The principal trope of the zombie genre is that in an undead-infested apocalypse, it’s the other humans you need to worry about.  A zombie plague is a natural disaster, an epidemic, an indiscriminate force of nature that is driven by neither logic nor passion–it just is.  What the genre has to teach us is that in a desperate situation, the worst atrocities will be perpetrated not by the infected hoards, but by the surviving humans.  This is the central appeal of a zombie movie: it puts characters in a high-pressure situation to see how their humanity holds up.

The craft demonstrated in this title is unbelievable.

The Last of Us Part II proudly continues this tradition.  It is a character drama set in a post-apocalyptic America that is more interested in the lives of the protagonists than the events that caused the destruction of the world.  Its choice of genre forces it to treat its characters like real, believable people, and the title puts huge emphasis on making its cast convincing.

It does this with flair.  The Last of Us Part II is an unprecedented production achievement, a dramatic outlier on many qualitative axes, outshining other AAA games on everything from graphics technology to script writing to voice acting.  It is clearly standing upon the shoulders of Naughty Dog’s previous releases (in particular the Uncharted series), but surpasses those titles in every category.  I was thinking so hard about the story in this game that I spent several nights dreaming about its conclusion.  But it is the very quality of the production that also reveals The Last of Us Part II‘s central challenge: walking the tightrope between character drama and gameplay expression.

All games with storytelling aspirations must find a way to deal with the inherent mismatch of narrative and player choice.  Most forms of narrative in other media are passive: we’re told a story and our role is that only of an observer.  But video games are played with controllers because we are supposed to control them: it is the active participation of the player that separates the medium from other forms of entertainment.  Narrative game designers are often confounded by the melding of passive story consumption with active participation in the game.

Many resolve the conflict by simply keeping the story and gameplay separate.  The story is metered out as the player progresses through the game.  Finish the tasks, watch the cutscene, head to the next room and get yourself a new task.  Beating the level is a turn of the page.  The player steps from one room to the next, performing whatever actions the designer has asked of him, but ultimately has no influence over the arc of the story: he can only move forward along a predetermined path, or not move at all.  He has control over his avatar, but his avatar is itself simply a rat in a maze.

If the activities are fun and the narrative compelling, this is perhaps enough.  At the macro level the arc of the plot can be controlled by its author to ensure that it is interesting.  At the micro level the player still feels that he has agency, at least over moment-to-moment choices.  We often don’t even notice the discrepancy between the macro and the micro because the input at the player’s disposal is not expressive enough to affect the macro-level story.  We never have to worry about what happens if Mario decides to run off with one of the Toad girls he rescues before making it to Princess Peach.  The player simply doesn’t have the input grammar to make such a choice, just as the rat cannot choose to bore holes in the walls of the maze.

Of course, there are games that also attempt to allow the player to influence the macro thread of the story.  Often these are simple binary choices: are you on Path A or Path B, good or evil, rescue or harvest, paragon or renegade, a branch at the macro level that selects one story over another but nevertheless keeps the player on a predetermined path.  Games that attempt a greater level of complexity risk producing a nonsensical story as a result.  But in most cases, story choices in video games boil down to just one more level of abstraction between the macro story path and the micro gameplay mechanics.

The Last of Us Part II’s design philosophy eschews such obvious binary choices and instead relies on small bits of implicit choice within a fixed, predetermined outer layer.  Our avatar is Ellie, and as rats in mazes go she’s got quite a few tools at her disposal.  We can choose to send her left or right, investigate that dilapidated bar down the street or keep moving, sneak past the guards or stab them in the neck, engage the zombies directly or pick them off one at a time, climb up the side of a car to reach a second-story window or slip through a storm drain to access a basement.  We control Ellie’s path through a level segment, but not the sequence of segments itself.  The entire design of this title can be seen as a spectrum from high-fidelity, high-choice micro-level interactions to fixed, macro-level predetermined encounters, level progression, set piece sequences, and overall story arc.

In fact, this design philosophy is a pretty common one, and while the execution in The Last of Us Part II is impeccable, it’s hardly unique.  The appeal of this design is that it makes the player feel as if they have choice while maintaining the impact and the coherency of the story by bounding that choice within a local maximum.  In the ideal case the player never notices the facade, and his perception of agency increases the punch of the story because it feels personal.  But there’s a problem, one that affects The Last of Us Part II more than it does Naughty Dog’s other games, an issue that is intrinsic to the framing of the story as zombie fiction, which is exacerbated precisely because the production quality of the game is so high: it is the moment that we realize that Ellie, our avatar, our rat, is not ours to control at all.

There are moments in Uncharted where the curtain is pulled back and we accidentally catch a glimpse of the rollercoaster ride machinery that powers the game.  Moments when we see that our controller input is more for our own benefit than the game’s, that in order to deliver the dramatic action sequences the series is known for our agency has been quietly suspended.  That’s ok in Uncharted because hey, we didn’t really want to screw up the timing on that train car jump and have to redo it anyway.  We might be on a rollercoaster track but the peaks are high and the drops are fast and that’s what we came here for.  Nathan Drake may murder people every few minutes on his way to raiding yet another tomb, but that’s not so incongruent with his character because he’s an Indiana Jones comic book person.  We don’t need to think too hard about his motivations to have a good time.

