On his blog, Raph Koster wrote an essay titled Narrative is not a game mechanic. His main point, somewhat over-simplified: a game’s narrative is not part of the player’s interaction with the game, it’s merely feedback for that player’s interaction. Whenever the narrative “content” outweighs the player’s interaction, that’s bad game design, because it ignores the strengths of games as a medium. Narrative-heavy productions may make for great experiences, but they’re not games.

Essentially, it’s the same arguments that have been repeated for years, but presented clearly and filtered through a more purely academic analysis. “Game stories suck.” “Games aren’t stories.” “If you put more emphasis on story than game design, you’ve made an interactive movie.” When I saw the essay linked on Facebook, I had to double-check the date to make sure that it was written recently and not pulled from the archives from 5 years ago.

I really don’t want to sound too dismissive, since Koster makes his points well, and it’s clear that the essay is more of a definition than an indictment. But I can’t agree with most of his assumptions or conclusions, partly because I think Koster uses an over-exclusionary definition of what constitutes a “game.”

A few days later, he addressed similar concerns with a follow-up essay: Narrative isn’t usually content, either. This one’s a bit harder for me to over-simplify, but essentially: he acknowledges that there is a category of games where the gameplay consists of piecing together a narrative, but says that the narrative is still being used as a resource, not a mechanic. You could change the individual pieces of narrative to fit any theme, but the actual game mechanics, what the player actually does with those pieces of narrative, don’t change.

That’s an insightful observation, but it still doesn’t do anything to address my problem with the claim that narrative isn’t a game mechanic. Most of my objections are rooted in his insistence that a “game” is by definition systematic, replayable, and deterministic.

If it were just a simple matter of definition, I wouldn’t really mind. He doesn’t use “experience” as a pejorative. He does use “interactive movie” as a pejorative, but I think we’ve all got a general idea of the distinction between a satisfying game story and an interminable series of cut-scenes interspersed with limited interaction. And if Koster were to say that, for example, a traditional adventure game was a great “experience” but not a good game design, then what’s the harm?

For one thing, it continues the tradition of drawing a clear do-not-cross line between “gameplay” and “storytelling,” which is bad for both. Back when I was working on narrative-driven games, I was looking for ways that games could be used to tell stories. But even then, I was starting to get more interested in systematic games, and I started to develop a greater appreciation for board game design. Now I want to understand how stories and games can work together.

Understanding how games work and how to design games as systems, instead of just cranking out minor iterations on the same set of established game genres, will only make for better games. There is one point that I agree with completely: it’s foolish to ignore or diminish the interactivity in a game, because the interactivity — the game design — is what’s unique to games as a medium.

But my main objection to Koster’s basic assumption is this: If you treat game narrative as nothing more than feedback, then your game narrative will never, ever be satisfying.

A digression about replayability

Most of Koster’s arguments reduce to the core problem that narratives aren’t repeatable:

There’s nothing wrong, to my mind, with using narrative as feedback. But we have to keep in mind that all that narrative and visual content is the expensive part of making the game. It is also consumable, whereas a systemically driven game system can provide many many problems to solve and heuristics to develop (and therefore fun to be had), with relatively few rules. Because of this, narrative content is destined to be expensive, short, and over.
[…]
You’re not going to get people to keep playing unless you keep releasing more content. This will matter quite a lot for any service-based game, be it MMO, F2P, social game, whatever.

I was a little baffled by his last statement, since “service-based” games are exactly suited to supplying more content. Downloadable content, in-app purchases, expansions, and sequels — an entire industry has been built on players wanting to buy more content. It’s very rare that these significantly change the game mechanics, and a lot more common that they provide more narrative. (For the simple reason that designing genuinely novel game mechanics is a lot more difficult and time-consuming than developing narratives).

Still, Koster insists that a systematic game that is indefinitely replayable is the ideal. That seems reasonable enough on the surface: if the player’s interaction isn’t meaningfully changing the outcome of the experience, then it would seem to be non-interactive by definition.

But consider a partial graph of a simple game, where the circles are the player’s decision points and the squares are the moments of feedback that lead to the next decision:

But whenever a player plays the game, he has to walk the graph:

The game itself is a system, but no matter how complex the game, the player’s experience with the game is always linear. Like a story. That’s just as true for the most systematic, abstract game as it is for the most linear, narrative-driven game.

You can never actually circle back to a junction point and make a different decision to arrive at a different outcome. Even if you go back to a saved game, you’re still arriving at the decision point with more information than you had the first time you made the decision.

