Death Wave: Cities: Skylines and Simulatory Failure
by gamer_152
This is Springdale, a Cities: Skylines sandbox, as it existed on 2nd March 2036. It's a sprawling hive of commerce and leisure with a population of 135,000 and has all the organs of a functional metropolis: roads, hospitals, shops, universities, and so on. But that population is about to fall precipitously, and all of Springdale's services will grind to an abrupt halt. This city is about to und...




This is Springdale, a Cities: Skylines sandbox, as it existed on 2nd March 2036. It's a sprawling hive of commerce and leisure with a population of 135,000 and has all the organs of a functional metropolis: roads, hospitals, shops, universities, and so on. But that population is about to fall precipitously, and all of Springdale's services will grind to an abrupt halt. This city is about to undergo what players call a death wave.
On 6th March, a retail worker, Edward Campbell, posts on social media:
"Who should I call if no one shows up to pick up the dead?"
With a single click, we can trace this complaint back to the northeast corner of a district called Prospect Park. Here, a red icon emblazoned with a skull pulses over a residential block. When someone dies in a Skylines city, a nearby cemetery or crematorium should send out a hearse to pick up the body. The red skull symbol shows that the corpse has sat in situ for days, and it's not the only one. On the other side of Springdale, a cadaver in an apartment goes neglected. In a distant suburb, a dead resident rots for the neighbours to smell. These skulls start spreading like flu in winter, popping up all over the map in areas entirely unconnected to each other.
The basic problem is that while the city's morgues can swallow a lot of bodies at a time, they're not made to dispatch whole convoys of hearses at once. Each crematorium has seven cars in the garage, meaning I don't have enough vehicles to transport corpses on this scale. Soon, every block in the region has a few stinking ex-humans sitting in it. Something invisible is culling the population, and it's getting worse. As black vans fill the streets, the traffic slows to a crawl, which only prolongs the collection of the bodies. I demolish shops and homes and construct more morgues in their place, but the citizens are expiring in surging numbers, and the new funeral homes aren't close to keeping up. Worse, every hearse is another four wheels on the street, clogging the concrete arteries.


As the highways and avenues freeze to a standstill, emergency vehicles fail to reach their destinations, and employees struggle to get to work. The population plummets, and everywhere, from hospitals to shops, wave goodbye to the staff that they'd need to keep their heads above water. It incubates all the societal diseases you'd expect. Without adequate healthcare, more people get sick and die. With fewer firefighters, more residents burn to death. With a dwindling police force, more crime occurs.[1] As refuse stops getting picked up, we get pandemics. None of the new dead can be removed from the buildings where they died, and a plague of bodies elsewhere only further spreads germs.
The whole supply chain breaks down. This city is meant to import raw materials for industrial workers to process into goods. Those goods should then get sold by shops. But the materials are not delivered to industrial sectors, fewer secondary labourers are arriving at their jobs, goods are not being driven to retail zones, and retail staff are not showing up to stores. The economy enters a debilitating recession at the same time as governmental costs are rising to get more hearses on the roads. As business revenues shrivel, so does the amount of tax collected, meaning there's less to spend on the city, creating a spiral of decay. Soon, my residents are living in the equivalent of a failed state: A sprawl of endless, impoverished apartment blocks filling with dead bodies and no one coming to help.
Illusions in Media
There are almost no wholly accurate simulations in video games. We're accustomed to programs like "Flight Simulator", "Farming Simulator", or "Cult Simulator", and yet, all these pieces of software employ skeuomorphism in the sensory and mechanical modelling of their subject matter. There is a discrepancy between the verbiage and aesthetic experience of the simulator game and the mechanisms and aesthetic experiences of the real tasks they represent. Additionally, the underlying mathematical and logical systems behind these pieces of software do not mirror reality. Games, like films, are powerful illusions, illusions more convincing than teleporting your card into the middle of an apple or having the Statue of Liberty disappear. Games create the impression of whole worlds, characters, and systems through only the sliver of them they depict.
That hallucination is possible because we infer conclusions from limited data. When you hear the sound of the rubbish truck outside your house, you don't go to investigate and learn what a rubbish truck is and the sound it makes all over again. You access your existing memory, and you fill in the blank of the rubbish truck based on the sounds you remember. If you see someone checking their watch and then running quickly down the street, you don't have to speculate much about their motivation for moving so fast. You fill in the blank from previous experience: they're late for something.


