Teaching to Fish

This talk was given as a joint keynote for the Sweden Games Conference and VS Games, in Skövde, Sweden, on Wednesday Sept. 16th 2015.

Slides

Video

PDF

Transcript

How many of you are here for Sweden Game Conference? Almost all of you. How many of you are here for Serious Games? Oh. How many of you are here for both, then? Okay, all right.

I decided that maybe I should talk a little bit about how games teach—not just how they teach in the boring “serious games” way, because a lot of serious games are kind of boring, but also how they teach people who play ordinary, everyday games: the kinds of games that are set up back there.

There’s an old saying: it’s better to teach someone to fish than it is simply to feed them. Although they’ve been feeding us fish here, and it’s much appreciated. Hopefully the conference is teaching us to fish in some way. It’s teaching us to make games better. It’s teaching us to practice our craft in more sophisticated ways. We learn things by interacting with one another.

Of course, we very quickly discover that there’s far more for us to learn than we can ever possibly get hold of. The thing that’s a little distressing is that, in practice, we have fairly little codified knowledge about how to make games. There’s a lot of stuff that we pass on through apprenticeship—not even teaching, but apprenticeship. Old-fashioned medieval guild apprenticeship, sometimes complete with the indentured-servitude part.

So it’s worth thinking more deeply about exactly how we make games, what games do, and what games mean. We’ve already had several talks in the last couple of days about that kind of thing. Mike Sellers, for example, spoke about game loops, atomic structures, flows, parts, and so on. We’ve had several talks referencing the fact that a lot of educational games aren’t very fun. All of these subjects have been coming up.

But it’s a keynote, so I thought, “Well, you don’t really want to see diagrams.” Instead, I thought I would tell a fairy tale.

THE FOUR FISHING GAMES

How many designers are in the room? All right, you’re in the front row, so you’re going to be Designer Number One. Sorry. Any others? Who else? More designers? All right, you’re Designer Two. You’re Designer Three. Any more? Please.

All right, no, Josephine—we need a woman. We really should. Drop one of the guys who was back there. You’re out. You’re in.

You are all bringing me games about fishing. Sorry if you don’t like fishing; you’re bringing me a game about fishing.

Designer Number One, you’re going to show me this awesome custom controller you have designed for the game. Tell me: how are you using the accelerometer in order to make it really cool?

[Audience response.]

Did you not play Sega Bass Fishing? It had a custom controller. So why did you decide to put a real fishhook on it?

[Audience response.]

Realism, I guess. Right. We’ve seen this game: the fishing game that is all in the controls. That’s a perfectly valid kind of game. There’s nothing wrong with it. I mean, don’t ever make a game that relies on a custom controller—it’s business suicide—but other than that, we’ve seen that game, and it’s kind of interesting.

There are a lot of games that depend on custom controllers. Everything about the Wii was based essentially on control patterns. The games themselves were not very sophisticated; it was mostly about the controls.

Now let’s see. You’re next in line, so tell me about your fishing game. You gave me this giant spreadsheet of all the different things you can pull up. I see tuna, I see sardines, and I see boots and treasure chests. Why did you do that?

[Audience response.]

To find new things hidden at the bottom of the ocean. Yeah. But how do I get the boot instead of the tuna if I really want the boot?

[Audience response.]

Oh, that’s too realistic. It’s too realistic. Okay.

Who was next? You were next. So this pitch is all about sitting by the dock and talking for a really long time with my dad, who is imparting life lessons that are supposed to be meaningful. He keeps drinking beer, which I’m not old enough to have yet. Can you tell me why you spent two years on the chat system for your fishing game?

[Audience response.]

So you’re really about the experience here. You really want to capture that feeling of utter boredom.

Josephine, I understand that you can tell me the exact caloric intake of every minnow in this river.

[Audience response.]

Of course. You need to get everything correct. You need to have every detail, because then you can learn as much as you want. If you’re a real nerd, you can go really deeply into it—and we all like nerds, right?

So how are you solving the problem that I can’t see the water for all the graphs in the way?

[Audience response.]

Why would you want to see the water?

THE MAGIC FISH

Fortunately, something comes to the rescue. This is when the magic fish appears.

The magic fish says, “You know what? Each of you has made a game that represents one part of fishing. They’re not bad kinds of fishing, but none of them necessarily encompasses everything about the experience.”

That is what we normally do when we make a game. We have to choose to represent only certain parts of the thing we are mimicking, and we choose to emphasize different aspects of it. In this case, whether it was the controls, the surface experience, unpredictability and chance, or a rich and deep simulation, we’re actually building radically different games about the exact same thing.