I’d even take it a step further and say that because humans aren’t “stateless,” we can never truly replay a game from the beginning. Whether we’re re-rolling a character in an RPG or starting a new game of Tetris, our subsequent play-throughs are actually just a continuation of our last. Because this time, we know that enemy X is more vulnerable to heavy weapon attacks, or we’ve learned when and where is best to place the straight-line tetrominoes. All of our experience with a game lands on a linear timeline from the start of our first play-through to the end of our last.

I believe it’s a mistake to ignore that when we’re talking about good game design vs bad game design, to focus on an academic definition that requires replayability instead of looking at the player’s experience as a whole. I’ve got a board game called Macao that’s a perfectly enjoyable game, somewhat similar to Puerto Rico. It’s lightly themed, only slightly dependent on luck, and is primarily driven by a system of well-thought-out game mechanics (too many game mechanics, you could argue). I can’t imagine a definition of “game” that wouldn’t include it.

Still, I’ve played it exactly one time. I have no compelling reason to play it again as anything other than a pure diversion. Not because it’s a bad game, but because I feel as if I got everything I wanted out of it from my first play-through. It’s systematic, but I won’t get anything new out of interacting with its systems.

On the other hand, I’ve played Sam and Max Hit the Road at least five times, even though little has changed from one play-through to the next. It’s not just to watch the cut-scenes, either; I could go to YouTube if that’s all I wanted. Instead, there’s something inherently satisfying about putting the pieces of the system together and making them click, whether or not I’m still getting the “a-ha” moments of first discovery. Based on Koster’s definition, that’s just be a case of my getting excited about feedback. But I’d say it’s slightly deeper than that: it’s the interaction with the system that’s fun, just as much as that of a player going through the same course in a driving game even after he’s played it dozens of times.

To put it another way: I say what makes a game is the act of playing it. Being presented with a choice, deciding on one, and responding to that decision’s outcome. As long as the choices aren’t over-simplified, the knowledge that I could do it differently a second time has no bearing on how much I enjoy playing it the first time.

Weapon of Choice

All of my graphs above reduce the decisions a player makes — the inherent gameness of the game — to simple circles, each with the same weight. You could say that putting all the interesting decisions into what are effectively black boxes, and dropping them all onto the same line after the decision has already been made, ignores what it is that makes a game a game: the ability to choose between several different predictable outcomes.

Koster reminds us that our brains love feedback; it’s why slot machines are so profitable, and it’s why long strings of cut-scenes in a videogame can trick us into believing that we’ve actually accomplished something, even when we actually haven’t.

I’d make the claim that even more than the chemical jolt we get when we receive feedback, our brains love having the illusion of choice.

We hate the idea of choices being closed off to us. We love the idea that we’ll be able to explore every possible opportunity and get the chance to see every possible outcome. We agonize over restaurant menus, even when we know we’ll be back. We throw bachelor and bachelorette parties as a kind of wake to mourn a person’s decision to enter into a stable, committed relationship with only one person. We even value the potential of a choice as much as the outcome — the old expression reminds us that eating a cake means losing the option of having a cake to eat. It’s a big part of why we like stories, simulations, and games in the first place; they give us the chance to try out the choices we wouldn’t be able to make in real life.

The prospect of having multiple choices is so compelling, in fact, that we think of it as a virtue even when we’re not equipped to handle it. One of the terms you’ll hear frequently in regards to board games (and occasionally with video games) is analysis paralysis — it’s when a player has so many equally appealing choices available, she can’t decide which one to take. The best games circumvent this by introducing mechanics that limit the player’s choices at any given opportunity, scale the number of choices up or down as the game progresses and the player gets more information, or give context to the player so that some choices are more appealing than others.

So we love having plenty of choices with lots of possible outcomes available to us, even though we’re generally bad at predicting what those outcomes are going to be. And “anecdotal data” from my own experiences with games, and from reading and hearing other players’ accounts of games, leads me to believe that we value the potential of multiple outcomes more than actually seeing multiple outcomes.

I’ve heard plenty of accounts of players praising the Grand Theft Auto series by saying “I solved this mission using one method, but my friend solved it completely differently.” But I’ve never heard of players saying how great it was to successfully solve the same mission multiple times in multiple different ways. (I’ve absolutely no doubt that there are players who do exactly that; I would just categorize them as “completionists” and still say that playing the game that way isn’t the series’s main selling point).

And I’ve heard lots of accounts of players finishing BioShock, or any BioWare RPG that presents a series of binary good/evil choices, and then playing through again on the “opposite” path. But I’ve never heard anyone claim that it significantly changed the gameplay, or even that it was as satisfying as the original play-through. It’s more like my replays of Hit the Road — going back through a game I enjoyed, in an attempt to see what I missed the first time.

