A lot of the earlier articles on Bumbershoot Software were written “from my old notebooks”—old notes or forum posts or conversations that I’d kept around and then worked into full articles. That’s fine for retrocoding stuff, where the state of the art moves fairly slowly, but it’s not useful for fields that are still rapidly innovating. My old notebooks still have a lot of material from my 2004-2014 span being involved with the interactive fiction community, but a lot of the figures there are now reasonably well known names in academia and industry, so much of my notes on what we’d now put under the umbrella of “interactive narrative” are extremely outdated. I do sporadically check in on Nick Montfort’s and Emily Short’s blogs to sample the academic and avant-garde commercial viewpoints in the space, but even there I’ve drifted far enough away from both the state of the art and the goals of the craft that there isn’t much I can talk about or grapple with.
However, Emily Short has a new post up on “Choice Poetics” that struck a chord. Back when I’d write nine or ten thousand words each year about the annual Interactive Fiction Competition, I ended up refining some principles and approaches to the entries in the competition that ended up serving me pretty well in subsequent years when approaching computer games more generally. The papers linked in that article—particularly the one formally analyzing Papers, Please—have forced me to re-evaluate some of those old conclusions.
So, let’s talk about complicity in video games. I’m using this term in both a narrow and a broader sense. In the narrow sense, it’s talking about things a game can do to make a player feel bad about atrocities they commit or have their character commit within a game. More broadly, I want to move this beyond the crimes that the word “complicity” suggests and consider situations where the player feels responsible for their actions within or to a game world.
I first attempted to formalize this in 2010 when reviewing Jason McIntosh’s The Warbler’s Nest, because it was brief and extremely pointed and thus crystallized for me what made me feel like it was laying the responsibility for the events of the game directly at my feet. To wit, if a player is to feel responsible for an atrocity:
- The atrocity must be optional. Otherwise the player is no more responsible for these events than a reader is made responsible for the events in a story by the act of turning a page.
- The atrocity must be the easy path. Otherwise it’s just Bonus Cruelty you have to work harder for.
- The player must know the stakes. Otherwise their actions do not have the necessary import. Lock a player in a room with two unlabeled buttons, and make it clear that the game will do nothing else until one of the two buttons is pressed, and the player will feel no responsibility for the result. Indeed, this is only fair; were such a situation to obtain in real life, the responsibility for the consequences of this action would be firmly on the head of whoever rigged up the buttons, or locked the button-pusher in the room.
That was almost a decade ago. By now, I don’t think any of these hold up as well as they did when I originally devised them. Here are the main problems or complications that, I think, have emerged with it:
- The player’s interaction mode may be too far removed for the analysis to apply. If game offers a predefined good path and evil path, the default assumption for many players is that they are expected to be completionist. The atrocities on the “evil path” may be optional, but if the player seeks to see the whole of the game’s content—particularly if certain maps, challenges, characters, or other such content is gated behind progress within it—they cease being optional when in pursuit of that goal. More simply, a player who has decided to rain meteors and dinosaur attacks upon their SimCity metropolis will feel no more remorse about this than they would for knocking over a tower of Jenga sticks. A player interacting with a game as an artifact to experiment with or as a library to exhaust has already made a decision that removes them from the analysis.
- People feel responsible for their actions in real life even if they didn’t know the stakes. Blind or actively deceptive choices still fail, but there’s a lot more leeway here than I had originally formulated. Allowing the player character to feeding a starving child a peanut butter sandwich and then rewarding them by having child die in front of them from anaphylactic shock would be a rather mean thing to do, but a player might feel some responsibility for the action, since they acted without some obvious information (the child’s peanut allergy) that they in principle “should” have known. On the other hand, one could imagine a hospital/surgery simulation where the epilogue reveals that the player character was not a surgeon at all, but a deluded lunatic who was just cutting people up in a barn. In such a game the player would be unlikely to feel even fleeting remorse for the actions presented at the time as saving lives.
- If an atrocity is difficult enough to perform, the analysis breaks down completely. This is hard to explain without actually working through examples, so I’ll defer that to my worked examples before.
