Object Orientation
I’ve been playing around with the Source Filmmaker animation tool a bit; it’s a strange experience. We spend so much time crafting illusory people, characters constructed piece by piece out of textures and polygons and recorded lines of dialogue, and at the best of times we are genuinely emotionally affected by the experience. You know, art. And yet, we can so easily strip all that away, and give you the same characters, the same lines of dialogue, like digital dolls, for you to play with as you choose. While I supposed the same is kind of true of other media, in that we can create fan-fiction and fan-art that creatively interprets and repurposes our favorite characters, there’s a big difference between creating fan-works echoing your understanding of a character and being given the exact mold that character was created from. It’s a bit unsettling, like being given an actual person as a doll, and that strangeness only increases as ever more sophisticated and high-poly models crawl out of the uncanny valley.
At the same time, there’s this ongoing discussion of how video games can abstract and objectify their characters, with a particular but not exclusive focus on the sexual objectification of female characters. Objectification in games is kind of a weird subject, because in so many ways the characters who populate games literally are objects. In terms of presentation, they’re a collection of polygons and textures and dialogue; in terms of programming, they’re a collection of coordinates and behaviors and states, usually connected together into a construct which is actually called an object in most programming languages. In terms of game design, they usually function as either a book with an accompanying audio recording or a vending machine – in every available and obvious sense, they are an object.
But we try to make them more. Sometimes. It’s never easy. Let’s be honest: It’s incredibly difficult to learn how to see other humans as truly human, to see them as not merely their religion or their ethnicity or their body or even their history or skill-set but as another living consciousness like our own selves. Though sexual objectification is the most commonly understood and railed against form of objectification, we dehumanize others daily by hastily generalizing them into narrowly defined narrative roles. The kind of objectification which lets us view women as existing for our pleasure is not fundamentally different from the kinds of demonization that allow us to view certain nations as existing solely for our displeasure. It’s a simplification, a cognitive cost-cutting measure, that reduces other human lives solely to the role they play in our own naive daily narrative.
Make no mistake, if this so-called ‘Candy Man’ were literally made of candy, those children would be devouring him where he stands
However, it’s not entirely fair or accurate to describe objectification in wholly pathological terms. Breaking the world down into understandable symbols is a necessary step for our coming to understand and interact with it, and information will always be lost in that transition. We never completely understand one another as human beings, merely build up an increasingly sophisticated network of interrelated symbols to represent each other. Objectification is a matter of degree and of attitude, not an on/off switch. It’s also worth mentioning the inverse of objectification, anthropomorphization, where we start to understand and relate to something that is physically an object as though it were another human being. The fundamental question is, when faced with a behavior from the given object or creature, do you attribute that behavior to intelligence and will, or to their unalterable and foundational nature?
Even as it’s obvious we can’t understand each other as wholly human, it’s rare that we completely relate to each other as objects as well, no matter how weird and abstract the interaction. Even as we treat a person as sexual object, in fantasy or discussion, their particular and unique properties as a person are what make them interesting as sexual objects – and, even as we develop fetishes for objects and pieces of clothing, those fetishes are formed out of the cloth of our history and connect us personally to something, however fuzzy and distantly understood. These two concepts always exist in our mind, no matter what we’re thinking about: Other-as-agent, intelligent and willful, and Other-as-object, something to be utilized or discarded based on its inherent usefulness. In many ways, the problem with regarding one another as objects is not so much the fundamental incorrectness of this viewpoint, but in our belief that all objects are for sale, all objects can be owned and controlled. This belief is what makes objectification inherently diminishing, that makes the not-me something lesser, a possession, a trophy. Not everything is made to be owned, colonized, exploited, and the fact that we forget that so often in regard to our fellow humans follows naturally from how readily we’ve forgotten it in regard to the world we live in.