In fact, these four games could even look identical and play completely differently. That is entirely possible.

So when the producer suddenly comes along and says, “I don’t understand what this game is,” we can have real trouble communicating it to an audience, to management, or even to fellow members of the team. These days, if you can’t describe what you’re doing, the whole game can spiral out of control. It’s actually pretty easy for that to happen, particularly as teams get bigger.

Then, all of a sudden, the magic fish appears emergently out of this mess of a system we’ve built, and we generally don’t even know how to cope. Then the producer says, “No, that’s the part I like. It stays.”

He’s exhausted. It’s 3:00 a.m. You report the magic fish of emergence, or whatever other weird thing is now in your game, and he’s so tired that he says, “It’s not a bug. It’s a fish, sir.” Sorry.

SERIOUS GAMES AND HIDDEN LESSONS

Here’s the thing. When we look at games, it can be incredibly difficult to convey what we’re trying to get across. The place where this is super important is in serious games.

In serious games, we typically make games that fail to get across both the lesson and the game. That happens all the time. We either spend too much time chasing the lesson and fail to get a game, or we spend too much time chasing the game and fail to get the lesson across. It’s really easy to make that mistake.

But this also happens in regular games, because in regular games we don’t tend to stop and think about the fact that we are teaching things all the time. That is actually what games do. The most fundamental thing games do is teach you how to play them.

So we come back to Designer Number One, and I guess the critique is: you have this awesome control system, but I’m pretty sure that after I cast the rod, you literally just spawn the fish on the fly to make sure I catch something. Right?

That’s actually how most fishing games work. It’s also how we spawn most encounters in AAA games today. There’s no there there. There’s literally no lesson being taught except how to maneuver the controller.

When was the last time you played a racing game? Have you noticed that knowing where to go with your car is no longer part of the game? They put a stripe on the road telling you how to drive optimally.

[Audience interjection.]

Are you trying to derail the talk now? Okay, moving on.

One of the key things racing games used to teach was how to take curves optimally. But we have actually taken that lesson out and solved it for you, in the name of giving you a better experience.

Of course, that isn’t how fish spawn. We are providing a very false picture of reality, and an underlying lesson that says, “If you just do what I tell you, a reward is going to pop up.” If you think about that, it’s actually kind of a pernicious lesson to be teaching people. Nonetheless, it is an implication of that game design. It’s a lesson we aren’t taking the time to consider.

For that matter, if we’re just hitting a loot table, there isn’t any skill involved. If there’s no exercise of skill whatsoever, all we’re doing is leveraging a bug in the human software system: the fact that we’re lousy at projecting odds.

Plus, it doesn’t give much credit to the fish. If you have ever tried to catch a fish, you know they’re really not that stupid. It’s actually a little hard to catch one. They dodge and that kind of thing.

Reducing everything to chance is a super-easy design tactic. It’s an easy design trick, but it also teaches us false lessons.

Then we have the chat person, who says, “But the experience is incredibly important.” Many, many games these days are built entirely around experience, and it could be argued that the experience trumps whatever is in the mechanics. Lots of people make that case. Certainly the entire AAA industry makes that case, because there are hardly any mechanics in AAA games anymore.

We spend most of our budgets, most of our time, and most of our effort in large games creating experience-based games. Further, this has trickled down into indie games quite a lot. Indie games these days are heavily narrative as well.

The interesting thing is that it’s actually a bit of a trap. Narrative is expensive, so from a business point of view, you have to be super careful about it. And of course, there is definitely no teaching somebody to fish if what you’re doing is simply aiming for the cool experience. No tuna sandwich for you.

Then, of course, we have Josephine’s game, which says, “No, but I’m teaching you all about the fish. This is the real one.”

This is the one that is mechanically rich. It’s the simulation. It’s what leverages the power of computation and all the rest. There’s certainly a case to be made for it. This is where classic systems design lives, and it makes the fish feel great.

But it often doesn’t make the player feel great, because if we can’t tell that this rich simulation is happening, we won’t learn a damn thing. If we can’t get the surface to convey the message of the mathematics behind it, it isn’t going to teach us anything.

Each of these games is teaching lessons. Some are teaching implicit lessons that we didn’t intend. Some are teaching lessons that we did plan. At the same time, they’re often failing to teach us critical pieces of information.

The answer isn’t necessarily to make the game with the custom controller, the voice-chat system, and random chance layered on top of a simulation, because you’re never going to finish it. Trying to do everything is going to produce a game that doesn’t have a point.

And if there’s one thing that any work of art—and yes, I’m going to use the word art—needs, it is a point: a reason to exist. Otherwise, you might as well go fishing instead.