Below the fold, I’ll work through a number of games and show how the original and the modified analyses interact with it. That also means I will, by necessity, be spoiling some of the narrative mechanics in them, so I will list the games here as the spoiler warning. Proceed at your own risk if you have not played Utlima IV, Bioshock, The Warbler’s Nest, Spec Ops: The Line, Fallen London, Papers Please, and Undertale. I’ll try to be vague about exact details that are not commonly discussed, but it’s still going to require going further than I’d like without spoiler warnings.
Ultima IV (1985)
I’m using Ultima IV as a control, more or less. Generally speaking, when we talking about players feeling responsible for their actions, the idea is that the game designer is punishing the player with negative emotions for their bad or neglectful behavior in-fiction. Ultima IV was expressly designed for the opposite purpose: Richard Gariott had tired of the way adventurers in role-playing games all seemed to be glorified kleptomaniac murder hobos and wanted to design a game where the point of the game was to become and be a good person. The general design of the game is that a player who wished to accomplish the challenges they were faced with and generally conquer the game would thus turn their efforts towards kindness and justice, while their expeditions into the wilderness and ancient monster-infested ruins were necessary acts for victory but always considered to be means to an end.
Time has not been kind to these design goals. Once the virtue/vice mechanics are understood, they may be cynically gamed for dramatic increases in capability and safety, to the point that this story usually ends, as I ended it in my discussion linked above, with the observation that “the accidental moral of the story is that the correct way to become a paragon of Virtue is to lead a highly profitable life of crime and then use the proceeds to finance various good works.” The default assumption is thus that the player would be expected to have a very low sense of responsibility for their actions, as one would play in phases; in the first phase, one behaves profitably sinfully and maximizes their material resources; in the second they restore their reputation with virtuous actions made easier by the previous crimes; and in the third they accomplish the goals of the various wilderness expeditions to actually beat the game.
We should expect the game to not fit well into the rules I outlined up top, and indeed it does not. Crime is only barely optional; without extremely plentiful gold for good equipment and large numbers of spell components the midgame is not particularly survivable. The “non-sociopathic” route I devised for my guide involves artificially triggering the endgame, equipping most characters with the very good and completely free endgame equipment, and then selling the rest at high prices to afford the remaining equipment needed. Despite this, by the late midgame crime becomes not merely not the easy path but actually an impossible one—starting the endgame requires maximum virtue scores and anything that lowers any virtue in the endgame will throw you back into the midgame. Finally, it turns out that the player doesn’t start out knowing the stakes but quickly learns them, and in particular that the stakes are zero. Nobody is ever permanently helped or harmed by your actions. Beggars’ conditions never improve, robbing blind little old ladies never causes them financial troubles or disrupts the supply of goods in their shops, and even outright slaughtering innocent townsfolk causes them to reappear none the worse for wear and with no memory of the combat once you leave and re-enter the room they are in.
Ultimately, the player is encouraged by the game’s mechanics to treat the inhabitants of the game world as the means to their end, and whether that is by swindling them or risking your life in their defense will vary based on those short-term goals. They learn early on that their actions do not matter to their world, but only to themselves.
This game was much discussed for its “moral dilemmas”, and it’s also gotten its share of derision over the years for that. The primary moral dilemma is that this game is a fairly traditional first-person shooter and instruction-following simulator where occasionally we are offered the opportunity to slaughter a weeping child for some powerups, or to not slaughter them and get a smaller number of powerups. The designers have repeatedly held to the position that this was intended to actually be a serious question, challenging the player to consider what they would be willing to do to acquire power, but I have honestly never encountered a player who took this choice in that way. Carl Muckenhoupt suggests that it is giving the player an opportunity to continuously reaffirm the correct choice, noting that moral behavior usually is the easier and more worthwhile path. Another player of my acquaintance was so offended by the suggestion that one would take the “kill the children” option that they found the nearest backdrop that seemed to encourage it (a banner reading THE GREAT SHALL NOT BE CONSTRAINED BY THE SMALL) and riddled it with bullets to leave a mark of their contempt embedded in the game world itself.
