The ideology hiding in SimCity’s black box
polygon.com
polygon.com
> The art project Magnasanti revealed the contours of the black box when it attempted to “beat” the game that famously doesn’t have a win state [...] Artist Vincent Ocasla accomplished this feat through rigorous trial and error, eventually creating a crime-free city that maintained a stable population of 6 million sims, but only by creating a dystopian nightmare. [...]
> Magnasanti reveals the contours of the black box, showing how citizen health and happiness were ultimately unimportant. What the game considers a successful city doesn’t look at all like one we would actually want to live in.
SimCity presents an open-ended simulation with a wide range of things players can decide to optimize for. An artist once built a city optimizing for population density and low crime to the exclusion of all other factors, and this resulted in a dystopian nightmare where all the simulated citizens were unhealthy and miserable.
But by describing what the artist did as "trying to 'beat' the game", the article pins the blame for the artist's choices on the game itself. Even though SimCity clearly does not think Magnasanti is a good city -- it's constantly flashing up warning messages and graphs showing that all the citizens are miserable and unhealthy! One could just as easily have argued that SimCity's black box is hiding a progressive ideology, one that warns us of the dangers of industrialization and over-policing! (Much like how the board game Monopoly is about the dangers of capitalism.)
In essence, an artist set out to 'beat' SimCity, the artist decided what 'beating' SimCity meant, and the article tricked us into thinking that said something about SimCity.
But try to zone something for mixed use.
SimCity itself tells you that certain things will happen under certain conditions, and it is generally wrong about those predictions. It's not intended to be an accurate simulation; it's intended to be amusing and to conform to people's intuitions about what they want to happen when they plan their city in a certain way. But people play it and feel they're learning something about how cities actually work.
It absolutely is trying to be an accurate simulation, for its time. That was 28 years ago. Do you remember how slow computers used to be? You can't zone for mixed use because of technical limitations. The simulation takes place on a grid; each grid cell is limited in memory by computer technology; therefore the performance requirements of the simulation dictate that only one thing can exist in one place.
It's really not a sleight of hand. The idea that it's natural for it being impossible to have a city with low crime and a high population without making everyone miserable is absolutely ideology, and it's just your own ideology that perceives this as somehow a neutral fact.
Its not progressive to say that without over-policing high crime is inevitable, or that industrialization means we have no choice but to be miserable either. Those are solidly reactionary beliefs.
There is a lot of intentional and unintentional emergent states in games, but trying to push narration of something being a win state as if that was a minor jump is not doing that critical reading any favours.
As far as myself, I did take care not to assume that large stable crime free happy cities are inherently desirable in my above comment. Simply the fact that they are impossible still communicates ideology, whether you think they are desirable or not.
In this case whether a stable large city is to aim for or not in real life or in the game is a bit irrelevant, what's important is that the game's black box allows large, crime free cities if and only if they are a dystopian nightmare. This is how the state was defined and it's backed into the game by design. It's worth noting that someone made it like this and behind that ideology there's a message being passed on through the game.
By testing to find the win conditions in a certain scenario you basically discover what the designer of the system considered to be necessary to win. The coder who wrote the code responsible for this (or their manager) either put their own ideology in code, or put what they believed to be the most accepted by the target audience.
Almost any such creation embodies either the ideology of the creator, or that of the consumer. Sometimes it's boring and sometimes it's interesting to let the genie out of the black box.
you realize, that there are none in SimCity.
You can't make (usable) chairs without conforming space to human physiology. The ideology is in making a chair, not in what chair one makes.
You can however put different ideologies into city builder and get different, perfectly fine city builders.
I think you have this backwards. The ideology in SimCity is only a tiny reflection of broader ideologies that pervade almost every aspect of modern/urban life and ought to be much more interesting due to their much larger impact. If anything is interesting in SimCity's ideology, it is only what it says about ideology outside of SimCity, as the game itself is trival.
You could probably create a nice big city that had low crime without dystopian policing, but you might need to build a public park in space that could otherwise be used to pack in a few more apartments.
The idea that what people are willing to suffer and keep the city in a state of self-sufficiency has nothing to do with health, income equality, and so on, but only about crime and the amount of commerces and parks, is also ideological.
The assumption that the highest density city that people are willing to live in means that reducing crime is more important than having healthy and happy people is ideological.
The idea that this is an expected result is also ideological.
I'd say the population limit stems from the game's map size limit. The largest SimCity 3000 map is 32 square kilometers, and Magnasanti packs 6 million residents in there. That's 187,000 people per square kilometer.