It’s not that any of these games is wrong. The key thing we need to be able to do as creators is make the right game for the right situation.

I’m impressed nobody is groaning. There’s a reason for that, right, Mike [Sellers]?

THE FOUR GREAT AREAS OF GAME DESIGN

Designer Number One, in your case, you have actually made a game entirely about haptics.

Haptics are literally the mechanics of our movement and how they interface with a machine. That’s what they are. And that’s great, because haptics are also pretty much all of sports.

We are a machine, in one sense, interfacing with the physical world. Learning how to master aspects of our own bodies and our own interface and control systems is one of the four great areas of game design. It’s actually one that we tend to neglect. We tend not to think about sports as part of our discipline, and that’s wrong.

Designer Number Two, you made a game about chance. There’s nothing wrong with chance. Chance is an enormous equalizer. Chance is a thing that gives the young and the inept the opportunity to learn and improve. It is incredibly powerful.

It’s just a little bit like sugar. It’s great in moderation, but if you make your whole meal out of it, it’s kind of empty calories. It’s a lot of fun in the right doses, and it is in fact one of the four great areas of games: chance-based games.

Designer Number Three, you made a game that was really about other people. That’s wonderful.

Finding courses in multiplayer game design is hard. But I personally have never seen a course at a game school on sociology, team building, anthropology, and so on. Those are the fields that connect to what you’re doing here. Psychology, and so on. All of a sudden, we’re getting into pretty deep waters.

This is also an area that games as an industry have historically chosen not to ignore, because it makes lots of money—but how much do we think about it as design? Not very much. Otherwise, we wouldn’t end up with all of the social problems we have. We wouldn’t end up with voice chat on Xbox Live.

Social games—not in the Facebook sense—are in fact the third great kind of games.

And, of course, the fourth kind is games about complexity.

When we talk about complexity here, we mean it in the strict mathematical sense. We’re talking about problems and systems that are intriguing in their own right. They’re like complex machines. They’re like mobiles hanging from the ceiling: we poke one part, they wiggle in interesting ways, and we try to figure out why poking it here makes it wiggle like that over there.

That’s great. This is traditional, classic systems design. And frankly, as game designers, we have radically failed over thirty years to codify what we know about this as a craft. That was kind of what Mike [Sellers]’s entire talk was about yesterday, because it’s a really hard problem.

Each of these is a very large problem. In some sense, they’re all about complexity in different ways. They’re all about interacting either with a complex system—people are complex systems—or with the illusion of a complex system, in the case of chance.

As I mentioned earlier, chance is actually a bug in our brains. We deal with probability by thinking there is a pattern. “I have a system for the horses, I swear.” But there isn’t actually a pattern there. Our brains fool us.

Some anthropological researchers speculate that the reason for this is that, if we were good at calculating odds, we would never have left the cave, because the odds of that saber-toothed tiger getting us were actually pretty good.

Our brains are terrible at chance, but it’s so easy to use that we go ahead and use it all the time. And that’s fine. What we often don’t do is make a considered choice about when to use one kind of mechanic, one kind of system, or one kind of approach rather than another.

The fact is that they serve very different purposes, and they have very different strengths.

HAPTIC SYSTEMS, COMPLEXITY, AND SOCIAL PLAY

Haptics are the realm of roller coasters. They are the realm of cool motion-based controllers. They are the realm of Dance Dance Revolution. They are the realm of playing football. They are running around and playing tag with your friends—although, of course, tag is also a social game.

There are lots of reasons why the body wants us to practice using it. A lot of these things boil down to survival skills. Games, in a lot of ways, are about making sure the world doesn’t eat children too quickly. Kids need to play games to learn how to walk. They need to play games to learn how to navigate a piece of food onto a fork, and from the fork into their mouths, without stabbing themselves in the eye. So we make a game out of that.

Kittens do it. Puppies do it. This is how all mammals work: they play games in order to learn how to solve these problems.

We also need to learn how to solve mathematical puzzles. By and large, playing catch involves something rather like calculus. Being able to predict motion is what you need in order to catch a ball, so you have to learn how to do that at a very rapid rate.

You need to master non-Euclidean geometry in order to play Snakes and Ladders, because that space is absolutely not a two-dimensional Cartesian coordinate system.

In fact, mathematicians have told us that most classic board games fall within a formal mathematical complexity range between PSPACE-complete and NP-hard problems. You can go to Wikipedia and find the page on NP-hard problems, and it’s like: oh, there’s Sudoku, and there’s building a raid group in World of Warcraft, and so on down the line. That is actually at the heart of game systems design.

