The Rhetorical Effects of System Design, And Toilets

A friend of mine tells a story about Ang Lee at a Q&A session after the release of Lee's Hulk. One questioner asked this: when the Hulk hulks out, his body expands and all his clothes are torn to rags. So how come his pants stay on?

Ang Lee, the story goes, spent some time thinking about the question, and then he looked the questioner in the eye and then offered this explanation: "Because if his pants came off, you'd see his cock."

It's a good line with a real point. The Hulk is a mainstream comic-book character, so you are never going to see under his pants. To all intents and purposes, there is nothing under his pants. This is one of the natural elisions of drama, like the way that toilet breaks don't happen in films. They don't usually happen in games, either, unless a henchman is conveniently distracted so you can assassinate him.

There are exceptions - like human-scale 24/7 simulations where you watch little digital people do their thing throughout the day. If there wasn't a toilet in a Sims house, you'd notice, so toilets there are, and Sims have needs. Even more so in something like Prison Architect, a sim game where you're building and managing a prison. Toilets are important in prisons, so in Prison Architect you have to plan plumbing, put a toilet in every cell, watch for prisoners Shawshanking their way out.

"Of course every simulation has to leave something out. In fact, every simulation has to leave most things out. The world is a lot more complicated than anything we can fit in software, especially on an indie budget"

Of course every simulation has to leave something out. In fact, every simulation has to leave most things out. The world is a lot more complicated than anything we can fit in software, especially on an indie budget. So Prison Architect has toilets, but no wildlife. Meanwhile, RimWorld, a sim game where you build and manage a colony on an alien world, has wildlife but no toilets.

It's a slightly odd omission in a game with a lot of realistic features. Food rots, at a speed affected by temperature; floors become dirty; colonist injuries are modelled down to the level of ear and finger location. But toilets, and indeed water management generally, aren't a thing. You won't find latrine trenches in RimWorld. You can die of heatstroke, but not thirst.

And that's fine. It's caused some opinions in the RimWorld community, but the game is made by one person pursuing a personal vision, and that one person - Tynan Sylvester - has pointed out that with limited dev time, any feature comes at the cost of any other feature. The guy doesn't want to put water management in his game, so RimWorld has no water management.

But more controversially, RimWorld also has no bisexual men - although it does have gay men and women, and bi women. When Claudia Lo wrote on this last week, the current culture war in gaming immediately polarised opinions: Sylvester was a bigot, Sylvester was a scapegoat, Lo had done good investigative journalism, RPS had run a hit piece.

(Interlude! There are a lot of claims and counterclaims about other aspects of the code, and about Sylvester's motives and politics. This isn't a piece about Sylvester, and I don't want to get bogged down in that, but I know if I don't express an opinion it'll sound like I'm taking a side anyway, so here goes: I think Lo's piece was interesting; I think it was a mistake to exclude bi men from RimWorld, and I've got lost in the detail of the rest; I think that Sylvester has the right to make whatever kind of game he chooses, and to make his own mistakes; I imagine he's had a bad week. End interlude!)

What caught my attention is that Lo demonstrated the point with an analysis of the game's source code. This isn't unique, but it's still unusual, and it got me thinking about the extra difficulties of cultural commentary and criticism when it comes to game mechanics - and especially to simulations.

If Sylvester had written a book that featured no bisexual men, it probably wouldn't have caused a fuss. Representation is (rightly) a hot topic, in film and books as in games, but not every cast of characters in every work can satisfy every expectation about diversity.

But when you write code, you make a categorical assertion, whether you want to or not. There will never be bisexual men in (unmodded, unpatched) RimWorld. Even if Sylvester adds them later, no-one's going to forget he made that assertion. If he'd written a book and added a bisexual man in the sequel, the same thing wouldn't have been true. Narrative thrives on subtlety and ambiguity. Code is rarely subtle or ambiguous.

Or, to come back to my initial point about the Hulk's junk: we can't say whether it exists in the film. Eric Bana presumably possesses junk, but we can't definitively say whether the Hulk does. In a game, though, the question can be answered. Either the clothes are just part of the model and there's nothing underneath, or the clothes are modelled separately and the Hulk is junked or junkless. If the Hulk is junkless and female characters in the same game have breasts faithfully rendered under their clothes, then the devs might be faced with awkward questions about priorities.

