by putting down lots of jails and lots of policing, and deal with it at a symptomatic level, or the player could elect to deal with it by improving the education level of their population. And so that's a much more interesting -- from a simulation aesthetic perspective -- choice that you give the player. The player has agency and multiple ways of dealing with something. Each one of which has its own effects, which are also kind of interesting.

How do you avoid -- or do you avoid -- a right answer to things? Especially because it's a very open-ended game. How do you avoid going down those paths?

OQ: That's really just game design. By figuring out, well, "Why wouldn't I just do this as a player? If it turns out that crime is caused by education, why wouldn't I just invest a whole lot in education, and have education be the right answer?" Education might have a really slow payoff, right? So, it could be that if you got a bad crime problem right now in your city that your city's going to fail before education can do its work. So for all of these components of the game, you want to make them not channels, but surfaces that the player can basically traverse. So you don't want to make them just like a rail the player has to go down. You want to make them a landscape that they can explore.

And there are certainly places on the simulation landscape that might temporarily be the best place to be, but the whole landscape is always switching around. Even with something as simple as education, you maybe can afford to up it at one point, and you'd still very much like to have a little bit later on, but you just can't afford it. You've got to deal with your fire issues, or you've got an economic collapse of whatever sort. So it's a shifting sort of multistate landscape that you're exploring and there's really no -- or we're doing our best to make it so there's no -- "just do this and you'll be fine" kinda answers.

Also you've got the unique position of that your audience is very broad, so you're designing for people who are idealists trying to create their ideal city, to people who are min-maxers.

OQ: Absolutely. People who are basically thinking like states and optimizing the city for, say, the production of ore, or the generation of power, or just the accumulation of money, for example. Yeah, no question.

That's one of the reasons that we have to make the simulation as transparent as possible, so we don't have a lot of hidden secret things that are going on in the game. We've surfaced them all to the player, making them front and center in the art content, so the player can directly see -- for some of this stuff, sure, you have to go to data views, but you can directly see what the state of your city and the consequence of the choice you just made were. You watch it play out immediately or over the course of minutes of gameplay.

When you came up with this concept of making everything transparently visible through just the game graphics, did you go down a lot of dead ends? Did you have to do a lot of testing to get to a point where people could actually perceive and glean that information? Or was it intuitive for you as a creative process?

OQ: We didn't approach it as a monolithic problem. It's more like, every time we wanted to put in some content we asked ourselves, "What is this content trying to say?" Whenever we put something in the game, we never put stuff in the game naively, as it were. No piece of content just goes in because we need some filler. Everything that goes in there goes in there for a reason. And then if there's something that we really need to show the player -- something that the simulation's showing the player -- we'll figure out what kind of content we can use to express that.

So it's a very iterative and incremental process. It didn't just spring forth fully formed as a finished project. It was something where methodically, brick on brick, we put in this content and said, "Well, what's the motivation behind this piece of content? What does the house really need to tell you as a player? What does the moving truck really need to tell you? What does the lawn need to tell you as a player? What does the driveway need to tell you as a player? What does the color of the car need to tell you as a player?"

Literally you go down the list through every piece of content in the game and figure out what it's trying to do. What is the simulation motivation for putting it in front of the player? And if there isn't a simulation motivation for it, either figure out, well, it's an opportunity to bind it to some simulation meaning or don't put it in there.

And so an accumulation of those sorts of decisions mean that, sure, there is stuff that we'd like to represent that we're having a hard time making literally represented in the ordinary realistic view of the city. We're going to have to put some of that stuff in bar graphs that overlay the city. But that does mean there's nothing in there that is superfluous. Everything in there is doing a UI job to tell the player what's going on.

When it comes to designing that, is that something where art and design have to really collaborate, or go back and forth? How does that work?

OQ: That chunk of problem was something [we tackled] when Andrew Willmott and I were first getting this game roughed in and prototyped to think about the stuff we were trying to do with it. We basically made that a rule from the get-go. That was kind of my core problem for the first maybe six, eight months of the project, to figure out what is the visual aesthetic of the game and why, what is it for.