Unfortunately, it’s the kind of thing you don’t really encounter until you have completed five or six years of mathematics study. We really need to be teaching it in freshman game-design classes, because it is at the heart of game systems design.

Then there are social problems. Almost everything we do in the world when we relate to other people is about making sure our tribe comes out ahead, and that the people in our tribe know that we are the monkey on top who gets to fling the most poop.

An enormous amount of social behavior is about establishing relative status, mentorship, tribalism, and identifying the people who are not us. It’s kind of baked into our psychology to dislike people who are not like us. We also have enormous brain bugs around simply going along with anybody who does look like us.

Door-to-door salespeople use this. Marketers use it all over the place. There is a pile of known techniques that exploit these things.

THE CLASSIC LOOP

Most games make use of a little of each of these areas, but all games share a common classic loop.

They start with the player needing to decide what to do. The player needs to be able to tell whether there is an affordance. That is a term from product design and psychology: whether there is something on the screen that they recognize they can press or manipulate in order to create an input.

The player maps that affordance to a verb. The verb goes into a black-box system of rules that mediates the simulation and the interaction, typically with an opponent or another player. That could be an AI. It could be another person. Whatever the verb was, it goes in and causes a simulation update, which then results in an opponent response.

This changes the state of the environment. Then we send back some amount of display information that tells the player what changed. Maybe we send the whole map. Maybe, as in chess, the player gets to see the entire board. In other games, perhaps all the player sees is a bullet flying toward them.

The result is that various forms of narrative are constructed. A huge amount of that narrative is something we build ourselves while attempting to make sense of the universe, because that is how our brains work. Some of it is probably narrative that the game designer is throwing at you in a classic storytelling attempt at mind control, because that is what stories are mostly for.

Then we use all of that information to form a new sense of intent and rebuild our mental model. Rinse and repeat, sixty times a second.

Actually, not sixty times a second. Human response time is much slower—on the order of a few hundred milliseconds—so we only get to make a handful of meaningful decisions per second.

But if we want to become good at our craft, that simple model is inadequate. It is hugely incomplete.

That seems really unfair, especially if you are an undergraduate game designer. How many fields did I just go over? And that is not even close to what games actually look like. Not even close.

Do you want to know what games actually look like? Do you want to get really depressed?

THE EXPANDED MODEL

First, you do have that mental model, and that’s good. But when you approach a problem, you usually do not have a solution. You try to form a heuristic: an approximation of a solution to the problem.

Using that heuristic, you form an impression of the designer’s intended objective. That is not necessarily the same thing as your own objective.

The information about that objective may provide you with high or low agency. That is a massively consequential choice, because high-agency games and low-agency games teach radically different things and use radically different craft techniques.

Finally, you form an impression of what your own goal is.

In Minecraft, which is a high-agency environment, your goal has barely been dictated to you. There is hardly any sense of an objective. You are told, essentially, “Don’t get killed by a Creeper,” and not much else.

In a low-agency game like Call of Duty, you are told, “You have to go out through that door within two minutes and thirty-five seconds.” In that case, objective and goal line up pretty well.

But even then, once you form the goal, you have to find the affordance, identify a verb, and figure out how that verb maps to an input. Maybe your keys were remapped. Maybe you are asking, “Where is the stupid red X on this controller?” Whatever the problem is, you finally manage to perform an action.

The action gets fed into that black box of rules. Inside that black box is a complex structure that I have taken to calling a ludic structure.

A ludic structure is any kind of system that reaches a certain complexity threshold. It isn’t just the rules of chess. The laws of physics are a ludic structure. Music is a ludic structure. The system of complex waveforms vibrating in the air is a system complex enough for us to play with.

It is not an accident that we play an instrument. It is not an accident that, when we do theater, we put on a play. Those are also systems complex enough to have unpredictable, emergent results.

Once we find a ludic structure, we begin to see the different categories I have been describing. The thing they all have in common is that they offer a broad enough possibility space.

If they are trivial, we find them boring. If they are too noisy, we also find them boring.

There really isn’t anything totally noisy in the world. Even the static on a dead channel is mathematically predictable if you happen to have enough data about local magnetic fields, interference, the weather, and so on. We simply cannot process it, so we call it noise.

Noise is just data we do not have the equipment to understand. There isn’t really any noise out there. That is our failure.

When we find a system that is complex enough, we begin to try to interpret it. So what lies at the heart of interesting systems is interpretability: the idea that perhaps they have some kind of inner life, that perhaps there is something inside them talking back to us in an interesting, complicated way.

We anthropomorphize everything.

Once you understand that, it is fairly easy to see why gamification usually fails. It is usually stacked on top of non-ludic systems: systems that are trivial. “Fly enough and I’ll give you more points.” There is no interpretability to that. I interpret it as, “You want my money,” and I am done.