We need this inductive reasoning to operate in our day-to-day life. Otherwise, we'd be researching everything from scratch each time we encountered it. Media exploits the assumptions we make based on context clues. Film creators only compose what sits inside the frame. However, the frame's contents and the audio that cohabits with them encourage us to imagine whole universes outside the frame that the creators have not constructed in detail.
If we hear a scream from behind an on-screen door, we imagine a person back there, terrified. For a more detailed example, consider a film scene in which one character shoots another. An actor can use a prop gun to fire blanks towards another actor, and an effects team can then detonate squibs on the target to create the impression of bullet impacts before the target actor falls and remains perfectly still. The impression given is of a gun that can fire live ammunition, a round that soared through the air, and then reactions of force and lead poisoning in the victim's body, which caused death.
Of course, none of those objects or phenomena are real, but we made assumptions about systems and technologies based on observations. The film creators don't need to add any of the above "real" elements to the scene. What matters is the audience's perception. If a technically realistic inclusion in the art does not change that perception, it's superfluous to requirements. Part of the practice of TV, films, radio, and video games is knowing what you do and do not need to show your audience to get the impressions you intend.
Illusions in Games
Video games are a systemic medium, so we cannot just discuss what sights and sounds are perceptible at any one time. Not that video games are just films with mechanics added, but given that most games have us interact with systems, we should pay attention to what games do and don't leave in the systemic frame. What events or mechanisms within our vision suggest which events and mechanisms outside of our view?


If guests come sauntering through our door in PlateUp!, we will assume they travelled from a home, workplace, or place of leisure to get to our restaurant. We imagine they have lives outside of our business that have earned them the capacity to visit our restaurant and that they will take memories of their meal out into the world with them. This is true within the narrative of the game, but in-engine, the guests are spawned just outside the business and will be deleted shortly after leaving.
Or we can look at what a barrel roll in Microsoft Flight Simulator constitutes. The physics for the manoeuvre come across as shockingly real. We can feel a true-to-life sense of momentum, drag, and gravity on the plane as we turn it through 360 degrees of movement. But however realistic the physics of a game feel, they are never immaculate recreations of real-world physics. A complete simulation would require modelling every quantum particle that makes up the world, as well as all the interactions between those particles. It is not only outside of the scope of real-time processing, it's pointless, and is an excellent example of how more realistic systems do not necessarily enrich the user's experience.
The Illusion Dissapates
Of course, no creator is perfect, and time and resources are never unlimited, so the illusion that media presents us sometimes shatters. In the cases when it does, those hidden elements that create the false reality become apparent to the audience. Sometimes, it's those elements coming to the surface that constitute the fourth wall break. Maybe the viewer of our film can see a camera operator in the corner of a mirror. Perhaps the gun is obviously a cheap prop. The reader of a novel might work out that its timeline is broken so that a character would have to be in two places simultaneously. Here, the strings by which the writer is puppetting their characters become visible.
Just as we must analyse the systemic nature of games when looking at the illusions they create, we must keep the same aspect of them in mind when picking apart how their simulations disintegrate. Because illusions are created both through gameplay systems and technological ones, the shattering of illusions can occur through the failure of either. Technologically, maybe our character falls through the ground, or we see polygons z-fighting. But systemically, maybe someone who is meant to be weak acts as a bullet shield, or an entire city spontaneously drops dead.
The Mechanics of the Death Wave
So, what happened to Springdale? We know the city's flaw can't be too few crematoriums or cemeteries. Our increase in neglected bodies happened very suddenly, with no complaints of a lack of deathcare beforehand. Sure enough, if we check the city's analytics, we can see a spike in deaths. Now, there are a lot of things that can cut a Skyliner's life short: Poor sanitation, noxious pollution, an unsafe water supply, meagre medical facilities, and fires are all killers in waiting. But none of the communication from my citizens nor metrics in the UI report those problems with the city before the death wave.
This is very weird because if no plague or fire spontaneously gripped the city, that means tens of thousands of perfectly healthy citizens dropped dead out of nowhere. It gets stranger. If we look at the statistics behind Springdale, we can see this has happened before, multiple times.
Purple: Population.
Yellow: Births, Red: Deaths, Brown: Influx.The birth rate tises relatively steadily, then suddenly, there is an equivalent sheer drop-off of citizens, and around the same time an influx of citizens in proportional numbers. The menace of it is that every time this reaper, whatever it is, returns, it takes more people than the last time. As the jolts in deaths grow higher on every occasion, so do the spikes in population increase, until now, when deaths finally outweigh births and immigration. That relationship between the population booms and busts is the clue as to what causes the death wave.
Have you worked it out yet? Because I didn't until I started reading the posts of crafty hackers who'd gotten access to the numbers underlying the simulation. I have to give shoutouts to Spec. Tater's thread, The Short Life and Random Death of Cims on the Paradox Forums and to Blake Walsh's guide, Preventing "Not enough workers" and minimising Death Waves, which, more than anything else, formed the basis for my understanding of death waves. The essential piece of information you need is that in the original Skylines, when citizens would immigrate into a residential area, they'd all start at the same age. People generally move into a metropolis that will satisfy their needs. Good public amenities, a place to live, and job opportunities can combine to make your society oh-so enticing for an out-of-towner. Because they'd immigrate at the same age, they'd die at roughly the same age.
That explains the deaths, but not the waves, and why the waves amplify each time. We can, however, explore the causes behind those patterns by looking at how typical player behaviour influences NPC behaviour. Around the dawn of a new city, we're likely to expand it hastily. Building a quiet suburb into a bustling mid-sized town is manageable and unlocks some of those urban essentials like buses, taxis, and plazas. That early growth entails a sudden influx of new residents. Because all these people enter the town simultaneously, they'll die around the same time, so just through the act of expansion, the player is arming a ticking time bomb.