For comparison, Manhattan is one of the densest areas in the US, and it has 28,000 residents per square kilometer. Manila, the densest city in the world, has 41,000 per square km.
I am unsure I would describe a city seven times denser than Manhattan as "reasonable", at least in a game designed to replicate cities as they currently exist in the real world, and I am also unsure I would describe the decision to limit the map size in a video game as "ideological".
To note, 187k people per sq. km is not actually unrealistic - there are neighborhoods in the world with this level of density such as Ayal Nasir in Dubai, and even some with this level of density and the same order of magnitude in population such as the Imbaba neighborhood of Cairo in Egypt. But this is a detail and mostly besides the point.
Put it that way - in SimCity as long as crime is down and some other factors are adequate, they won't leave even if they are miserable. But if crime increases, then they leave, even if they are in good health and make a nice income nonetheless.
Doesn't seem that unreasonable, actually. High-income people leave crime-ridden areas, happens approximately everywhere
Also, do you claim to know exactly how and why crime happens?
Yes, but since you've been snarky now I won't tell you.
I think Simcity makes a political comment, and there was literally no way for the game to avoid that.
No, it "absolutely" does not. It could simply be a shortcoming of the developer's reach. You try to simulate a large city and get it a stable utopia! Do I have to remind you that this is something we've yet to accomplish in the real world, either. You expect a single game developer to do it successfully? Will Wright is (presumably) not a mayor, city planner, nor politician; he's a gamedev. Talented and famous, I'll grant you, but not infallible. And we're talking about a game he made 28 years ago from today, at the age of 33.
SimCity 2000 isn't your typical illusion as many games are, either. It's trying to be a fairly accurate simulation of its time. Of course it embeds its creators biases, but it's not really trying to "convey a deep message" using an elaborate facade, like some games are, or theatre, for example. It's a sandbox.
In that game, people care more about still pretty low levels of crime than they care about their health, having a nice income, and so on.
And in addition, that game has crime as something that's simply immutable and that you remove by totalitarian policing.
I never said the game conveyed a deep message. I said that it conveyed the ideology of the creator. Those are very different things - often the second is accidental.
Video games are art. Adding crime in the way it was implemented is an artistic decision. Making crime dependent on policing instead of other, more realistic factors is also an artistic decision (and it's not true that police is the only way to handle crime from a gameplay perspective either). Balancing what exactly the sims are concerned about is also an artistic decisions.
Artistic decisions, especially when they seem obvious, are often ideologically motivated without someone even realising it, and this is both what the writer of Polygon was touching on as well as explicitly on of the things the author of Magnasantis was targetting.
I love that you called it a sleight of hand because that's exactly what it is; just by assuming someone could beat the game, the authors have misunderstood Sim City.
Sure, not all the design assumptions align quite so neatly with American and right wing sterotypes: suburbs are really inefficient to provide services for and Sims aren't at all bothered about religion in many versions. But they are designed limitations rather than the product of players' imagination.
Because it's a 28-year-old game with the computational limitations of the time. It does not simulate people walking. It does not simulate multiple types of buildings existing in the same plot. It does not simulate building styles (I think?). And it has nothing to do with ideology. No-one could make such a simulation in 1993.
Buddy,
One example: "Scientific simulations, like those used to model climate change, don’t keep their formulas hidden in a black box because replicability is a big part of the scientific method."
But the black box nature of climate models was the core issue in one of the biggest disputes between sceptics and researchers of all time. It went to court and the researcher preferred to lose the suit than reveal his model (Michael Mann). The COVID model from Imperial College that justified lockdowns had to be extracted from the university via FOIA requests which were stonewalled for months. The scientist who wrote it openly claimed in interviews nobody except him understood the model equations, which were all entirely undocumented. Probably the author doesn't know these things, but it shows a rather naive misunderstanding of the true nature of science. Black box models that encode ideological beliefs are common throughout serious research.
The comment about how SimCity doesn't simulate race or gender is also rather ideological. The belief that race and gender are central to everything is both quite new, very much a view of the left and totally USA centric, but SimCity was designed to sell to international markets too. "Cities are built on history, politics, and racial bias" is the sort of comment that gives away the article for what it is: a complaint that a a popular video game wasn't woke. We learn more about Polygon's ideology than that of Will Wright.
Which court case was that?
SimCity does have an ideology, and it does have something to say with how it has weighed it's measurements of citizen satisfaction, but the author seems to fail at capturing that ideology and instead just talks about how minmaxing is bad.