This was the game that motivated my rule regarding complicity requiring the atrocity to be the easy path. In Bioshock, taking the murderous path actually locks out options within the game, and storywise changes only the ending cutscene. A player that takes this route is very likely to only have been doing it in the first place to see what changed, and it seems to me that they would likely end up disappointed. Indeed, the answer to “how far are you willing to go for power”—if that was the question they were intending to ask—turned out to be answered by their own game’s event system with the answer “you should be willing to forego atrocities so that you have more resources at your command thanks to support from the allies you win by the virtue of not being a child-murdering monster”. This is not precisely a sharp dilemma, but it does at least provide a coherent moral universe in which players can act and react.
The Warbler’s Nest (2010)
This was the game that crystallized my initial three rules for me. The setup is very simple; the player character is convinced that her child has been replaced with a changeling and she needs to ritually slay it to prevent supernatural doom from befalling her home. That is fairly grim on its own, but the strain on the player is that the game’s text is extremely careful to never suggest that this concern is legitimate anywhere outside of the player character’s own internal monologue, where it is entirely unquestioned. If one goes through with it, the game’s final replay carries no reproach, just the joy of a mission well-completed. The aggressive mundanity of the actual world, though, subtly encourages the player to treat the player character as dangerously deluded, and the game from that point on can be interpreted as the player attempting to permanently dissuade the player character from her mission. Succeeding at this produces a very different final reply, more uncertain from the point of view of the player-character, but much less uncertain from the point of view of the player.
Revisiting my notes from 2010, it seems that even then my phrasing “the player must know the stakes” is a bit problematic; the player is never actually told here whether changelings are real in the story’s universe, and it never really openly invites them to consider the fact that they might not be. The player knows only the character’s crystalline certainty. Based on the rest of my notes then, it seems I was reacting to a certain kind of bungled attempt at complicity that was occasionally loudly attempted in the RPG world; usually the setup involved a traditional hack-and-slash fantasy quest and then revealing in the last scene that—surprise!—you’re actually just a crazy person and have been running around a city massacring innocents, you monster, you. (At the time I had written it, the interactive fiction game “Bliss” and the table-top roleplaying game “PowerKill” were the most visible examples of this.) I have never been impressed by the power of this argument, and indeed I turned it on its head in my prelude: one could just as easily have the main game be a game about life-saving surgery or emergency first responders only to then reveal in the last scene that—surprise!—you were actually a crazy person cutting people up for no reason instead of performing lifesaving surgery, or kidnapping them and delivering them to human traffickers instead of rescuing them from burning buildings and seeing them safely to hospitals. Hiding information about the import of the player’s actions destroys any responsibility they might feel for their actions, and condemning the actions of a player for not knowing information that the narrator was intentionally completely hiding from them rankles. If The Warbler’s Nest actually included actual visitations by faeries in its events, then the idea that the child at risk was in fact a changeling would be high enough that the player’s sense of personal responsibility would, I think, greatly suffer.
That’s two games in a row about killing children for bad reasons. Let’s broaden our violence horizons a bit for the next one, shall we?
Spec Ops: The Line (2012)
I’ve felt like this was the game that got the broader game-playing community talking about player complicity—Bioshock wasn’t really aiming for this and it also had the inability of the player to influence the game’s events as a major theme. This is, however, like Bioshock, a case where the developers’ interviews about the game undercut their actual goals.
The part of the game that everyone actually talks about is a point where the game essentially locks the player character and his party in a room and doesn’t let the game progress until the player pushes the “commit atrocity” button. This fails the first of my criteria completely, and indeed this was a point where the player is in some sense invited to stand outside of the story and begin viewing this, not as their own vicarious story, but to be an observer in a character study as they watch the main character’s descent into madness.
I originally would have phrased that as “fall from grace”, but in my own playthrough, that fall happened six levels previous, in which our main character was locked in a room full of fellow American soldiers and did not let you proceed until you had killed them all—a feat that neither the game nor the rest of the characters seemed to call out as more than slightly unusual. So by the time I’d reached “Press X to commit atrocity” I was entirely disengaged.