That quickly leads to boredom and the failure of the game. Gamification tends to fail because it tries to use the rest of this loop without implementing the hard part: the part we lack the most language and knowledge to talk about, which is the ludic structure.

LUDIC ARTIFACTS

Ludic structures are sometimes designed, but we did not design physics. If we design a real-life sport, we can effectively write “Import real-world physics DLL” at the top of our game design. Somebody else already wrote that code. That’s great, because so far no major bugs have been found—although we do not understand it all.

I call consciously constructed ludic structures ludic artifacts: the deliberate attempt to build a system like that.

Those artifacts fall into a number of different kinds of problems.

The first is the problem represented by the cool controller. It is a visceral problem. It is a haptic problem. It is the problem of interacting with our own selves.

We have very few resources on this enormous subject. Thank God one of them is excellent: Steve Swink’s Game Feel.

Keep in mind the kinds of things hidden underneath this. Every single best-selling platform game in history gives its characters a remarkably similar amount of airtime during a jump, and nobody planned that.

Isn’t that weird? Mario’s jump timing feels like the jump timing in many other successful platformers. And guess why. We unconsciously tuned the games for feel and matched the real jump time of humans.

Talk about not knowing what we’re doing. Talk about being completely unaware of the systems underneath things.

TRAINING AND AUTOMATICITY

Feel feeds back into the notion of training. That is what most reflex-based, visceral problems do. We use them for training. The military has known this for years. Athletes have known this for years.

We use training to turn actions into abstractions, chunks, schemas, and patterns. We internalize them all the way down to simple things.

If I ask you which leg you put into your pants first this morning, you probably don’t know. We internalize actions to that degree. We run them as background processes.

But visceral problems are only one kind. We also have interaction problems. We have the chat game. Hidden behind chat are social problems.

Social problems are about othering. They are about identity. They are about mentorship. They are about tribalism. They are about hierarchy. Pretty much everything we do with other people.

The number of relevant academic fields is enormous. You can get PhDs in psychology, anthropology, cultural history, and so on, just to try to understand how to build the economic system in your MMO properly. You have to understand these fields, or you will fail.

Then, of course, there are complexity problems: game systems.

One interesting thing about complexity is that it is relative. To a little kid, tic-tac-toe is pretty complicated. Then there comes a moment when you cross a threshold and it becomes trivial.

Hidden behind that are the systems of rules that lead into complexity problems. Sometimes those systems are designed. Sometimes they are created through a folk process, as most board games were over the course of centuries.

They lead us into complicated problems involving topology, non-Euclidean space, simulation, stochastic and nonlinear behavior, economics, and much more. All of those problems lurk back there.

Then you have chance, which mostly dead-ends, thank goodness.

INFORMATION, METAPHOR, AND STORY

After the system updates, we still have to tell you what changed. We have to give you the state—the delta, the things that were modified.

We have to give you a view of the overall state and decide whether to present perfect or imperfect information. That can change feature by feature. It can change control by control.

Here is one of the hideous realizations you reach when you get this far down the curve. You start thinking: wait, pressing the fishing-controller button was a haptic game. Talking to my dad was a social game. Then I have my chance-based complexity problem—what am I pulling up?—running against the nonlinear simulation.

You start realizing that every single action you take in a game has all of these things. Every verb. Every button press. Every single thing you do.

From imperfect information, you start cloaking things in metaphor, because people learn better through metaphor. They learn better through stories.

You have to read every book Edward Tufte has written on information design in order to learn how to communicate these complex mathematical landscapes.

From that, you start forming a sense of story, narrative, and myth. You combine it with the kind of story the designer is trying to tell.

These days, of course, that can be a pretty imposing kind of story.

Stories are primarily meant for enculturation. When you watch a fun comedy, the fun comedy is trying to brainwash you into agreeing with the norms of society. We have used stories for that for centuries.

Stories are primarily about cultural transmission. They are primarily about building a certain degree of conformity and a certain degree of empathy among people within a society.

Empathy is their sweet spot. It is what the great stories do.

EMPATHY, INTERPRETATION, AND ART

You take empathy, and that leads you to the notion of play from the semiotic side rather than the fun side: the notion that signs and signifiers do not quite match up, and that there is space for interpretation.

All of a sudden, we connect play and art all the way back to the possible inner life of a complex system.

We begin to realize: when I try to figure out how the river works on the inside, I am also trying to figure out how my dad works on the inside.

We start realizing that this notion of play—of possibility space, of interpretation—lies at the heart of understanding pretty much everything.

