Forget Protagonists: Writing NPCs with Agency for 80 Days and Beyond

meghna jayanth
19 min readJun 5, 2016


This is the text and (most of) the slides from my talk at the Game Narrative Summit at GDC 2016, the video of which is available to subscribers of the GDC Vault here. Thanks so much to the GDC organisers, track organisers and my Narrative mentor Susan O’Connor for their help, and for everyone who came to my talk. I really wanted to make the text of the talk public for people who couldn’t make it to GDC and/or don’t have access to the GDC Vault.

My talk is meant to be complementary to CJ Kershner’s talk The Lives of Others. CJ will be telling you all about how to use NPCs to increase player empathy. Whereas I am going to spend the next twenty or so minutes trying to convince you to limit, hurt, stymie and stifle your protagonists to give your NPCs room to breathe.

So my talk title is really a lie — we are going to do the opposite of forgetting protagonists. I’m going to try and convince you to make them pay, because they’ve had it their way for too long — at the expense of other characters.

So it might seem a bit strange, that I’m up here talking about writing NPCs with agency — when the game I wrote that you’re probably familiar with is inkle studios’ 80 Days — a massively branching anti-colonial steampunk adventure based on Jules Verne’s novel Around the World in 80 Days.

Fundamentally it’s about two white(ish) Europeans going around the world on a silly bet, encountering people, cultures and politics in fleeting glimpses. The set up of our story and our mechanics almost lends itself to shallow characterisation and the playing out of protagonist power fantasies — it would have been very easy to make a game that just retold The Last Samurai over and over –

a chewed over story of a white dude coming into a native culture and saving the day. But that was exactly the story we didn’t want to tell –

and we developed strategies to give our NPCs life, agency and integrity, and balance that with the demands of telling this particular kind of story. And today I’d like to share some of our design thinking and our strategies with you, in the hopes that you’ll be able to find it useful. Basically, if we could make a game full of engaging, diverse NPCs with agency given the original novel could have easily been adapted to be a tourism simulator — you can do the same with whatever limitations you’re working with.

We’ve spent too much time fridging girlfriends and little brothers because we’re afraid of hurting protagonists. We only dare hurt them by proxy. We think players will be turned off by experiencing a protagonist’s loss, or confusion, or stupidity or failure — who wants to play a “loser”, a failure? That’s something that seems true but I don’t think it is. In fact, it’s winning all the time that seems fake. True immersiveness necessitates loss, failure, pain, frustration.

Remember that bit from the Matrix when Agent Smith tells Neo about the very first Matrix

– a world without suffering, where everyone was happy? That’s one of my favourite bits of the film. Of course a simulation of perfection was bound to fail, of course we’d keep trying to wake up from it.

But I think it also fails because immersion is social –

we want to play with others, we want to engage with people who feel real — and people demand things of us. Being the speciallest snowflake is a lonely fantasy. There is a reason that Truman Show

is about a guy realising that his surroundings and relationships aren’t real and trying to escape. Everything in the Truman Show is all about him — but eventually that starts to feel shallow and unstatisfying — he doesn’t want the simple, transactional, easy relationship with his actor/wife Laura Linney –

he wants the thrill and unexpectedness and risk of Natasha McElhone.

Of course he does, of course we do — so why do games keep giving players Laura Linneys with perfect hair and saccharine smiles when what we clearly want is Natascha McElhone and her crappy apartment and deflating blowout? Truman deserves someone real and messy — and so do we.

But the thing is games protagonists aren’t Truman, who wants to wake up from the self-absorbed dream — they’re Gregory House from season 1.

the obnoxious jerk who gets away with his behaviour because he can somehow solve all the problems. Basically the fictional equivalent that one rude guest at the dinner party that never shuts up.

Everything orbits around House, everyone else is swallowed up in the supernova of his ego, and totality of his brilliance. And it’s a great TV show — but what if all TV shows were like that?

Okay, maybe this was a bad example. Asshole genius is clearly a popular fantasy.

But jokes aside, when it comes to television it’s so easy for us to understand how limiting it would be — to the kinds of stories we could tell, the kinds of characters and ideas and themes we could explore, if every TV protagonist had to be like House. So we have to stop games protagonists from getting away with it. This kind of structural centrality, these assumptions about what it means to be a game protagonist and have agency can lead us to unwittingly reinforce unhelpful cultural and social norms. To keep making the same game protagonist over and over again — because we are mistaking variations on a theme for variety.

We need to think about “agency” as an effect, a technique — within a wider context –

Agency should not necessarily be a goal in of itself.