Artists have an uphill battle, then, in using objects to emulate humanity – or at least they should. In practice, in modern culture, we are so primed towards media consumption that we find it easier, sometimes, to understand the actor on our screen as a person with their own motivations and goals than we do in seeing our waiter or garbageman or camgirl as the same. Though we know that one is a character and the other real, we have learned to engage the character as a human and the person as a utility. I mean, this sounds controversial, but these are literally their jobs: The actor’s job is to emulate another person and put their struggles on-screen so eloquently you feel for them, and the service-person’s job is to do whatever is required of them quickly and competently and pleasantly, like a machine. We engage the object as person and the person as object. Thus, the arc of art: A person creates an object meant to be related to as a person, a deeply personal connection created through the dry conduit of a deeply impersonal medium. This is all the more challenging in game design, since our objects are that much objectier.
“Mr. Treehorn treats objects like women, man!”
It’s dangerously easy to forget how much we’re abstracting when we create characters, especially game characters. In effect, a character becomes a stat sheet: Some numbers are high and some are low, and this tells us everything about the character, how capable they are, and how they behave when simulated or stimulated, their preferences and their biases all broken down into a digital statistic set which is run through a parser which creates the behavior. This is several layers of abstraction away from humanity: First, breaking our understanding of a human down into a set of symbolic attributes, into their physical strength and their intelligence, health and friendliness and race and height and so forth. These concepts are not neutral: We value certain forms of strength above others, and the same for other aptitudes and quirks: Merely giving any one of these a name and definition is a cultural-political statement. If we fail to even see colors that we don’t have names for, how can we possibly understand beliefs and attitudes and strengths and weaknesses which we have yet to name?
So what is it we talk about when we talk about intelligence? What is it we talk about when we talk about beauty? When I say a dog is intelligent, I mean that she ‘understands’ what I want from her and is, moreover, willing and able to perform to expectation. A dog who’s intelligent enough to ignore her owner in pursuit of her own interests is not considered intelligent by most humans. With that in mind, does it mean anything substantially different when we say a human is intelligent? Either way, we are quantifying their ability to perform to a certain expectation: Forms of aptitude not constrained within this expectation are disregarded, or never noticed at all. In other words, while ‘intelligence’ is what we like to call problem solving ability, what we consider to be problems, and what we consider to be solutions to those problems, will vary from person to person, culture to culture, moment to moment. Thus, not only is the measurement of intelligence via testing suspect, the entire concept of intelligence is suspect on the same premise. Intelligence, like beauty, is in the eye of the beholder, and its shape is a manifestation of our individual experience and our culture.
So after we break down the simulated human that is a character into these different strengths and weaknesses, we then further abstract these concepts into data that can be parsed by a computer. This is, again, not a neutral act: How data is broken down and stored, and which data is stored, changes its meaning and usage. A game that tracks a character’s lineage through 100 generations is very different than a game that tracks the strength of each group of muscles on the character’s body – well, unless that game is Dwarf Fortress. Most games don’t even track that much – in most games, a character is basically the sum of their model and their recorded lines, the quests they give and occasionally some kind of special ability. We still overwhelmingly communicate character through the tools of film and literature, rather than by actually modeling their behavior. There’s lots of reasons for this, but prime amongst them is most likely just that players aren’t primed to notice how a character behaves and understand it to emerge from the character’s persona. When a player sees an AI-controlled character behaving strangely, they’re much more likely to think ‘buggy’ or ‘cheating’ than panicked or confused.
This is unfortunate, because we could do so much with characterization in games. Though I don’t think the tools of film or literature will ever fade from primacy in game storytelling – sometimes it’s vital to communicate understanding of a character by putting them into a very particular situation in a very specific way, and abandoning the ability to do that won’t serve anyone – by ignoring the other channels we could be communicating through we’re hollowing out our portrayal of a character. While a movie can provide a ‘portrait’ of a character, as they behave under specific circumstances and influences, we can provide a ‘sculpture’ and demonstrate how they behave as those circumstances shift and flux dynamically, an actor improvising her lines as the play changes and melts away, wholly becoming her character. Still, perhaps, less complete an understanding than we can ever have of a real person, but more and more robust than we could create through any past media.