That leads us to art. It leads us to the MDA framework and the notion of dynamics. From that, we start building intuition, because almost none of this happens at the conscious level.

We use it to build background learning, and the reward our brains give us for that background learning is what we call fun.

But fun isn’t the only thing you can do with a game. You can practice with it. You can train with it. You can use it as a form of exercise. You can use it for meditation or comfort.

How many of you have sat down to play an old game you have beaten many times because you had a cold and you just really wanted to visit Mario again? Right. You can use games for comfort.

You can use them for meditation. You can use them to put yourself into flow. Or you can use them as yet another way to be bored.

THE DESIGNER’S LIBRARY

This is the moment when the designer says, “Holy crap. How long am I supposed to study in order to do this?”

All of a sudden, a dual emphasis in high-end 3D rendering and distributed network programming sounds easy.

Yes, I am here to tell you: if you want to tackle the toughest problems in the game industry, game design is where it’s at. But you’re going to need a really big library.

That’s what you do. I have no shortcuts. There aren’t any.

But you come out the other side—or, really, you get partway in, though it looks like a stopping point—and you say, “Wait. That means my game with this custom controller has a unique, magical strength. It teaches a kind of lesson that none of the other fishing games can teach.”

None of those other games is going to teach me the precise number of degrees I need to twist my wrist. That might not seem incredibly useful unless, of course, you make your living fishing, in which case it is critical.

We shouldn’t denigrate that. Sometimes we do. Sometimes we look down on games and say, “That’s just a reflex-based shooter thing.” We do that.

We do the same thing with Peggle. Sometimes we get very snobbish about it: “There’s nothing but interface, feedback, and ‘Ode to Joy.’ There’s no game there.”

But each of these kinds of games actually has great strengths.

The thing you need to do is be aware of what you’re making. Be thoughtful enough in your design to know which tools and techniques you’re going to leverage.

DESIGNING FOR TRAINING

If you are attempting to train somebody, you are trying to teach them not to have to think about a particular problem set. You want them to react based on reflexes and experience.

The key tools for that include cutscenes, triggers, loss of control, and faking everything.

Faking things is incredibly important. Almost everything from flight simulators to firefighter training is based on creating a lot of fakery that does not have a full simulation behind it. If you start simulating everything, you will never get done.

Training is about constraining choices, because you do not want people to start developing their own goals and wandering off.

I watched San Andreas on the flight over. The movie begins with a Los Angeles Fire Department helicopter pilot. An earthquake hits Los Angeles, he is flying over the city, and he says, “Oh, I guess I’m going to fly to San Francisco.” That is not what you want their training to lead to.

So you take away choices.

You also have to use classic educational scaffolding. How many of you know that term because you are on the serious-games side? The rest of you need to Google it after the talk. Scaffolding is the process of building learning on top of prior learning.

One of the classic examples is World 1-1 in Super Mario Bros. It is speed and reflex training delivered through selective agency.

And this might sound shocking: liberal use of tropes is absolutely critical in training.

Tropes, stereotypes, and generalizations are absolutely vital to how the brain functions, as distasteful as that is. We cannot operate on large volumes of raw data. We need to be able to cluster information into clumps.

So all training methods rely on getting people to recognize broad classifications and respond to them quickly.

DESIGNING FOR EMPATHY

If, on the other hand, you are trying to build empathy, you have a completely different set of tools.

You use loose symbolic relationships. Think Limbo. Think Braid. You use ambiguity.

The mechanics you use start to change. You begin doing things like asking the audience for its opinion, or giving people two choices that are equally valid or equally good, or creating moments of surrealism that shock people out of seeing only tropes, symbols, and generalizations.

If you do that, you can build a story that is richly interpretable. You can build a story that people can find themselves in. You can build a story in which they learn things about other people in ways they did not expect.

DESIGNING FOR ENCULTURATION

If, on the other hand, you are trying to bring people into a culture—how many of you are making serious games about teaching tolerance, for example? Anybody?—that is an example of an enculturation game.

You are trying to get a person to conform to a new social norm that you are attempting to build.

There are tools for doing this. There are established approaches. We do things like repeat a phrase over and over, because there are things in our brains that make us respond to repetition.

Go study rhetoric and you will find a giant library of tools and techniques that have been used for centuries to get people to agree with you.

For example, I can stare into your eyes, mimic the movements you make, take exactly the same pose, and lean back when you lean back. You become more likely to agree with me.

There is also the use of frames and labels.

How many of you have seen a war movie lately? Those almost always carry enormous amounts of framing in order to start you out with presupposed points of view. Before long, you find yourself already half agreeing with whatever the movie is telling you to think.