Let’s talk specifically about 80 Days’ protagonist Passepartout and how we approached his place in the game world: in 80 Days Passepartout perceived whiteness and masculinity and Frenchness never go unremarked. We don’t treat these as though they are “normal”, or automatically powerful. We mark bits of the map as off limits because of them, we have NPCs deny you authority and question your motives because of it. Whiteness and masculinity make you an outsider in 80 Days more often than it allows you within the circle. We remind players of Passepartout’s social and cultural limitations. NPCs will often discuss their plans, goals, and ideas with Passepartout — but we very rarely allow Passepartout to make a choice for them. He gets involved in the revolutions, riots, plots and adventures of the people he encounters but he does not drive them — yes, he can meet a beautiful revolutionary and shout alongside him in protest against a despot — but he can’t lead the revolution. It was already happening without him — he wasn’t the one oppressed, so he can’t do the liberating — and often, the issues our NPCs struggle with are political and systemic — beyond the ability of one person to solve.

What that boils down to is that Passepartout — in a traditional choose-your-own-adventure sense — actually goes around the world “doing” very little at all. We limit his actions and his power — but NOT the player’s agency. Let’s define player agency very simply as “the ability to interact meaningfully with the game world” -

let’s even qualify that — it’s about making significant changes to the game world. The words we need to look long and hard are “meaningful” and “significant”.

Look at your friendships in the real world — you can probably remember a time when a friend of yours was going through something — a breakup, trouble at work or at home — and they told you about it. You listened to them. You might have even offered an opinion. They might have done what you suggested, or not. But that doesn’t matter. You were still part of that story — your relationship isn’t based on the other person following your directives exactly, or on you making the decision for them — the meaning comes from your relationship itself. The conversations you’ve had. The connection you share. The fact that you care about their problem. You might have made a significant change to their life simply by listening.

So this is something I want to talk about: player agency does not have to translate into actions. Even if the player cannot directly affect something — if they can have an emotional response or reaction — for a game to allow them the space to have an opinion can be as powerful as allowing them to “do something” — and giving protagonists this type of agency allows NPCs to have more development and depth, to pursue their goals without being constantly overriden by the protagonist’s overdeveloped sense of importance.

Let’s look at an example of this, from Bioware’s Dragon Age II.

How many of you here are familiar with this game? Now, it gets unfairly maligned, I think — Dragon Age II is actually my favourite Bioware game, possibly one of my favourite games of all time — and all of that is down to the NPCs. I can still load up any one of my thirty or so saves, and feel right at home — not because I remember exactly where I am in the main story, but because I know the characters around me, those relationships are what pull me into the world. They feel real in a way that many NPCs in games don’t — and that is because they often act independently. And sometimes, they lie — which almost NEVER happens in games.

So — SPOILERS — but in Act II of DA2 you the player find out that Isabela has been lying to you almost all the way through the game. No matter what the state of your relationship with her — how friendly you are, even whether you’re in a romance, she betrays and abandons you because she has pressures acting upon on her that you didn’t know about.

It is entirely possible for Isabela to leave permanently — leaving your companion quests hanging indefinitely, denying you the player and you the protagonist any resolution to her action beyond rage or belated justification. And even if she does return you have to deal with the consequences of her betrayal — her independent action. You cannot control what she does, the only thing you can control is how you react to it, how you feel about it — and that’s powerful.

We need more NPCs like Isabela if games are ever going to be more than

“entitlement simulators” -

Matt Boch talked about this at his GDC microtalk last year, that the stories, characters and mechanics of games are often designed to give us exactly what we want, uncritically. And that’s a problem — because who ever heard of that great novel where the protagonist got exactly what they wanted all the time? It sounds like a real page-turner.

Looking at games in this way, as “entitlement simulators” also highlights what’s wrong about most romances in games — why they feel transactional and goal-oriented.

How can we believe in an NPCs romantic desire if they don’t have any other kinds of desires, if they feel like skill-checks rather than people? We need to give NPCs agency and integrity if romances in games are ever going to feel like more than just…playing with ourselves.

Let’s keep talking about Dragon Age — pretty much every single romantic option in Dragon Age II stymies the protagonist in some way. I’m going to talk about the most dramatic example of this — which only happens to be, uh, my Champion’s romantic partner –

Anders: a mage possessed by the spirit of Justice who occasionally gets all glowy eyed and kills people. Definitely NOT Fenris –

a runaway slave with tattoos which occasionally get glowy and then he kills people. Just want to make that distinction clear.