It’s an open question how convincing we can make these ‘sculptures’ though. Certainly we can model certain AI quirks through programming, make them likely to perform a certain technique or make the same rash decision, but eventually the player will see through those, and eventually the canned lines will dry up, and the character who was so full of life at first will begin once more to be a mere automaton. This is a problem which, for the foreseeable future, is only potentially resolvable through multiplayer. Most games use multiplayer as a battle of wits, but the potential is there to make as all actors, players in theatrical play, understanding and working to manifest the character we have been given. The barriers between the current state of gaming and this kind of play are, at this point, primarily cultural rather than technical. Players are used to playing a role in the tactical sense, but not in the sense of character – and, since it’s difficult to enforce something as nebulous as being ‘in character’ (this would lead us back to the same problem of quantifying characteristics as before), players would have to operate solely off of the honor and reputation systems to ensure a satisfying experience.
The boundary between human and object defines the line along which art lies, and our parallel perception of the two will only get stranger and more abstract as the population grows, our ability to connect becomes more distant and mediated by technology, and our simulacra of humans become more sophisticated and realistic. Divisions which were once clear, who was part of or apart from our tribe, what was a tool and what was a person, a servant or slave, a pet or a friend or a pocket calculator, all of these are fading into the distance as we speak, eroded by technology and technique. It may be difficult, as we progress, to continue to be able to see through the intervening layers, veils of technology, and find another human being behind that – but, for the sake of our common happiness and humanity, it may also be vital that we learn to do so.
And I, for one, welcome our new robotic and occasionally vaguely squid-like overlords.
Problem Machine 
This was really eye opening, especially when you drew a parallel between the relationship between people and in-game objects. It was a pleasure to read, thanks.
Interesting take on character design, and entertaining to read as usual.
By the way, tools like Myer-Briggs indicators (or more recently true colors) are often used to model complex personalities. I wonder if in the near future such tools will give birth to behavioral softwares accurate enough to be actually used…
I think that sort of thing is a bit too general for most game situations. I mean, you could use it to create a general behavior profile for, say, a soldier in a strategy game, but it wouldn’t really offer much in that instance over something specific like caution/aggressiveness stats.
In a more narrative context, what we really want to see as an audience a lot of the time, I think, is seeing how a PARTICULAR life experience leads to a particular decision — general personality profiles are a nice baseline, but don’t provide that. Something that generates traits based on good/bad experiences, like fetishes and phobias, would probably be more likely to lead to narratively interesting character moments. I think Dwarf Fortress may have some stuff like that, which leads to some really incredible generative story moments even when the actual events are related very crudely.
Myers-Briggs is actually quite complex, with a large number of personality types combinations and insights on how human psyche works. Like how when under pressure we tend to display the opposite of our usual behavior. Of course it’s not perfect, but it’s good enough to simulate emotions or desires at a game level. I’m not advocating Myers-Briggs though, there are other ways of mimicking human decision-making
I do agree though, that in itself it wouldn’t create a “living”, and more importantly interesting character. Behavioral units lacks both evolutivity and the depth created by a past. However, I feel like strictly generating or recording events isn’t enough either. We are more than the sum of our experiences. Well, it’s debatable but it is at least safe to assume that different humans would react differently to the same experience. It’s as interesting to see how a particular decision is made DESPITE a particular experience – just because this guy won’t give up..
It’s as you said : a character won’t feel real as long as it only acts rationally. Once the player begins to see the pattern, once he can reliably predict his actions, what was alive suddenly collapses into a lump of pixels. No matter how many phobias and hobbies, it’s just an illusionist trick. Now it’s good enough for a film, where the set of characters and events are tightly controlled. Storytelling games like The Walking dead are very similar in this regard.
However, if we’re talking about more open content, the mechanics may begin to unfold as the player advances through the game. An example would be Darkest dungeon, where stressful events and ensuing traumas lose their narrative qualities the second time it happens. I didn’t dig deeply enough into Dwarf fortress (yet) to make any bold statement about it, but their characters do seem to behave based on more than their past. They follow rules we don’t know about, complex enough that we have trouble figuring them out, and that’s why they feel alive – in my humble opinion.