When that bomb goes off, and the population plummets, residential demand rises as homes empty out, and taxes on homes go unpaid without owners in those properties. We may also notice that our population, and likely our economy, are on the decline. So, how can we balloon the citizenry, and how do we usually respond to residential demand? We build more flats and houses. The shelter with the best ratio of nests to price is high-density residential housing, but such housing is more likely to attract young people and is not where citizens generally choose to start families. So, while building skyscrapers for people to live in might bounce your city back to health after a death wave, it doesn't create the conditions in which people might raise the children who would replace them after they die.
Keep in mind that at this point, we don't necessarily know that the reason there's all this residential demand is because of a rush of deaths. We might just be responding to the direct incentives of falling income and population and a rising call to fill housing. But if we're relying on immigrants to fill those empty domiciles and then building whole blocks of shiny new apartments alongside them, we're going to get a bigger population jump than the previous time we expanded our city. So, when the death wave for this generation of immigrants hits down the line, it'll be even more devastating than the last.
If we keep using this same technique of building new housing, especially cheap, low-footprint housing, every time we're low on humans and high on demand, we're going to keep amplifying the death waves. Eventually, a point will come when we can't move new citizens in fast enough to replace the old ones dying off, and the house of cards will topple.
Death Waves as Simulatory Failure
Death waves were a common blight on Skylines games, and part of what makes them interesting is that the player creates and perpetuates these mass extinction events, probably without realising. Why the developers thought citizens should all arrive in their new homes at the same age, we don't know for certain. I'd hazard a guess that the designers biased the systems towards the young so that players would get the most out of all their urbanites. If every alternate citizen out of our intake pipe were a senior, we'd spend more resources taking care of them than we'd get back from their labour. If everyone that enters our apartments comes in as the equivalent of a bright-eyed twenty-year-old, things will go much better for us.


Further, flattening the ages of the immigrants prevents erratic patterns in supply and demand that could make cities unmanageable. If details like whether immigrants will need education, whether they'll add to the city's GDP, or how long they'll be a net drain on resources are pseudo-random, it becomes harder to plan for the future. Ironically, what may be a plan to keep citizens' activity more controllable can make it impossible to work with.
What happened to Springdale was an example of the simulatory failure I talked about in section two. The systems intended to generate an illusion failed. Interrogation into that failure further makes the systems visible, stopping us from suspending our disbelief. We are incentivised towards that interrogation because Cities: Skylines is a game about systems engineering. It is a rough simulation of a city: trains carry passengers to destinations, farms produce food, loans need repayment. Almost everything in the game works, fundamentally, as it does in the real world, and so the game is an exercise of identifying the kind of problems we'd see in the real world (e.g. A lack of demand for products, a district with too little recreational space), and responding with the solutions we'd see in the real world (e.g. Exporting surplus goods, zoning more parks and entertainment districts).
If we could pop the game's hood and see the data that runs it, we'd have a very different view of our city. We'd be able to review the dice rolls and hard maths that underlie what looks, from the air, to be an organic and human series of choices and relations. We'd perceive our city less as a world of cafes and offices, and more as a structure of variables and algorithms. We would also notice all the missing components we'd expect to see in a real city, causing the diegesis to fall away.
The lack of a window into the base logic that drives the systems is, more often than not, no ball and chain on our planning. The underlying logical operations performed by the game's code translate into phenomena we can understand because they correspond to real-world states and events. Those states and events are relayed to us by the UI. Heatmaps highlight where parks are lacking, icons above retailers show where product is backing up, fires appear on the facades of buildings, and so on.


