I want to talk about The Last of Us Part II, or rather the conversation popping up around it, with the big old caveat that I haven’t played the game yet. (We didn’t receive a review copy, which is fair enough; reviews are not the main thing we write around here). For the moment, my interest is more in the broader traditions of design and critique that The Last of Us Part II fits into.

So to begin, let’s discuss the history of games that tell you you’re an asshole for playing them.

Video games in general, and big high-spectacle tentpole blockbuster action games in particular, tend to be violent by default, and there’s a now-longstanding tradition of breaking the fourth wall to turn an accusing eye on the player on this count: “You enjoy all the killing, that’s why,” Metal Gear Solid said to us—directly to the camera, confrontationally—in 1998.

In 2007, BioShock’s shtick was a bit different, drawing a direct line between committing acts of violence in-game and doing what you’re told. That was the thematic meaning of both its dual twists, first that Atlas was using you for his own ends and then that Ryan was Manchurian Candidating you with the codephrase “Would you kindly?” Your mileage may vary as to whether this is an incisive switcheroo or merely an act of callow dickishness. After all, there’s some flagrant disingenuousness in this approach—a Fight Clubesque dedication to making violence fun and engaging, and then saying, Ah! But actually: Not-that!

This the actual and frequently ignored importance of ludonarrative dissonance: What if the player already knows that violence is fun in a game but fucked up in real life? What if the player already knows that the game is leading them by the nose in all sorts of ways, both obvious and subtle? And what if the player’s enjoyment of the game is dependent precisely on knowing these things, feeling the friction, and then choosing the play along anyway?

In that case, to look the player in the eye and say You were willing to play along?! You monster! is at best cheap. (At worst, the twist effectively eats the rest of the plot; at worst, when you tell the player that they were dumb for giving you the benefit of the doubt, they believe you). It’s a narrative parlor trick that becomes a little less effective each time a player sees it, both because it becomes less novel and because it depends completely on being a surprise.

Indeed, the very same year as BioShock, Portal contained what probably should have been the Blazing Saddles moment of this trope, with the Weighted Companion Cube. For those who haven’t played Portal, briefly: You get a cube with hearts on it, and you carry it with you throughout a level, using it to solve puzzles. All the while, you are told via voice-over that this cube is your loyal friend (but also an inanimate object, you sentimental dolt). At the end of the level, you’re told that you have to drop Companion Cube into an incinerator in order to proceed. And you really do. There’s no way around it. You monster.

I call this the Blazing Saddles moment for the “You enjoy all the killing, that’s why” trope because it’s simultaneously affecting and absurd—to the point that, just as it was more or less impossible to make a straight-faced Western after Blazing Saddles laid all the genre’s hypocrisies bare, it shouldn’t have been possible to say “You enjoy all the killing, that’s why,” with a straight face, post-Companion Cube.

On the one hand, a lot of players (me included) really do feel bad for the fucking cube. And even those who don’t will nonetheless resent being told what to do, which is where the absurdity comes in: Whereas BioShock’s twists rely on you not noticing that you’re being railroaded, the Companion Cube joke expects you to notice both that killing your Cube is an act of rococo cruelty and that this is only true because the game told you so. It foregrounds the artificiality, even the arbitrariness, of our affective relationships to virtual people (who are, after all, no less virtual if they’re shaped like people rather than cubes).

For my money, a twist that calls the player a monster only works well (or at all) when it reframes the game’s themes, as Companion Cube does. The twist in Braid isn’t just that the protagonist is a jerk, actually, but that his perception of time—the whole basis for the game—has enormous, tragic drawbacks. (There’s something in there about how treating the world as a puzzle to be solved can blunt a person’s empathy). The twist in Nier isn’t just that the antagonist has a sympathetic motivation, but that he and the protagonist have the same motivation, even though their concrete goals are nonetheless mutually exclusive. (There’s something in there about our infinite capacity to justify violence against a capital-o Other).

You-Monstering doesn’t need to be limited to a single moment or twist, either. Shadow of the Colossus famously employed a slower burn, peppering your adventure with melancholy and dread. There’s a rising squall of tonal cues that your quest/killing spree might be doing more harm than good, but more than that, that your whole hero’s journey is—and by extension, maybe all monomythical hero’s journeys are—selfish to the point of literal solipsism.