And Andrew Willmott, in addition to writing the simulation, the underlying GlassBox simulation engine, is also a graphics engineer. He got his Ph.D. from Carnegie Mellon in global illumination. So he and I were able, just between the two of us, I as an artist and Andrew as an engineer, were able to kind of have the volley back and forth of, "How about this?" "Well, we can't really do that." "How about this?" "Well, okay." I would mock up and prototype stuff in Maya and in-game and say, "This is what I'm thinking about." He'd say, "Well, hmm, how could we hook that up to the simulation? What would be the mechanism for doing so?" That was just a core part of getting the game rolling.

I mean, the game can look like anything, right? It's SimCity; you can stylize it kind of however you want to. And there has to be some reason for making one aesthetic decision or another, and that is what we sort of settled on as the grounding aesthetic rule that would control everything. We just iterated on it back and forth, and back and forth, and back and forth. And now it's just become how we do things. It's ingrained after that.

In terms of getting the whole team on board with it, that was not just at the lead prototyping level, right? That's what I'm curious about as well.

OQ: Right. The way that we usually work on this stuff is -- for years and years and years I was a production artist. I'm a fairly technical production artist as well. So what I tend to do is -- first I make a prototype of what it's supposed to do, and then I'll build a few examples to work out the bugs and the kinks with it, and then I'll step out a workflow that lets us do that sort of thing.

And then I'll train up a lead artist in the workflow methodology and what the motivation behind that is. And then they take that to their team of artists, and do it at scale. What it kind of means in practice is I go out, and for each bit of content that we want to put in there, think through what it is that it's supposed to do, how we're going to make it work, what its job is, what the workflow for it is and then hand it over to a team of production artists to do it at scale. We've had the luxury of this project of a relatively slow build-up.

Andrew and I started this after we shipped Spore. That's a few years back, so we had the luxury of being a team of three or four people to work through this stuff and accumulate all these answers that we could roll out to the team at scale. When the project got green-lighted we could do it for real with a big team as opposed to, say, our little secret project, our little side project.

I wanted to talk to you about the aesthetic choices conveying some sort of perspective. I remember seeing in the demo, the coal plant, for example. First of all, it's nice looking, but it's also old looking. What are you trying to communicate with that, if anything, meaning-wise? Is there a value judgment there?

OQ: Yeah, there is a little bit of a value judgment. So, with the coal-power plant, it's trying to represent 1950s, 1940s-era technology, right? The base coal power plant takes coal in, cheap coal, burns it, and then generates electricity and lots and lots and lots of pollution, and the process of mining that coal is also pretty filthy. So, the whole thing is kind of a dirty and kind of a primitive machine, a coal-fired power plant.

If you look around we're not really making new coal-fired power plants in the world anymore. Maybe the Chinese are, but by and large in the West we're not. And so I decided that I'd echo the technological level of sophistication of different infrastructure elements you put down in their aesthetics.

So the coal-power plant looks like it's from the 1950s, 1960s, and oil-fired power plant is a slightly more modern image aesthetic and maybe it looks like it's from the '60s or the '70s. The gas-fired power plant, more modern still, looks like it could be from the '80s or the '90s.

Nuclear power plants, well, everyone knows what they look like, right? So we went with the classic big cooling tower ones, and then as you get into more fanciful power sources.

We're probably not going to do fusion. At least, we haven't done it yet, but the more modern power plants are aesthetically more polished and gleaming and modern. And so we're trying to represent the technological level of a given piece of infrastructure in the individual aesthetics of that piece of infrastructure.

And so, similarly, if you have a poorly educated city that's got all these low-tech industrial buildings in it. All those low-tech industrial buildings, they look first half of the 20th century. They look like they're Lowell, Massachusetts in 1890 or 1900. They look like they're old mills and stuff like that.

And then as you increase the education level of your population to, say, middle education level then the industrial building, the industrial construction of your city starts to look more contemporary. Or maybe not quite contemporary, maybe more like 1970s or 1980s. And then as you got a very high-tech population, a very well education population, the industrial buildings that are populating the landscape wind up looking more like Silicon Valley office parks. Even somewhat more futuristic stuff. It'll look like --

Like the projected Apple building that's going to be built? That kinda stuff?