The rest of the article is interesting. There's plenty of people out there that don't even think about how the rules of the game frames the discussions you can have about it. It's just kind of a bad example.
Ho hum. Who lacks ideology? Well, the dead, for starters.
My observation is of the "who smelt it, dealt it" variety: those offering the heaviest charges of bias and ideology are communicating the depth of their own.
Their job is to make something feel enough like reality to be plausible, not model real world accuracy to 5%. Just like they may cheat on what warped room reflections are rendered in a mirror you see on the wall, they're going to come up with cheats and simplifications to things about how cities, governments, people act and behave to get it "real enough" to be believable.
And just like you have to take with a grain of salt that you cannot jump through a portal and land on the other side of a building, or that women in virtual bars are interested in going on a crime spree, or that flying a plane in GTA IV makes you capable of being a pilot in real life, you have to do the same when a game tells you your choice of a property tax has caused your residents to rebel and lower your approval rating by 8 polling points.
Maybe the problem is that some game experiences are obviously not from the real physical world, while these are just enough on the edge of believable to be dangerously convincing. Perhaps the people who want to ensure realism could try coding up an engaging game, complete with IRS forms 8960 and 8825 to get people's behavior right...
I did find though that the game forces the player to limit the availability of education to provide a sufficient supply of low education workers, which I didn’t find very fun.
Jokes aside, maybe it's hard to balance. I mean how do you treat the zone as?
In all seriousness, I think the main issue is that it would require more computation. C:S already runs into computation issues because unlike SimCity, which more or less just calculated a bunch of statistics and then simulated things on screen to reflect that, the former actually treats an individual resident as an agent and does all the calculations for each resident it simulates. IIRC there is a hard limit of 65,536 (a 16-bit unsigned int) agents being simulated so that they don't kill the average consumer computer trying to calculate Djikstra's for more than that amount of agents.
Games with macroeconomic simulations (SimCity 4 for example) use sampling-based methods to get statistics about the city and then plug in some simplified equations to simulate things like demand and growth. Your simulations generally behave in the way you want to, with all your assumptions and beliefs about how economics should run. The problem with this is that it fails to capture the nuance of local economic effects, which leads to your game becoming too simplistic (exemplified by the Magnasanti project which is referenced in the article)
Games with microeconomic simulations (ex. Cities Skylines) treats every citizen as an independent agent, with their own jobs, families, and movement for work/shopping/travel. The problem with this is that you have less control of how the game will actually play like, and that it's incredibly hard to balance the game (Cities Skylines has a reputation for their systems being "too easy" except for transportation, but I imagine that they made it that way because any further fundamental economic tweaks that make the game harder might actually break the whole game). There's also the technical issue of simulating too many agents.
NewCity, one of the more recent city-builder games on early access, is trying to achieve some balance between the two, and it's interesting to read their design decisions (https://lonepine.io/2021/01/15/macroeconomics.html). It seems like a really hard gamedev design problem though.
SimCity (2013) is the first SimCity game to simulate individual agents. And it takes some pretty ginormous liberties there, too. Citizens, for example, return to any random available house after their workday. That is, they switch homes every day. It's less human city and more... anthill.
- In C:S agents are mostly used to allocate work/home slots, and to depict traffic. SC used agents for everything, including simulating how individual poops would reach a sewer outlet, or how power or water reach homes. So they used computationally expensive algorithms to simulate things that no one cared about actually seeing like that.
- Because of the very heavy computation, to make the load workable they limited city size to 2km by 2 km, or 4 sq km. To put this in perspective, the land area of Manhattan is 59 sq km. SC:2013's scale would be as if the city you built would only get to be built as big as Central Park (3.5 sq km). This is fine if you want to simulate a medieval town, maybe, but probably not a modern city with trains and highways and stadiums and things that take up a lot of space.
The site contained a pretty amazingly comprehensive detailing of the game's mechanics and various algorithms scattered throughout the articles, to the point where it seemed to me like it'd be possible to implement a lot of the game's engine using it.
Some good examples of the more detailed articles:
The economy https://community.simtropolis.com/omnibus/other-games/the-ec...
Land value specifics https://community.simtropolis.com/omnibus/other-games/land-v...
Traffic and transportation specifics https://community.simtropolis.com/omnibus/other-games/traffi...
Zone development rules https://community.simtropolis.com/omnibus/other-games/zone-d...
I wonder what kind of simulation model would produce a game where the 20 Minute Neighborhood design is optimal.