All of which leads me to conclude that it’s the player’s agency that’s most important. Koster says that

The bar that designers should strike for should include a rich set of systemic problems precisely because that is what the medium of games brings to the table. It’s what lies at the center of the art form.

That’s game design, which is the field that Koster is interested in. What I’m interested in is “interactive entertainment,” which is a combination of game design and storytelling, and therefore isn’t constrained by the definition of a “pure” game.

The appeal of a simulation is that it allows the audience to do both A and B. But a game, much like a story, has an end goal. What lies at the center of the art form is presenting the audience with the information to make an intelligent choice between either A or B, then showing the outcome of that choice. The act of solving the problem is what’s important, not the chance to see an array of alternate solutions.

The responsibility of the game designer is to present the player with a set of interesting choices. The responsibility of the storyteller is to make a sequence of interesting decisions and form a narrative from their repercussions. They’re not mutually exclusive. Anyone making a narrative-heavy game has to be good at both.

Press B Button to Ruin Game

So far I’ve side-stepped (I hope) the question of authorial control. All of the discussions I’ve read on that topic tend to be prescriptive: because games are by definition like this, you’re violating the integrity of the medium by doing that.

But I’ve played outstanding games that give the player absolute control over the storytelling (like The Sims). I’ve played outstanding games that give the player no control over the storytelling (like Half-Life 2). And I’ve played outstanding games that have no storytelling at all.

The only absolute must-follow rule is to be consistent. If you make a game with the intention of making the player the storyteller, then don’t impose narrative decisions on him without his consent (as in Grand Theft Auto 4 and a couple of unfortunate quests in Skyrim). If you’re making a game on the assumption that the “gameplay” is the focus, and the narrative is there solely to provide context and feedback for the game, then follow Koster’s advice: don’t let the narrative overpower the player’s actual interaction with the game.

In his example from Arkham City, Koster points out the problem of having a very small, very stupid game (turn camera and press A) followed by a huge cut-scene. But the problem with the Quick-Time Event isn’t just one of scale, but intent. It fails as both gameplay and narrative.

My interaction with the game is inconsistent with everything else I’ve done to that point: the A button doesn’t mean “leap through window and glide safely to the street below as the cathedral explodes behind me.” If I remember correctly, the A button means “jump.” And my interaction with the narrative is inconsistent with the story as presented up to that point: even though I’m ostensibly the Batman, and I could safely predict that I’d be walking into a trap, I was given no way to prepare for it or to circumvent it. I wasn’t an active participant in the storytelling; I was simply moving the character into place for the next cut-scene the designer wanted me to see. That’s not storytelling, it’s blocking.

Quick-Time Events happen when the developer wants to show the player an elaborate non-interactive cut-scene, but then feels the need to throw in some token amount of interaction because hey, videogames.

To fix the problems with the gameplay, you’d need to make the interaction more interesting and more consistent with the game as it’s been presented so far. Let me take any of the dozens of bat-gadgets I’ve acquired and learned how to use, and figure out a way to either defuse the bomb or open the window. That still doesn’t fix the problem with the storytelling, though. I’ve accomplished what I set out to do by making my way to the top of the cathedral, but I haven’t actually learned anything useful to advance the story; I’ve simply completed one in-game objective and been assigned a new one. In storytelling, you’re supposed to eliminate scenes that do nothing to advance the story. Even if they have cool explosions.

Sharing the Role of Storyteller

I’ve played plenty games that use narrative as feedback or “reward” for the gameplay. Solve a puzzle, get a cut-scene that introduces the next puzzle. Shoot a bunch of bad guys, get a cut-scene that introduces the next bunch of bad guys. It can be perfectly entertaining, but it’s still the least interesting type of interaction.

When Koster limits narrative to the role of just feedback, he also limits the amount of interaction the player has with the narrative: I’m making decisions in the gameplay that have little to no bearing on the story. He dismisses what I believe is most exciting about the potential of interactive entertainment: to create games where the gameplay is the story.

If you’re on board with the assumption that an indefinitely repeatable system with different outcomes isn’t required for something to be called a “game,” then it’s easy to see how a player’s walk-through of a game can be mapped directly to a plot-driven story:

Nothing particularly earth-shattering there; that correspondence is why game developers thought it would be a good idea to use games to tell stories in the first place. The interesting part is figuring out exactly how you map the game design to the narrative, and deciding how to divide the roles of “game designer” and “storyteller.”