OQ: No, because that one doesn't look industrial enough. That one looks too much like an office park. The ones I'm thinking of are more like big biotech pharmaceutical factories where you see big silver vats of stuff. And you see piping and tubes. You're not exactly sure what's going on there but it doesn't look like it's a steam-powered mill from 1930, whatever it is. So, broadly speaking, that aesthetic of the toys, the poppables you put down, and of the industrial buildings are deliberately contracted to convey to you the technology level that's in them.

And we're not really tracking time in the same way. Like, in SimCity 2000, for example, you always got the fusion-powered plant in 2050, for example. But for us it's more -- you could invest in education, for example, and have a highly educated city and be able to run highly sophisticated power plants and get high-tech industry going in there.

So, you could let your education standards slack and the only new factories that would come into your city would be lower-tech, and then if you get rid of your schools altogether, then your city reverts to something that looks much more like mill-style factories. So it's not a linear march of time that we're binding to the aesthetics. It's just that we're tracking the education level and the technological sophistication of your city, which can go up and down.

There's a certain degree of overt politics which go into the making of a SimCity game, and it predates this game. People have talked about Will Wright's politics as inferred from what the best way to run a SimCity city is. The clearest and most obvious political message that came out of EA's Game Changers event at GDC was about anthropogenic global warming. Do you make these political choices? How do you feel about that kind of issue?

OQ: So, I think that the intent is to make a simulation with enough internal integrity to it that the player can explore those sorts of choices and see what happens. As far as I'm concerned, the goal is to make something that's robust enough that you can push it in all these different directions and get plausible outcomes from it. So, my agenda with this game is absolutely not to make something that's got kind of a pious, holier-than-thou lesson to it. My goal is to try to represent reality with enough fidelity that you can do all sorts of things and there's really no proper way to do it.

You can make a sprawling city, a broad shoulders-style industrial city, like Carl Sandburg's description of Chicago. If it's all factories and industry, and it is making lots of money, and growing and is polluting like crazy and people have a pretty low life expectancy and so forth, and that's perfectly fine. That's exactly as it should be.

Or you can make a city that's just a fully extractive industrial city that's just pulling coal right out of the ground and then burning it, and then sending some coal to all the other people and sending electricity to your neighbors, and so forth. And that's perfectly fine as well.

So that the core agenda that I've got with this is to make a simulation that is a landscape that the player can explore, metaphorical landscape -- you push on it and see what it happens. It's absolutely not to make an ideologically channeled experience where you have to get all pious and make the right environmental choices to win because that would be, frankly, absurd, as a gameplay experience. That becomes an exercise in propaganda, as opposed to giving you a simulation to explore.

Like I mentioned earlier, we can't help but make some choices for the cause-and-effect stuff towards. Like, we have to bind crime to something. People cause crime. Why, you know? We have to do in some cause-and-effect way. So we say, well, "We could make the plausible argument that crime is caused by a combination of unemployment and poor education, for example." Maybe that's not the case. People have argued that crime is cultural -- that crime is caused by snowballing effects of other criminals around them.

There are lots of other arguments that you could make about what causes crime, and so our decision that crime is caused by a combination of unemployment and poor education is ultimately a political assertion, right? But we're making it because the player can do something about it and because it's at least parse-able. It at least makes sense.

But beyond that we are not attempting to encode our ideology into the game and force people to believe what we want them to believe in order to succeed at it. It's a landscape for them to explore. It's a little model world for them to push on and see how it responds. We're not preaching to anybody.

You talked about simulating the Sims, and everything coming from that, rather than necessarily being a top-down simulation.

OQ: Definitely. The bottom-up. It's worth pointing out that it's not just the Sims who are simulation elements. The buildings are simulation elements as well as are, say, the vehicles, as are the map. The important point there is that it's coming up. It's a simulation that's built of interacting parts as opposed to a modeled, top-down simulation.

At the presentation, someone did you ask about Dwarf Fortress and things like that. Has that kind of thinking had a profound effect? And it is an under-the-hood effect, or is it a paradigm shift for how things operate, in terms of the ways the players will perceive really clearly?