The developers are apparently on the record saying that the player is actually being encouraged to quit the game at this point instead of being party to all of this, which is a claim I neither accept nor respect. Even leaving aside the rather impressive chutzpah required to say “we think you should buy our $60 game and then quit after playing the first mission so as to not be a terrible person”, a linear plot does not change if the audience leaves partway through. If I stop reading The Game of Thrones after half a page, that doesn’t make its body count zero. And if there’s no way to avoid the atrocity in Spec Ops: the Line, that means that the story that game tells centers around that atrocity and it is the author, not the player, that insists on it being there.
Much later, long after the central atrocity has been committed, entirely unrelated to that atrocity, and long after even the most unobservant player will have noticed that something is seriously wrong with the player character and their ability to interact with reality, a choice is offered that meets every one of my criteria perfectly. The player is, as is so common in games of this nature, blocked from proceeding and is confronted by a large number of hostiles. The player is also aware that they really should be proceeding through these hostile people, and that under ordinary circumstances it would not actually be acceptable to kill them. Killing them, however, is clearly the easiest path; they show up as enemies in the game interface, and it’s a first-person shooter. The stakes are also quite clear; if you shoot them, they will be dead, but also will not be blocking your way.
There are three aspects that make this scene work brilliantly in the way that its most famous “dilemma” fails.
- It is not obvious that the atrocity is optional. Most games follow the Bioshock model for scenes like this; a menu would pop up and you would be making an out-of-band decision to slay or spare your foes. Not here. You have your full, normal freedom of movement, and full, normal access to your arsenal, and you may aim, shoot, or otherwise attack freely. It is up to the player to consider that they may have nonlethal or even nonviolent options.
- The atrocity is the easy path no matter what mode the player is in. By this point the character is obviously mad, has already committed multiple atrocities, and is in the process of being severely provoked. A thoughtless player might just assume that this is simply yet another atrocity to commit, if they thought about it at all even more deeply than “here are things that show up as hostile, and I still have ammunition left.” A player following along with the character’s descent might well conclude that the atrocity would be in-character at this point and go along with the action, accepting it as the action they choose even if they do not think it is truly optional. But to avoid this atrocity requires the player to actively rebel against not only the game but the basic interface and interaction modes.
- The game immediately passes retroactive judgement. This is where rebellion is rewarded, or thoughtlessness called out. There are two achievements tied to this scene: “A Line, Crossed” or “A Line, Held”. The design clearly contemplates a player acting more or less automatically as the game has trained them. This either sharply points out that there was another way, or acknowledges that they had found it and that they were meant to.
This was well-done enough, and effective enough, that players who did not receive the immediate judgement (say, because they played on a system without achievements) reported feeling terrible about taking the atrocity route here even long after beating the game, once they learned an alternative path existed. This is interesting because it implies that if the conditions set turn out to have been met at the time it can spark the complicity reaction after the fact at the point the erstwhile player learns that they had held.
Fallen London: Seeking Mr Eaten’s Name (2013)
To the extent that Spec Ops: the Line failed to achieve player complicity, I had generally ascribed it to the player deciding that the player character had gone uselessly mad, and had relegated themselves to the role of watching an atrocity unfold. This conclusion was premature. The first counterexample I encountered was the optional quest in the Fallen London web game usually called “Seeking Mr Eaten’s Name.” Its author, Alexis Kennedy, described it thusly:
In the early days of Fallen London, we added an experimental storyline. It gave the player the option of developing a ghastly obsession which would ruin their character’s life, requiring savage ordeals that chewed up their abilities and resources. It was initially very popular, and then as we tightened the screws and people realised we meant the warnings that no good would come of it, only the most determined stuck with it.
This is a point where the initial analysis I’ve been using starts to crack. Embarking on the SMEN storyline is not to be done lightly, and the costs get more and more devastating as the game progresses. Since this is a web-game based on time, this is a case where years of progress may be lost in seconds, or money spent to speed that progress likewise have been spent in vain. No player proceeding down this path is doing so thoughtlessly, and they are very much aware of the path they take and where it will lead. But they are also, obviously, not actually going mad, so they must be in some sense along for the ride, audience to their character’s descent.