So Anders, whether he’s in a relationship with you or not — chooses his political ideals over his personal loyalties and commits an act of terrorism that starts a war. It’s a great moment in the game — you sort of start suspecting that’s he’s lying and sneaking around, but instead of finding out that he’s having an affair he turns out to have been plotting a terrorist atrocity all along. What I love about this is that there is no way to stop him, or convince him out of it, or even really find out what he’s planning, no matter how many times you shag, or whether you move him into your fancy mansion — and if you are romancing him, the other characters give you a hard time about the fact that your boyfriend’s a secret terrorist. His actions reflect on you because you chose to romance him — usually the game protagonist’s actions ripple out into the world, the entire game is about seeing the consequences of your choices play out — but in this instance, you’re the one stuck having to cope with and deal with the consequences of an NPC’s actions.

What this does as well is begin to move us away from this goal-oriented conception of romance in a game — where you “win” a romance by getting a character to sleep with you. But in Dragon Age II shacking up isn’t the end of your story, it’s the beginning. I know it sounds pretty out there but the game gives you content and dialogue and quests with a character EVEN after you’ve slept with them, almost as if one shag doesn’t tell you everything about another person in perpetuity. I guess, if these romances have a win-state it’s staying together. Anders acts contrary to the interest and the goals of the protagonist, and the experience of the game is made richer for it — I really think Dragon Age II is a milestone for the treatment of relationships in games.

Giving NPCs control and power can create interesting effects and experiences in games, beyond romantic fulfillment as well. I wrote a couple of guest Islands for Failbetter Games’ Sunless Sea –

any of you played it? For those who haven’t — it’s a texty, literary game of darkness and exploration set in a vast subterranean underworld. I wrote the Isle of Cats — which is a smugglers’ haven, a really morally dubious place, a dangerous place. You have to pick a patron on the island and perform a series of quests to progress –at one point, one of the patrons — Zaira — demands you sacrifice a member of your crew to them — allow their memories to be harvested in a gruesome and horrible way. It is worse than murder, no way to access the rest of the content without giving Zaira what she wants. There is a real cost — to the protagonist and the player too — you can retain your morals, but you’ll have to give up on the story and the rewards. The protagonist doesn’t get to have everything, and certainly not for free — that is the entire point, that’s how we establish the creepiness & drive home the danger. But the thing is — the player isn’t entitled to the end of the story. And even though there is no way to subvert or trick Zaira the player is not left without a choice — there is a choice, and that is not to play.

Papers Please is an interesting counterexample that helps make clear the difference between agency and control, between agency and power — so in Papers Please you have all the agency in the world, to decide how to spend your time, what documents to pass, how to spend your money — but your power is hugely, pointedly limited. Part of that limitation is by making the NPCs demand and need more things than you can provide, forcing you to make choices, compromise, fail. Giving you the agency actually highlights how powerless you are — in playing you actually form a kind of kinship with the NPCs you nominally dictate to — you are ALL, in fact, powerless in the face of Aristotzka’s tyrannical bureaucracy.

All of these examples show us ways NPCs can help us make games that go beyond the power fantasy.

In fact, the way in which we treat our NPCs in 80 Days ties into the theme, but also is a huge part of how our players interact with and understand the world — in a very real way, NPCs are a key part of the worldbuilding.

We explore wider political, social and cultural themes through personal concerns and entanglements — the world feels complex because the way our NPCs think about, react to, engage with the world is complex and often contradictory — this is a much more elegant and natural way to unfold backstory, politics, plot. So in 80 Days almost every single NPC has a slightly different opinion of the Artificers’ Guild, they have their own biased and particular perspective, and that’s what the player experiences. Worldbuilding through NPCs in this way actually makes the world feel more coherent and alive than offering the player a journal entry or a singular, definitive perspective — it gives the player the space to make up their own mind, rather than having their mind made up for them. There is no definitive truth about the Artificers in a worldbuilding document somewhere — we developed our ideas about the Guild through writing the NPCs opinions and engagements with them — that’s what makes the fantasy feel nuanced and grounded.

The reason this works though — is that there are so many NPCs in 80 Days, and so many perspectives — no NPC has to bear the weight of entirely representing a culture, a gender, a race — this is the quickest way to end up with stereotypes and mouthpieces or merely vessels for plot or story. This is definitely NOT what we want.

One of the easiest ways to avoid this — beyond having more characters, and not just “tokens”, is also to do the research. Whether it’s about a historical period or a situation or a particular identity — look for real-world examples of dilemmas or people or places or cultures to help inform and add nuance to your portrayal. Don’t give your NPC the most obvious opinion you think they’d have — in some cases, you’ll need to do research in order to discover the stereotype in the first place. This is the very first hurdle, so don’t fall at this one — it’s lazy, and it’s boring, and it’s so, so easy to avoid.

Part of doing the research is also — as writers and creators — asking ourselves whether this is our story to tell. In 80 Days some NPCs stories are simply NOT FOR Passepartout — they are not for the Player — they are not for strangers, tourists or outsiders, and that’s okay. That can actually be a good thing. This is a question that we should ask ourselves when we are writing games and designing NPCs — how much of this is for the player? How much of this is for the protagonist? If we want to breathe life into our NPCs, there has to be something left over for them — something that is just for them, and not in the service of player, protagonist or plot. And that might mean telling the story in a different way than you imagined.