The death wave problem occurred because the game performed corner-cutting logic, resulting in a problem that we couldn't solve through intuition. We couldn't solve it because it didn't emerge from real-world logic and because the UI did not actively express what was going on. Therefore, we had to analyse the problem not as one of a city's population bottoming out but as a software hitch. This is why much of the community research into death waves had to be conducted by spying on the data within the game rather than playing the game as it was shipped. Modders, and eventually the official Skylines team, patched the death waves by making the age at which immigrants arrive in population centres pseudo-random, meaning new arrivals won't later all die at once. The fix was to keep age data as obfuscated as ever but bring the simulation in line with real-world expectations.
This is not the only scenario in which the game does not clearly externalise the information a player needs, but it is the most catastrophic example of it, as well as a highly common one. The logic of immigration and ageing in Cities: Skylines were the mirror that film director missed, the clashing chronology that took the reader out of the moment. Of course, the game's interactive nature makes the simulatory problem manifest very differently than it would in a movie or book. The believability of the world is broken, but the greater casualty is the sense of fairness and control that the game promises.
Lessons Learned
The specific simulatory failure we see in Cities: Skylines teaches us a few different lessons about how this phenomenon often occurs in games specifically and how its knife can be twisted. Firstly, it's useful to note that there would be no issue if all games simulated the real world perfectly, but besides being technologically and developmentally unfeasible, it would often be undesirable from a design standpoint for them to do this. Games simplify how the world works (e.g. A character can put on clothing in a fraction of a second, players can "fast travel" to teleport across the world) to more effectively achieve their aesthetic goals.
That usually means making the game funner and systems easier to understand than the real-world areas of activity they represent. However, by doing so, they violate the concept of simulation. That is permissible if the player is aware of the non-realistic element of the game or if the consequences for not being aware of that element are not significantly negative. However, if the player's success rests on treating the game as a simulation, and they're not aware that the designer may not be running a simulation, that's where trouble starts to arise. To streamline communication and sustain the fiction, the designers only surface the information they believe would be practical for the player to know and which would be emotionally appropriate for the specific game. The developers of Cities: Skylines believed player age to fall outside of that category, so we were left in the dark as to when a death wave was coming or why it was happening.


Secondly, the reason that a death wave is so devastating is because the ageing system is closely connected to a series of other gearboxes that are pivotal in the success or failure of a player. Also, because the ageing system pushes on these other structures with robust force. This will not be the case with every system that undergoes simulatory failure. If a character's face disappears in Assassin's Creed: Unity or their weapons clip into their model in Skyrim, that's not inconsequential from an experiential standpoint. However, it's not like the wider systems of these games are reliant on objects being kept from intersecting or the rendering of faces. Whereas, in Cities: Skylines, everything from the sewage lines to healthcare are leaning on zoning, immigration, and ageing. Therefore, when the player is unable to act in an informed and constructive manner with respect to one pivotal system, it creates a domino effect that fells all of them.
Conclusion
In review, all fiction is fundamentally reliant on simulation. Creators use sensory phenomena that we can perceive to lure us into assumptions about the areas of the world that we can't. Should this sensory level fail to convince us, then we will no longer be able to suspend our disbelief. In video games, this simulation has components of fairness and empowerment and exists on an interactive level. If the systems we interact with do not behave in a logically consistent manner, the metaphorical curtain falls.
Systems sliding out of alignment with our expectations can have dire effects because we rely on seeing real-world elements appear in games as cues for how we should act. The wrong cues can lead us to make catastrophically poor decisions in play without even realising it. In Cities: Skylines, the death wave was triggered by misinformed decisions. We can see that the failure is possible because this is a simulation, because designers hide play-critical information to maintain the simulation, and because the systems these failures occur in are closely related to other key systems in the game. Thanks for reading.
Notes
- In the real world, police are not to crime as firefighters are to fires, but this is the way it works in Cities: Skylines.