What I think we see here is one of the boundaries of the notion of “responsibility” I’m trying to develop here. Players that take this path are acting as creatures of pure determination, intent on making progress toward the goal simply because it is the goal, and not for any rewards or overall efficiency. Proceeding through the quest requires one to repeatedly refuse to turn back. The player is given a surfeit of options to abandon the quest forever, marking their account such that the quest cannot even be restarted without some dark miracle, and the player must keep choosing to stay on their doomed path just as a Bioshock player would continuously choose to spare the Little Sisters.
This is no small thing. But I think it lies just past the border of the concept I’m exploring here. It uses very similar means, but some slight differences in the player’s intent means that it leads to a very different end.
Papers, Please (2013)
I had not actually considered this game as a test case before reading the Mawhorter paper. It is interesting because it directly contradicts one of my conditions. I take it as completely undisputed that Papers, Please makes players feel a lot of strain about its corner cases, and generally sets things up so that there are common cases where the player is tempted to bend the rules for the sake of someone who claims to have an excuse, or great need, for why their paperwork is not in order. The goal of the gameplay design is to have the player observe themselves turning into a relentless and pitiless cog in a machine of oppression, and one of the ways it does this is to encourage the player to turn away what seem to be worthy cases because there is nothing in it for them and much to lose otherwise.
That’s interesting from my initial perspective because the goal of the game is to get the player character to do awful things and to have the player make excuses for them, as if it were intentionally reversing my complicity results. That said, the preconditions are mostly there. No response is ever truly compelled; accept/reject selections are always freely chosen by the player no matter how absurd the option is given the evidence or lack thereof. The “bureaucratic cog” option is always the path of least resistance, and the only ways to get in trouble that don’t involve failing at your stated duties involve taking unwarranted initiative. Finally, the daily newspaper reports provide hints of the larger consequences of some of your smaller choices.
What I had not considered is that the smallest choices ought to work in this framework as well, and they don’t. Simple cases where a person claims their children are waiting on the other side of the border are one-and-gone, and we never really know whether we had just been fined for “incompetence” in the service of a good deed, or to enable a swindler through. On the other hand, we know for a fact that some of the would-be entrants are swindlers, because their attempted swindles are extremely obvious. (One is so clearly incompetent that this is played for quiet laughs, but one does begin to suspect that the cases where his paperwork appears to be in order is just the work of an unusually good forger…)
But these lesser cases are just as stressful to the player as the more drastic plots that unfold around the character. This complicates my analysis: the player never truly knows the stakes, and also never encounters the consequences. Whether their choice was wise or foolish, cruel or kind, is never known. I think this suggests that the rule regarding the stakes needs to be relaxed a bit, to something like “the player must believe that they are being given an amount of accurate information that is reasonable for a person facing that situation.” A border officer reasonably could face someone with paperwork problems and a convincing sob story and not know if it’s a minor problem that should be overlooked or a swindler giving away their scam; a surgeon would not reasonably face the suspicion that they are secretly a mad butcher who is not even in the operating theater they perceive all around them.
I’ll be concluding my little tour here with the game that directly grappled with the issues of player complicity in the crimes or lack thereof of its player character. The Pawhorter papers discuss Undertale but do so restricting itself to two routes; the No Mercy route (described in the paper as the “power player” route) and the True Pacifist route (described as the “story player” route). For my analysis, though, I need to diverge from that framework, because from the point of view of player complicity, the No Mercy route is not actually the atrocity, despite it being the path in which the most atrocities are committed.
The general idea behind player complicity is that the player should feel bad about taking actions in the game world that cause harm. By this standard, the actions the player is encouraged to feel bad about are the ones on the “neutral violent” path; this is the path a “thoughtless” player would follow if they were to treat the game like an ordinary CRPG. This is a path where they are upbraided regularly—even by the villains of the piece—for their acts of violence, and where it plays some tricks to remember past runs that were saved over later to, if not rebuke them for doing so, at least to remind them that the past is still the past. My three criteria are met quite simply here: finishing a fight by combat is always the easiest route and it makes later combats easier; fighting is always optional, despite the taunts of the villains, and as acknowledged by the game’s initial presentation; and the consequences of combat are generally clear in the short term, even if the long-term effects are not as clear.