But once we’re out of the realm of parlor tricks and into the realm of theming, we’re skirting a different (though related) category: games that feel bad to play in an interesting way. This tradition goes back at least to Missile Command (1980), which used its inevitable fail state to argue that war is inescapably disastrous. If we’re willing to include tabletop, then the tradition really goes back to The Landlord’s Game (1901), the original, overtly anti-capitalist version of Monopoly—a simulation of rent-seeking that always, pointedly, descends into a protracted death march of financial attrition. (Modern Monopoly sucks in about the same way, but probably not on purpose. Its misery is vestigial).

More recent examples of good feel-bad gameplay often come from games that changed direction partway through development. Spec Ops: The Line and Demon’s Souls were both troubled productions that were probably past the point of ever feeling good in a AAA sense, and which therefore elected to take a hard turn into saying something with their rough edges, their abrasiveness, their jank. (The latter conceived a sub-genre in the process).

And of course, indie games and altgames are where the wildest and most focused feel-bad auteurism lives. Nathalie Lawhead’s RUNONCE takes the idea of killing a virtual being to its logical conclusion in a really affecting way; Camilia Gormaz’s Nanopesos is a succinct, grounded simulation of being in a genuine no-win situation; Austin Jorgensen’s LISA revels in giving violence meaningful, but not always fair or predictable, consequences. And we have to mention everything Ice-Pick Lodge has ever made, as well as Lucas Pope’s Papers, Please, and its tabletop cousin, Brenda Romero’s Train.

And I’m still leaving out entire genres of finely-crafted feel-bad gameplay, from traditional roguelikes to out-and-out masocore platformers—a sub-category that we could call brutality unto hilarity. Roguelikes dangle the possibility of permadeath over you at all times, a more credible threat than most games can muster; masocore games assault you with a seemingly limitless barrage of tailor-made misfortunes, wielding your expectations against you.

Nor is it anything new to feel disempowered in a survival or horror (or survival horror) game. These are genres that frequently feel bad on purpose, genres that make a meal of their oppressiveness. But it used to be the case that, once a horror series reached a certain scale, it would cease to be horror as such and default to something more like straight-up action. (This was certainly the case with the shift from System Shock to BioShock, and from Dead Space to Dead Space 3. And hell, the Resident Evil series only recently returned to form after 5 and 6 took an extended sabbatical from disempowerment).

So feel-bad games are a grand tradition, but The Last of Us (the first one I mean) is a bit of an odd beast even so. It’s a lush blockbuster production that is, quite intentionally, absolutely miserable to play. A Nathan Drakey guns-akimbo yippie-ki-yay approach means certain death, out-and-out conflict avoidance will leave you without necessary supplies, and methodical, one-shank-at-a-time stealth is a joyless grind (super-powered, but stubbornly unheroic and ceaselessly mopey, like Zack Snyder’s Superman). The plot is shot through with hideous misfortunes and (to put it mildly) regrettable decisions, albeit punctuated with little moments of quiet, numinous grace. It’s not miserable through and through, but it’s so deeply miserable as a rule that I have to think the misery is the point.

And it’s not merely a feel-bad blockbuster, but a feel-bad Naughty Dog game. That’s a remarkable thing, because Naughty Dog has a well-earned reputation for unfathomable polish. Even in a AAA context, there’s an impossibly high standard to their games’ character animation, level design, and environmental detail. This is possible specifically because Naughty Dog’s productions are rigidly linear (and relatively short, though The Last of Us Part II apparently breaks the latter convention). There is simply no practical way to be on the vanguard of lavish spectacle in a game with more branches—or to put it another way, a game with more choices.

The obvious tension there being that games are, you know, interactive. The player could always, at any moment, do something other than what they’re supposed to do. Naughty Dog’s games leave very little room for what Dan Olson has called “intentional disobedience.” As such, they demand that the player play a role in a sense exactly opposite to what we usually mean when we use the term “role-play.” Here role-play has nothing to do with constructing your character, or with deciding how they’ll act or how they’ll grow, and everything to do with falling into a role that has been fleshed out for you, in advance.

When I spoke to Part II level designer Evan Hill last year, he discussed this in terms of, well, level design: “We valve you pretty aggressively,” he said, valves being the points in a level wherein, once you’ve gone forward, you can no longer go backward. And naturally this approach, with its frequent, quietly enforced points of no return, extends to the studio’s philosophies on narrative.