First, you have to take the narrative and imagine it structured as an enormous game graph. (Or alternatively, start with your game design and assign narrative beats to it). Start with the setting, introduce the protagonist (the player), establish the protagonist’s goal (the win condition), and define what the protagonist can and can’t do (the rules).

Next, identify the significant plot points and translate them into decision points for the player. That’s the part that requires a familiarity with “pure” game design — determining how to present the player with a set of interesting decisions.

You could just stop there and turn the role of storyteller over to the player. Say that the player’s in charge of the decision points, the developer’s in control of everything else. In that case, the narrative is feedback for your game. I play a game, and when I accomplish an objective, I get to watch a moment in somebody else’s story.

Koster calls narrative a “parallel medium” to game design. But narrative is a lot more interesting when it intersects game design. To do that, I believe the developer has to share the role of storyteller with the player.

Earlier I said that a game designer presents the audience with a set of interesting choices, a storyteller decides among choices and pieces them and their repercussions into a sequential narrative. So in this case, the developer would have to:

1. Map the story onto a game graph
2. Walk the graph, deciding on a “best” subset of choices at each decision point
3. Encourage the player to reproduce the steps you took to walk the graph
4. Enjoy the accolades from your audience and your peers

My feeble attempts to translate this into the terminology of game design is making it sound more complicated than it really is. (It’s been almost two decades since I’ve been in an academic environment, so I’m not the best person to be talking about game theory). In the simplest, highest-level terms: turn the story into a game, “play” the game until you get the most satisfying result, and then encourage the player to replicate your play-through.

In the case of the Sam & Max games, it meant:

1. Figure out the narrative “arc” of the episode: typically opening, act 1, act 2, finale
2. Break that down into obstacles for the characters: getting into a room, getting past a character, searching for a key item
3. Since the games had only one solution for each puzzle, decide on the single funniest or most satisfying solution that we could think of
4. Putting context into the setup for each puzzle to make it feel as if our choice for the funniest or most satisfying solution was actually the best solution

Obviously, the magic happens (or doesn’t happen) in that last step. Adventure games are particularly susceptible to the “having to read the designer’s mind” complaint. But I believe that’s more due to the often ridiculous complexity of the puzzle solutions compared to the player’s actual interaction with the game, not the fault of the process itself. By which I mean: it’s kind of difficult to subtly suggest the idea “use sea slug on gong.” It’s not as difficult to suggest less specific things, like “this character is untrustworthy, you should check him out” or “that bridge looks rickety and is about to collapse.”

Most games do this kind of thing already, not just puzzle games. A huge part of level design is encouraging the player to do certain things instead of others: go this way, take this corridor, notice this object, climb on this ledge, get ready for the boss fight here.

Game stories do it as well, even though developers usually like to believe that they’re dropping huge unexpected plot development bombs on the player. In the Arkham City example: you’re Batman, heading to the top of an old building to meet the Joker, in the first 30 or so minutes of the game. Did any player, anywhere, not realize that he was walking into a trap? Instead of forcing me into a narrow corridor of cut-scenes and limited interaction, wouldn’t it have been more satisfying to reward me for realizing that I’m walking into a trap?

After the “read the designer’s mind” complaint comes the “interactive movie” complaint. If the developer is making all the so-called “best” choices for the player, then the player’s not actually interacting with the game. He’s just pressing buttons to advance to the next stage of the developer’s story.

But when you make a game that treats the narrative only as feedback or context for the game, instead of as an actual game mechanic, what you’re doing is making a non-interactive movie. It’s just one that happens to be separated by sections of interactivity. I get an objective, I kill a bunch of monsters, I watch the next scene of the movie to get my next objective.

That’s why I insist that a linear, non-systemic sequence of repeating decisions already made by a game developer can still be called a “game.” Because when it’s done correctly, the player’s doing exactly what the developer did while he was designing the game. During the design process, you start with a game state and a set of tools with predictable results, and then you have to decide “what happens now?” The “a-ha” moment comes when you decide “Of course! He should storm the rebel base!” Whenever I’ve hit one of those moments, it’s given me exactly the same jolt of satisfaction as when I’ve made a decision in another game.

If you were just to show the player a cut-scene, with the notice “New Objective: Storm Rebel Base,” then you’ve guaranteed that the player won’t get that same feeling of piecing together information to arrive at a decision. The narrative has no chance of being a game mechanic.

But if you give the player the exact same information of the game state, the exact same set of tools with predictable results, and an appropriately subtle nudge making the rebel base look like an awfully ripe target, then there’s a good chance he’ll get the exact same “a-ha” moment that the designer did. Assessing the game state, knowing the rules of the game, knowing the available moves, and making an informed decision about what to do next. How is that not a game mechanic?