"Everyone has unexamined, implicit assumptions, areas of constructive ambiguity, areas of legitimate uncertainty. But those implicit assumptions become explicit in code"

And here's the rub. Lo's piece says: 'code is never neutral'. That's true, and that's a problem. Everyone has unexamined, implicit assumptions, areas of constructive ambiguity, areas of legitimate uncertainty. But those implicit assumptions become explicit in code, and moving from an implicit to an explicit assumption can be tricky - especially when people are careless or prejudiced, but even when everyone involved has done their best to do the right thing. Here's a carefully fictitious example.

Let's say you're making London Marathon Simulator 2017. You've done your research, so you're aware that eleven out of the last fourteen (male) winners of the London Marathon have been Kenyan. And the other three have been Ethiopian. What are you going to do with that information?

You could do nothing at all with it. Individual differences are far more important than nationality. There are plenty of Kenyans who'll never run a marathon. But it could be reasonably argued that you were erasing the achievements of black athletes, especially if you were a white developer.

Or you could give some nationalities an advantage, in which case it could be reasonably argued that you were baking racist assumptions into your game. Are you comfortable perpetuating the stereotype that black people are naturally athletic? Or the stereotype that East Asian people aren't?

This isn't a complaint about cultural hypersensitivity. I don't know if these objections would be correct, but they wouldn't be obviously unreasonable. The problem is, again, that code isn't neutral. If you made a film about the London Marathon, you'd probably want to think about the issue, and you might or might not feature a black athlete winning the race, but you wouldn't be perceived as making a definitive statement about whether Kenyans are fundamentally better runners.

Back to a real-world example. This one's more subtle, but very topical. Cliff Harris, as Positech, has released three games in the Democracy series - abstract, high-level but quite complex simulations where the player governs a country, nudging the simulation by altering policies. The simulations necessarily make a lot of assumptions about how societies and economies work.

Some of these are very uncontroversial - for example, that better technology investment means more productivity. But the simulation also assumes, for other example, that government-mandated maternity leave lowers productivity at the national level. I happen to disagree, and I know people who would disagree much more forcefully, but it's not an obviously ridiculous point of view. Perhaps Harris put it in because his research suggested it was true, or because it seemed like common sense, or to make decisions harder for the player. I have no idea, and I don't think we can legitimately deduce anything about his political views from the game.

There are economic effects from immigration in Democracy 3, both good and bad. The game assumes (as I understand it) that too much immigration is bad for the economy and that too little immigration is also bad for the economy. This is pretty mainstream thinking, but exactly what constitutes 'too much' and 'too little' is the subject of intense and often ill-natured debate.

And of course this is where it gets really complicated. Some facts are more negotiable than others. Very few people would deny that bisexual men exist. Most people would probably be prepared to agree now that, no, Brexit won't give the NHS an extra £350m funding a week. But even if you think, for instance, that immigration is fundamentally a Good Thing or a Bad Thing, then as soon as a system tries to codify exactly how much of a Good Thing it is and what its effects are, someone somewhere is going to disagree.

"And most game designers won't tune those numbers to reflect the findings of their favourite think tank. They'll tune them to create an interesting game design experience"

And most game designers won't tune those numbers to reflect the findings of their favourite think tank. They'll tune them to create an interesting game design experience. That's their responsibility. On the other hand, they're also human beings and members of society, and that carries its own responsibilities.

Democracy 3 hasn't caused much of a fuss. 'Indie Developer Suggests Middle-of-the-Road, Mildly Contentious Technical Assumptions Buried Deep In A Game Consisting Mostly Of Arrows, Boxes and Text' is not a great clickbait headline. But RimWorld caused a fuss, and a few years back, so did Prison Architect.

Back in 2014, Prison Architect was still in (paid) alpha. Paolo Pedercini, who releases games as Mollleindustria, wrote a detailed critique of the game, and the ways in which he found its representation of prison life problematic [http://kotaku.com/what-to-do-with-prison-architect-a-video-game-about-b-1505204131]: that it was supposed to be a fairly serious game but that it (for instance) overrepresented prison riots and legitimised prison labour. Mark Morris and Chris Delay from Introversion recorded a thoughtful half-hour video response.