Here was Evan’s example: The player, as Joel in the prologue of The Last of Us the First, is supposed to carry a girl out of a burning house. The player could choose instead to stand perfectly still, or to run directly into the fire. If the player has control of the character, then there’s ultimately no stopping them from doing those things. But the game won’t respond in any interesting way to those other, stupider choices. And more to the point, the whole scenario is composed to make you want to play ball with the narrative. There’s only one reasonable thing to do as Joel in that situation—get the hell out of there—and indeed, said Evan, “the story is written so that it wouldn’t make sense to do anything else.”

The idea is to make every moment “livable,” with player and game and designer all on the same page, performing the game together in a way that feels natural, even inevitable:

The real trick of a Naughty Dog Game is not that oh, everything is linear and built perfectly and there’s no choice. It’s that, in the moment we present to you, there are no reasonable choices other than the one that’s actually going to progress the plot.

This is the heart of why Maddy Myers had such a bad time (in a bad way) with The Last of Us Part II. She found herself painfully, constantly aware that the choices progressing the plot were by no means the only reasonable ones, or even especially reasonable, period.

What’s worse is that the characterization of Ellie makes it seem like she should also understand this part of the journey. I kept expecting her to grow and turn away from a life of constant violence, but she never picks up on the obvious didactic nature of the game she’s in, even as the designers beat you over the head with a very simple lesson about the value of human life.

Ellie just didn’t seem far enough gone to keep making the terrible choices that the game needed her to make in order to proceed—and so Maddy Myers, as a player, resented having to act those choices out. By contrast, Joel’s big final choice in the first game is, for a lot of players, perfectly in character—even if it’s unreasonable in all sorts of ways, and perhaps even unforgivable. The player doesn’t have to consider it an inevitable decision in order to accept it as the decision that Joel would inevitably make.

Which is the crux of it, right? This stuff is subjective. What a character would do in a given situation. What you would want to do in that same situation. Where the two conflict, for one player but not for another.

That shouldn’t be an especially hot take, should it?—that art is subjective, that a technical marvel can nonetheless be a bad piece of art by a lot of people’s lights. But mainstream games journalism is in a transitional moment where that aspect of criticism goes only half-acknowledged. That’s how The Last of Us Part II can get scathing reviews from several major publications, and yet sit pretty at a 95 on MetaCritic. This game’s bad reviews are by and large unscored (and therefore excluded from the meta) specifically because they’re by and large reviewing the game as a piece of art rather than as a product.

And The Last of Us Part II is both a work of art and a product, to be clear. No contradiction there. But the language of shiny new products—so new! so expensive! look at all the labor that’s gone into it!—and the language of bleak, difficult works of art—are people this bad, deep down?—mean that a lot of writers are writing about the same topic but writing right past each other.

It might simply be impossible to make a one-size-fits-all series of livable moments, because something in there is going to feel dissonant to someone. The designer will have more latitude, and create less dissonance, if they can convince the player that it’s interesting to be Joel, or Ellie, or whoever. But they’ll never convince everyone. Some players won’t want to be the person you’re asking them to be, certainly not past a certain point.

So the strangeness of The Last of Us (and Part II, by the looks of it) isn’t exactly about linearity or choice or agency or whatever. It’s about the intense performative demands that the game is making on you as a player by curtailing your narrative agency: Not only to faithfully act out your part, but to buy wholesale into the creative team’s vision—to tune yourself to the same frequency as the work, to agree for a time to operate on some shared baseline assumptions about human nature. It’s not just a question of doing what the game tells you, but of believing what the game believes.

That demand isn’t unique to the Last of Us games, but their signature blend of pervasive misery and aggressive valving makes the demand unusually hard to ignore. Then again, that’s why feel-bad games are interesting in the first place: They ask you to be someone definitely and resolutely other than Captain Power Fantasy. They ask you to believe uncomfortable truths.

We reach a point of absurdity when we try to apply the language of the newest, the best, the slickest, the shiniest, the most innovative, to a work that ultimately lives and dies more by whether its worldview rings true. It’s not merely that different people are getting different things out of this particular game. It’s that we’re having a disagreement about what you can or should expect to get out of a game—any game—in the first place.