These are the tools of propaganda. But they are also the tools of community building and culture.

“Propaganda” is the term we use when we do not like what we are being taught. But the same tools are also the tools that bring us together. They are the same devices.

That extends all the way down to intentional othering, which we do constantly in games.

Those of you working on games for tolerance, or games in which you are trying to make people care more about others: think about how many times we have made a game trying to do that and still included a bad guy of some sort.

You are using othering to get your point across.

And, of course, you use high causality. If you are trying to enculturate people, you want them to believe that this is how things should be. That means presenting extremely plausible causal chains: this leads to that, because that is the conclusion you want.

DESIGNING FOR INTUITION

If, on the other hand, you are trying to build people’s intuitive libraries, this is where complexity lives.

We usually cannot cope consciously with complexity. Instead, we use heuristics, and those tend to be unconscious.

So you use systems that involve NP problems. You use multiplayer, because players almost always add complexity to a system instantly. You use orthogonal goals and multiple games piled into one another. You build sandbox environments or simulations.

All of those are tools for building intuition, as opposed to empathy. They are tools for building intuition, as opposed to training, because intuition and training can lead you to very different places.

There is a library of these techniques, and each style of game has very different magic powers.

These powers are kind of magical. We are literally rewiring brains, people. Come on. Who else gets to make that claim? Neurosurgeons. Right. Fine. That’s it.

TAKING OUR CRAFT SERIOUSLY

The key thing is that, if we want to take our craft seriously—and further, I would say, take our responsibility seriously—we should try to do this consciously.

Let’s be honest: we’re going to make games anyway. If we’re making games anyway, I personally think we should try to understand what we are doing rather than simply letting shit happen.

Right now, we do an awful lot of letting shit happen because we do not understand our own tools.

So we say to the fish, “Can you teach us magic?”

The thing is, it isn’t magic. We just need to expend the effort to think deeply about our own discipline.

We need to arrive at the underlying science: the underlying rules of harmony, the underlying choreography, the underlying physics of how these systems work.

Lots of people have been trying for quite some time to do exactly that. Mike spoke a little about it yesterday. It is a massively multidisciplinary effort, and there are not enough designers trying to solve these problems today.

There are so few of us that we get together for dinner at GDC every year, and we fit at one table. Seriously.

And I’m going to tell you something—with apologies to all the professors in the room. A music professor did not invent the well-tempered scale. A musician did.

This is on designers to solve, not academics. Academics will teach it after we figure it out.

THE ABSENCE OF A CRAFT

If we do not figure it out, somebody else will come along and enculturate and train us into their system, and we might not like it.

That has already begun to happen.

In the absence of a craft—a formalized method, a system of game design—other people showed up and said, “Guess what? We have all the answers.”

Their answers were plausible enough that they worked for a while. Certainly, we have seen a lot of business successes that lasted at least a few years based on those approaches.

But the fact is that they fall into the trap of looking through only one lens.

If there is one takeaway from this talk, I hope it is that you realize you need to learn to look through a hundred lenses—and, ideally, keep all of those pictures in your head at once—because that is crucial to game design.

Otherwise, you land in the realm of metrics.

THE TRAP OF AVERAGES

The metrics people just do averages.

Average is not where you want to be. Average is where you chase the magic away. It is going to leave. That is not what you want to be doing.

Fundamentally, when you take averages, you wash out the high points. You wash out the low points too. You turn everything into sameness. You reduce it all to the almighty dollar.

You collapse away all of that glorious, complex beauty that lies in the rich systems we have been talking about.

That is kind of a crime, because if you boil everything about the human experience down to an average, you wash away a hell of a lot.

That is not where game design should go. Please, God.

The metrics person will tell you that the little bit of magic did not matter very much. “Who cares? My average is still up.”

And when I take all the fish in the ocean, add them together, and divide, then on average no fish are magical.

So please: go forth and tackle the big problems. We all deserve it.

Thank you.

QUESTIONS AND ANSWERS

Moderator:

Thanks a lot, Raph. We have time for one or two questions, and then I know you’re off to another thing. Just one or two questions, please.

There’s a microphone moving back there.

Audience member:

Will the slides be available?

Raph Koster:

Yes, I’ll put them up on my site for sure. They have a copy here now, too. Also, I see a camera there, so I’m suspicious that this may actually have been recorded for blackmail purposes.

Audience member:

First of all, thank you very much for your talk. I never thought I would find someone who had read my mind and put it on slides.

I have been thinking about the same things for quite a long time. I just don’t have the credentials to go up there and say them.

I also come from an interdisciplinary background, and I recognize every single point you put on your slides. But I also see a big problem.