There are multiple times in 80 Days where Passepartout encounters a person or situation and can’t do anything to influence them, because of a particular set of cirumstances — and it’s been interesting for me to watch players respond to that in forums and blogs — I think it can tell us a lot about player expectation, and how hard we have to work to subvert that to make worthwhile NPCs.

In Brisbane you can encounter a girl from the indigenous Murri tribe, working as a maid in the hotel you’re staying at. You can engage her in conversation — and she tells you about the depredations her people have suffered as a result of colonialisation. And she has written a letter of complaint outlining such, which she wants to get printed in the local newspaper, The Moreton Bay Courier — to gain attention for her cause and her people. You — as Passepartout — realise that it is unlikely the white-settler owned newspaper will print such a note, especially not on the request of an aboriginal girl, and you can offer to take her letter to the newspaper offices yourself — and use your whiteness and maleness and position as protagonist — to help her. But she refuses. She does not trust you because you are an outsider, you are white, you are male — you are closer to the oppressor than you are to her, and all the good intentions in the world can’t change that.

And what’s wonderful and perfectly understandable is that there is a heated debate in various forums about “how to get the Murri girl to trust you with her letter”.

This isn’t surprising, games train you — the protagonist — to fix problems — or that, rather, they teach you that all problems can be fixed, and by you, if you try hard enough — but here, in this case, it’s not your problem, and you cannot fix it.

“Solving” it using Passepartout’s white privilege, actually would just make the solution part of the problem. Passepartout cannot ride to her rescue, cannot solve racism by posting a single letter. And the protagonist cannot get past years years of ingrained distrust of whiteness and his ignorance of local culture and politics in the course of one conversation — she is never going to open up to Passepartout no matter what item you — the player — try and use from your travels.

Awesomely enough, I checked the thread on the wiki as I was putting this slide together, and just seven hours earlier someone made this post:

Exactly, anonymous wikia contributor. This is not a bug; but this it the ENTIRE POINT of the game. But it is such an unfamiliar notion to gamers — and such an unfamiliar structure, that inhabiting the skin of the protagonist can be LIMITING as well as LIBERATING. That you will meet someone who does not immediately open up to the protagonist and want their involvement. It feels unfair to the player, to be shown a problem and not allowed to fix it. But it is important — we couldn’t have done it any other way without completely undermining the story that we wanted to tell. We had to deny the player their fairness fantasy, their white saviour fantasy.

We deliberately steal power from Passepartout and give it to the NPCs — that’s why the game works as is does. Because the way in which we understand and assign agency is not a “neutral” design decision — as though there is any such thing — how could we write stories of indigenous power and ingenuity with our words if Passepartout was mechanically solving all our NPCs problems for them?

It might feel unfair to the player — but maybe unfair isn’t the worst thing a game can be?

This is one of the myths of equality and multiculturalism and feminism that I particularly despise: that socially good and progressive outcomes will come without a cost. But you can’t deny this truth in game design — if you want someone who was silent to speak, someone else might have to shut up for a while. With 80 Days went in wanting to be respectful to other cultures, other peoples and characters — and that resulted in compromise. Balancing agency in a game as you would difficulty or challenge can not only be a question of design, storytelling or mechanics but also a question of ethics — giving NPCs agency can genuinely be an ethical imperative.

Specifically: one of our stated purposes going in was to make a game which undermined the power fantasy, the fantasy of white history, the fantasy of imperialism. The imperial fantasy requires that the struggles of colonised people, of people of colour, and women and working class people and minorities — are either ignored — or enshrined in pornographic narratives of victimhood and suffering. And victimhood all encompassing, it is reducing. Victims exist to be saved or pitied, but never respected. Victims are acted upon rather than acting.

Victims are quintessential NPCs — and if we don’t deliberately and specifically work to undermine that, all our NPCs are going to be victims even when we don’t intend them to be. Just as we need to deliberately and specifically work to make sure that “protagonist” is not synonymous with “hero”.

The more we make these distinctions, the more we interrogate these assumptions, the more we’re able to shift focus to NPCs, to allow them power and space and depth and importance. I think that this is vital to the progress of games narrative as a craft, and to videogames as a medium. We need to come come to an almost entirely new understanding of the meaning of protagonism and agency — a definition of agency that can assimilate a loss of control, protagonism that works outside of primacy, games that function outside of the simulation of entitlement.




meghna jayanth

Narrative design. Alternate worlds. Fantasies counter to capitalism-colonialism.