But the motivations of the characters in The Last of Us Part II are central to its story.  The point of zombie fiction is to examine the humanity of its characters, and Ellie’s humanity is the central axis upon which the story turns.  Is she a good guy, in the right for wanting to avenge the death of a loved one, fighting only out of necessity, or is she just as bad as the people she’s hunting, a psychopath willing to go to any length to achieve her goals?  The title puts huge effort into making the situation morally ambiguous, even going as far as to give the cannon-fodder bad guys who jump Ellie in the overgrown streets of Seattle proper names (“Oh my god, she shot Sean!” an enemy barks as I drop her companion with a shotgun blast to the chest).

I don’t like this scene, but not for the reasons the designers intended.

In every dilapidated building Ellie finds notes from folks who probably didn’t make it, describing the horror of their lives and the tragedy of their situation.  We’re supposed to feel empathy for these people, even as we direct Ellie to slit their throats.  With highly detailed animation and rendering we can read the expression on Ellie’s face and the faces of her friends.  We can see the blood splatter left by a man who shot himself rather than turn into a zombie.  The reason The Last of Us Part II has such an astoundingly high production value is that emotional expression is in the details, and this title is overflowing with detail.

My version of Ellie is as close to the moral high ground as I can get her.  She doesn’t like to take life unnecessarily.  When confronting humans I prefer to play her stealthily so as to avoid needless bloodshed.  If forced into a situation where the bad guys have to be taken out, I’d rather let the zombies have them than pull the trigger myself.  I only kill another human as a last resort.

For the most part, The Last of Us Part II gives me enough agency to make these decisions.  But upon closer inspection I notice that even my palette of options has been designed to force me to choose between binary extremes.  When confronting other humans there are no non-lethal options, and killing is often the easiest path.  I can’t knock a guard out, I can only drive my knife into her jugular.  The game has removed the middle road: I only have the grammar to murder or sneak, and sneaking is harder.

Of course, this setup is by design.  The game designers want me to have to make hard choices about whether my avatar’s humanity is so important to me that I’ll forego iterating a video game combat loop by route.  Like Metal Gear Solid 3: Snake Eater, this title has an opinion about the righteousness of taking human life, and wants me to reach the same conclusion in the course of play, without having to spell it out.  It has biased my options to produce that effect.  If, like me, you’re a fan of Metal Gear Solid 3, you’ll notice a lot of its DNA in The Last of Us Part II.

But there are points in the game where the macro-level story arc requires an even greater suspension of my choice.  If Ellie is to grow and change as a character, certain events must happen along the way to make that change a believable one.  To increase the punch of those moments, it’s better if they happen while I’m behind the wheel, rather than in a passive cutscene.  But when these events conflict with the way I’d prefer to play Ellie, the illusion is suddenly shattered and I’m left feeling more like the puppet than the puppeteer.

With realistic characters come realistic expectations.

That’s the central problem with the system that The Last of Us Part II employs: when the narrative supersedes the player’s choice to such a degree that the marionette strings the game has been trying to conceal are suddenly revealed, it feels like a breach of trust.  It hurts precisely because the game is so successful at making its characters real human beings who we empathize with.  The dramatic quality of the character drama in this game makes these moments all the more transparent.

There were points in The Last of Us Part II when I put the controller down because I didn’t want to pursue any of the options the game made available to me.  I hoped that inaction would be validated as a choice, that by electing not to Press X to Commit Violence I would be spared the pain of having my hand forced.  Unfortunately, that’s not how the story goes.  The show must go on.

It’s impossible to talk about these moments without spoiling the narrative, but there is a minor optional sequence that is a good encapsulation of this problem. In this sequence Ellie is attacked by three mercenaries in a hotel building on her way to a hospital.  They mistake her for their enemies and jump her.  She has no choice but to shoot these people before they shoot her.  But when you’ve killed all but the last guy, he will grovel and ask to be spared, or dare you to finish him off.  Ellie has no particular beef with these people, they don’t even know who she is, and up until this moment her reaction to their attack feels defensible.  But at the last you are not allowed to spare the survivor.  There’s no option to knock them out (a “strike” will slit their throat), or to tell them to take off, or to tie them up, or any of the other plausible non-lethal solutions to this situation.  If you do nothing they get back up and start shooting again.  It’s important to the designers of this game that Ellie actually kill this person, even though it isn’t necessary.  But the player’s agency must be suspended to enable this little bit of character progression, and it immediately feels like a betrayal.

Perhaps the designers assumed that most players would just kill the guy and move on without much thought.  Just iterate the game mechanic, point the reticle at the bad guy and pull the trigger.  But I think they hoped for more than that.  I think they hoped that the player would feel a little queasy, a little Shadow of the Colossus moment of inner conflict, when following the directions by route seems like the wrong thing to do.  And they are successful–it does feel like the wrong thing to do.  But they’ve also ensured that you have no choice but to go ahead and murder a person who cowers before you.  In doing so, the moral quagmire the encounter was designed to create is immediately defeated because there is no actual choice to be made.  Though these moments are few and far between, they undercut the power of the interactive narrative significantly.

The Last of Us Part II is an amazing game.  It offers deep, exquisite design and technical solutions to some of the problems that have plagued video games for decades.  The story is gripping, the game play is exciting, and the whole thing is beautiful to behold.  It smoothly blends gameplay and narrative in a way that few titles in the history of video games have ever managed to achieve.  But every once in a while the cranes are visible in the rafters, the strings pulling the characters forward catch the light, and we remember that it is, after all, an illusion.