So first of all, bloody hell, I wish this happened more often. Critic writes well-researched but sharp piece about a game, devs provide a substantial and good-humoured response: great! But I don't think it's remotely fair to criticise any dev who doesn't do this - games are creative works, people inevitably take critiques personally, time is short, devs aren't always temperamentally suited, etc. etc.

And second, this is another wrinkle: code in Early Access is still code. It still isn't neutral, and it still expresses assumptions. But it may be very temporary, and it's even less likely to make a coherent point about reality.

For instance: I mentioned that Pedercini pointed out that riots in prison are rare and serious, and usually aren't outbreaks of random violence. None of these things are true in Prison Architect. Morris and Delay explained that (a) games tend to converge to combat but also (b) in Early Access, less sophisticated AI meant it happened too often; and (c) they had deliberately amped up the frequency of violence so that players could alpha-test the prison guards' squad-based combat behaviour.

I think Prison Architect is a well-crafted work that intelligently navigates difficult territory between good game design and political awareness. But at the same time, Introversion had alpha-released a simulation on a serious topic, in which prisoners behaved like stereotypical delinquents in extreme law-and-order conservative rhetoric. At which point do the assumptions in their non-neutral code become political statements, not experiments? When the game hits 1.0? When they cease development? Never, because it's only a game?

One last example. Prison Architect added female prisoners in an update. Female prisoners sometimes have babies to look after, and male prisoners don't. This straightforwardly reflects contemporary UK reality. Prisoners who give birth, and mothers of children younger than eighteen months, can look after their children in prison, but fathers can't.

Female prisoners in Prison Architect also require better standards of hygiene, and more family contact, than male ones. Is this sexist? I don't think so, and my straw poll of female friends didn't find anyone who thinks so. But you might, and I wouldn't think your opinion was outrageous. It's a value judgement, not a clear-cut case.

"Simulations are a recent development. All our habits and intellectual tools rely on clear lines between reality and fiction. Simulations certainly aren't reality, but they don't look like the fiction we're used to, either"

If you did think it was sexist, would you think it was less sexist if I suggested the increased hygiene need was in part intended to model menstruation? Or would you think that was more sexist because it doesn't take account of transwomen?

Is it sexist for a sim game which models toilets, plumbing, and differences between men and women - as many do - not to model menstruation? Or would it be more sexist to do so, since if it had any game-significant consequences, female characters would probably operate under a disadvantage? In a game, all else being equal, you might take a female soldier off the barricades if a tool-tip tells you that she's suffering a weapon accuracy penalty because of cramps. But in the real world, it is outrageous and reprehensible that some employers discriminate against women because they might menstruate at inconvenient times. It's a live issue, not a point of theoretical anatomy.

Simulations are a recent development. All our habits and intellectual tools rely on clear lines between reality and fiction. Simulations certainly aren't reality, but they don't look like the fiction we're used to, either. As game criticism and audience reactions become more sophisticated, we're going to grapple with more and more of these problems.

Well that's a pickle, isn't it? Do you have any useful suggestions, Kennedy?

It would be silly to suggest that code - game mechanics - should be immune to criticism. Games are cultural artefacts, and that's what criticism is for. Difficult questions are interesting questions.

But it would be rash of us to assume that all assertions in code are simple expressions of a developer's opinions. (And all the examples in this piece have been indie works from small development teams! I don't even know where you would start with a big team on a shared code base.)

Both Lo and Pedercini made points along these lines: that simulations are necessarily simplifications, that code isn't final. That didn't stop a fight from kicking off in the comments.

So I propose a simple habit, as a patch for our outdated critical tools: when we're looking at the rhetorical effects of game mechanics, we should mentally insert an 'If...' or 'For the sake of argument...' before anything that looks like an assumption in the mechanics. 'If this were how immigration worked.' 'If we made this assumption about gender difference being irrelevant.' 'For the sake of argument, let's say that being Kenyan is a statistically significant advantage for athletes.' 'Call it poetic license, but I'm going to assume that Hulk dresses to the left'.

'For the sake of argument' doesn't mean every opinion gets a pass, any more than 'only joking!'. We can still discuss, we can still critique, we can still learn. But perhaps this kind of distancing to would help us to handle troublesome assumptions safely.... until we've got a handle on how the hell all this works.