OQ: The main thing that it lets us do is describe a much larger simulation landscape for players to explore. If you're doing -- I'm going to call it a "monolithic", or top-down simulation in advance, you define the bounds of that simulation landscape of what the player gets to do up front. You're really not going to do anything that's not pre-defined in that.

But with this approach, essentially an object-based approach, or a bottom-up approach, as you add new components you get to do new things. You as the player get to do new things. So the simulation landscape is potentially unbounded. You just add more components and you get more things.

Lego is a good metaphor for that. So, think of the difference between, like, a jigsaw puzzle, where you've got all those pieces and they snap together in that way and you're kinda done, versus Lego, where you can reconfigure those Lego pieces in a bazillion different ways and new Lego pieces all work with your existing Lego set and extend the things that you get to do with it.

So we've been kinda self-consciously going with this bottom-up, object-based simulation -- not just because it gives you more simulation integrity. Not just because the simulation is actually localized at a sticking place, and each house or each vehicle or each Sim or whatnot. But it's also because it lets you reconfigure it. It lets you add stuff to the simulation by adding new objects. And so I just think that's just a more exciting open-ended way of defining a simulated world. New objects come into the world and the world changes as a consequence.

Is that what inspired you to do it? The sense that it would make the game more open-ended, or more variable?

OQ: It's the combination of having it be more open-ended, so that you combine things and get new behavior, and it's also the desire to have something that's got more close-in integrity. You know, by way of contrast with SimCity 4, because it was a model with top-down simulation, we wanted to tell you details about what was going in a given building or on a given block or on a given street or in a given car. We essentially had to make it up, and we had to do some smoke and mirrors tricks to try and make it so we weren't contradicting ourselves. Because that data wasn't really there.

But with this new simulation -- the simulation's taking place in a building, and there's enough integrity to what's actually going on inside that particular building that no matter how we show you the data, we're not contradicting ourselves, because it's really there. It's like the joke about why it's easier to tell the truth than be a liar. If you're telling the truth, you just tell the story, and tell it from this angle and tell it from that angle and tell it from the other angle, and it all lines up because it's fundamentally the truth. But if you're a liar, you have to keep all your facts straight in your head, right?

So, with these much, much simpler, less-sophisticated simulations that we had to make due to limited CPU resources in the past, we had to struggle to keep our stories straight about what was going on in a given building or on a given block or on a given part of the neighborhood. With this new one,any way that we want to expose the data to you as the player about what's going on in a given area on a given building with a given Sim, with a given car, we're just surfacing what the simulation's actually doing. So our problem is more a UI problem of getting the data to you rather than a game-design problem of, "Well, what do we tell them this time?" If that makes sense to you.

When you make creative decisions on this game is it about creating, as you say, a simulation with integrity and readability, and that will automatically be satisfying to the player? Or is it about player satisfaction and then working backwards from that? How do you do that?

OQ: So, we think about things that would be a lot of fun to do. Like, we know there are things that are fun to do from previous SimCitys. We know it's fun to zone and see stuff come to life. We know it's fun to see cars drive around. There's a bunch of proven, as it were, SimCity satisfactions.

So, for starters, we need to make sure that we hit those satisfactions and we design towards those basic fundamental satisfactions that come with the game. Blowing up buildings, seeing traffic jams form. Solving traffic jams.

And then for the new stuff that we're doing, we think about, well, "This is stuff that we could do that naturally grows out of what the simulation is doing. What would be some fun things to do with this? What would be some satisfying, exciting things to do with it?" That's kind of the simultaneously top-down and bottom-up thing. We get this simulation running and then we iterate on it. "Where is this taking us? This is kinda fun. Wouldn't it be cool if we did this?" Then we kinda bend the simulation towards that.

So, something as complicated to pull off as SimCity is, as necessarily a kind of combination of those bottom-up discovery processes. You do stuff and see what's satisfying and fun and exciting to do. And also, kind of a larger top-down vision of what you might want the player to be able to do and then how you bend the capability of the simulation toward that end. So, it's a "yes, and." It's not the crisp answer you're hoping for but it's too big a project for it to be one or the other.