Sim City seems to capture this perfectly through the third-person view of a city. There's a difference between being happy and wanting changes that you believe will make you happy, and a computer simulation has captured this distinction by showing both a citizen's actual happiness separate from their desires.
Obviously this isn't an exact parallel to the real world as desires can have an impact on one's happiness as well. However, it asks the important question of whether or not we should be designing cities to match desires, or to match what city planners believe would make people the happiest.
There was a rule, that each neighborhood had to have access to all the services (healthcare, kindergardens, schools, "stores" etc) within a walking distance - and you can really see it in the neighborhoods built in that period.
Case in point: I played a lot of SC3000, and that game doesn’t automatically replace utility buildings when they “age out” of their in-game design lifespan. As soon as any plopped utility building hits year N+1 it’s immediately destroyed and replaced with a nonfunctional ruined (“Defunct”) version of the tile. That feature is annoying enough when a bunch of your water pumping stations all blow up at once and cause water service disruptions, but it means all in-game nuclear plants will explode and spread radiation like that exactly 71 years after being plopped unless you manually bulldoze and rebuild them. Doing it earlier than 70 years is a waste of money since the new plant costs the same amount regardless, so players will always be tempted to let it go as long as possible and only replace at the last minute.
On top of that, any building (plopped or otherwise) that catches fire and is allowed to burn will also become destroyed/defunct. If your city’s total electrical usage exceeds any one power plant’s electrical capacity, that power plant will catch on fire. If it’s a nuclear plant and you don’t put the fire out, the plant is destroyed and spreads radiation.
If you have multiple nuclear plants and accidentally let one of them get to be 71 years old, it will explode and irradiate its surroundings as mentioned. The sudden drop in total electrical capacity will almost certainly overload the other plants, then they will catch fire and may also explode and irradiate their surroundings.
Maxis plz
Any old city of Europe, and many of South America, the Mideast and Asia, beats the generic, parking-lot dominated mid-sized American city in quality of Life.
NY, SF, and a select few others illustrate the old adage about Americans always doing the right thing, after exhausting all other options.
But the vast majority, operating with what planners seem to perceive as endless scrolling in all all directions, are dystopian.
Sim City replicates that scenario. Neither its tools, nor depth of simulation, allow for any emergent behaviour that doesn't fit the predefined mould (pun not intended, but appreciated).
It’s the grid, the strict zoning, and the coal or nuclear plant built as close to the neighboring city as possible. If your map starts with any hilly terrain, you’ll better bulldoze it now, or you’ll regret it later. If there’s any river, it better fit the grid.
Good science is when you get out more than you put into you model. The same is true of simulations, and SimCity never did.
> I recall something about city planners using Sim City back in the day as part of their education, which horrified Will Wright since in his words (paraphrasing) “sim city is just cellular automata with some differential equations sprinkled on top.”
It was never intended to model the development of actual cities.
You can start a city from scratch on easy mode. When I haven't played for a while, it's common for me to get tripped up by something stupid and have things go sideways.
Fordlandia and other real world efforts to build a city from scratch in the middle of nowhere are generally infamous for having gone sideways. It's quite hard to create a functional city from scratch with all the right moving parts and once things go sideways, it's hard to come up with the money to fix the problems, assuming you can even identify them.
You overbuild X and it has knock-on effects of Y and Z that didn't predict or forgot was likely and now you've painted yourself into a corner.
Real cities tend to start as a nexus on some transit hub, such as a port at the mouth of a river where the terrain creates a natural port and you don't need much to make it a functional commercial port. Quite a few American cities can trace their origins to being a rail stop on a rail line.
Towns can get born or die when new highways get built, depending on whether they are on the new highway or whether the new highway causes traffic to pass them by. Historic small towns on Route 66 were born from traffic passing through them and many of them "died" (or greatly shrank) when Route 66 was replaced was modern limited access highways.
If you don't have enough water and electricity, no amount of tax breaks or pro-business government policies will grow your industrial base. Add water, electricity, trains and a port and now you have enormous demand for industry.
Of course it doesn't model "actual cities." Models always have their limitations and it's helpful to understand what those are, but you have start somewhere for developing mental models for the complexity of all the moving parts of a living, breathing modern city.
Our failure to wrap our minds around that seems to me to be a root cause of a lot of the small town blight and related social issues happening in America today. (This may also be a factor in problems in other countries, but I don't feel qualified to speak to that.)
It's a 28-year-old game. These are technical limitations of computers of the time. You try to simulate an entire city on 1993 hardware while catering to every 2020 social justice warriors' desire.
And this is why at least some countries make this kind of black boxing illegal.