That leaves the No Mercy route. The game goes to some effort to ensure that the No Mercy route will not be hit by accident—it requires not only that the player conclude each encounter lethally, it requires them to continue to farm (ever-rarer) random encounters until they cease entirely. A player simply proceeding through a zone as normal will complete it (thus ending all further encounters, and locking the player out of the No Mercy path permanently for that run) without actually reaching the point where this path would be satisfied. In the No Mercy path you are not merely showing no mercy by slaying those who cross you; you are slaying everyone.
The game world notices this, and has an opinion about it. Nobody remonstrates with the player in the No Mercy run; it is simply acknowledged by all that the player character is an inevitable force of destruction, and they all act accordingly. While there are a number of “rewards” on this path—in particular, a number of plot-specific boss fights that do not otherwise occur—the game does still work to dissuade the player from following the path. Neither the game itself nor the NPCs in the game world do this by attempting to appeal to the player’s empathy, and only rarely does it attempt to stop the player with an extreme challenge. The player is assumed to have switched their empathy off for this run, and is determined to see the No Mercy path through, and to be good enough at playing the game that they actively relish any extra challenges.
So what we see is that the challenges of the No Mercy path are designed to attack the player’s determination directly. The boss fights that are not extremely difficult or unfair are trivialized anticlimaxes. Noncombat obstacles are likewise removed. Actually triggering the last few encounters in each zone is an exercise in patience with no intervening rewards whatsoever, and failing even once knocks you off the path and may require a complete reset. Most tellingly, this attack even becomes explicit at the very end; if the player fails to stick out the requisite number of encounters in the final zone, then despite the fact that you are immediately victorious in the zone’s disappointing anticlimax bossfight, said boss’s dying speech tells you that it has determined that your murderous intent was lacking, throws you off the No Mercy path, and mocks you for failing at the final gate.
Viewed from this perspective, the parallels for the No Mercy path are not to the cases of player complicity we’ve seen above, but back to the mad, self-destructive determination of Seeking Mr Eaten’s Name. Like SMEN, the player must constantly reaffirm through hardship (and, notably, not the kind of “hardship” that a player would relish as a challenge, but rather mostly mind-numbing tedium and the removal of enjoyable aspects of the game experience) their will to see the path through. Like Spec Ops: the Line, though, the player’s affirmations are all done entirely through the normal game interface, and not through refraining from selecting some clearly labeled “Turn Back” option.
Conclusion: How Does the 2010 Theory Stack Up?
I’m pretty happy with it, overall. It struggles with a number of corner cases, but those corner cases end up lining up neatly with the rules themselves:
- A player may still feel awful about taking an action that is required to continue. Even if it is not a choice within the game world, this has parallels with vicarious dread or embarrassment that an audience might have in noninteractive media like books or films.
- A player may be filled with a mad determination to see some path or outcome through irrespective of the cost. This is definitely not the same thing as a player’s guilt, but it seems to be presented and manipulated by similar techniques.
- A lack of resolution can contribute to a player’s unease or discomfort with a decision. Having to act on imperfect information is part of our lot in life; the rule regarding the player knowing the stakes needs to be flexible enough to account for the player’s decision making process.
Over the course of this review, the moments that impressed me the most were the points where a player did not realize they were making a decision until the fact that their failure to live up to that decision was thrown back in their face—Spec Ops uses the achievement system to drive home to the player that they’d let themselves slip into casual violence, while Undertale pulls the rug out from a player who had started taking their murderous intent for granted.
I’m also not completely convinced that this is necessarily a desirable condition to induce in players—most of the examples here are essentially about putting the player in the position where they will feel pressured to take actions that they and the game (as the pressuring entity) acknowledge as shameful. This seems like it would only be effective in moderation, and that overuse of it within or even across works would risk players disengaging from the characters or the story more readily.