I’m not a designer. I’m a psychologist, and I have worked many times with designers. Of course, no one person is a whole library. You can develop three of these skills very well and maybe five of them moderately well, but not all of them.

Nevertheless, when I have worked with designers, programmers, and teams—especially on serious games—I find a huge problem: everybody thinks like an island.

Even if I say, “Okay, from a psychological perspective, this doesn’t work,” people end up doing whatever they were going to do anyway because it looks better, or the mechanics are easier to implement, or we don’t have the time, or we don’t have the budget.

They don’t listen to each other. Nobody can be a complete library and have all of these skills, but then people also don’t want to work together as a library.

How would you tackle this problem? What solution would you offer? Would you make a game to teach people how to make games?

Raph Koster:

I’m going to give you two answers.

The first answer is that you are describing a psychology problem—more specifically, a management-science problem. So go find a manager and have them solve the problem for you. There are, in fact, ways to solve that problem. It is tractable. It has been dealt with.

But the real answer is: yes, those library people do exist. Many of them are in this room, actually. I’ll go ahead and say that I am one.

Most of the best-known game designers whose names you know are people like that.

The practical solution, I hate to say, has been for autodidacts to train themselves obsessively, reading in every field they can because they get curious about questions like, “I wonder how quantum mechanics connects to astrology?” They are simply people like that.

They go on game-design retreats where they have those discussions for hours on end for no apparent reason. I’ll be at one next weekend, literally.

There are dozens of game-designer secret societies that hold retreats to do this.

There is a group of people who train themselves to be massively interdisciplinary Renaissance people. That is what they do.

And in this industry, what they end up becoming is creative directors.

So, to all of you who aspire to be creative directors: that is your career path. I’m serious.

My advice to people who want to reach high levels in the game-design profession, and end up running projects, is that they actually have to care about every part of that graph.

Make no mistake: in a large enough studio, a creative director is in fact in charge of telling the haptics designer what to do, the story designer what to do, the complexity designer what to do, and the social designer what to do.

They need to know enough to tell everybody on the team what to do.

On the great games, the creative director takes in ideas, but makes the calls and rules with an iron fist. That is actually the solution that has tended to work.

So my career advice to all of you is: go get a really broad liberal-arts education in every field known to humanity if that is the job you want.

All of the top designers I know are people like that. I think there are people in the room who can vouch for this, because they know those people too.

Miyamoto comes up with this stuff because he goes gardening. People at the top of this field draw from a huge array of inspirations.

Moderator:

That was a long question and a long answer. Is there room for one more? Okay.

Audience member:

I loved the talk, by the way. That was great. I like the library concept.

For the people here, if you were to recommend five books—only five—and only one of them could be about games, what would you recommend?

Raph Koster:

Five? Five. And only one about games?

All right.

For mental models, heuristics, and the like, I would probably read Thinking, Fast and Slow, or possibly Sources of Power. Either of those would work.

For the learning process, there is a pile of material in education. There is also a lot of material in psychology around agency and self-determination theory. That is not one book, but there are multiple things to read in that area.

For everything related to haptics, affordances, and feel, Game Feel is pretty good. But so are many different kinds of sports materials. A lot of studies have been done along those lines.

When you start getting into things like ludic structure, you really need to be reading mathematics. I suggest reading up on graph theory as a fairly accessible entry point.

Books about six degrees of separation are a good way into that. Duncan Watts is one of the key researchers, but I really like the work of Albert-László Barabási because it is very accessible while still presenting interesting, complex problems.

Complex mathematical problems are isomorphic across different fields. That means that once you learn them through graph theory, you can begin applying them in other fields of mathematics.

Really, it is about finding the version of mathematics you feel comfortable with, so you can dig into NP-hard problems and all of that.

For social subjects, the best training is history, anthropology, and that kind of thing. Jared Diamond is a fantastic entry point for looking at some of these questions, and is also very accessible to a general audience.

For complexity problems—well, I guess I already talked about that.

There actually isn’t that much to study about chance, but the best sources are in psychology. Just watch out for the strange problem in psychological research whereby much of it is performed on white male undergraduate psychology students.

Once you get into the realm of information display, Edward Tufte is absolutely the best place to go. The books are unfortunately expensive.

When you get into the realm of story, myth, and authored narrative, you pretty much have to study creative writing. Fortunately, there are a fair number of books.

Everybody starts with the Joseph Campbell-type material. I called it “crap,” but I don’t mean to say it isn’t important. It is important. There is simply much, much more out there to study, including many different storytelling traditions.

I have an MFA in poetry, and it has been crazily useful in my career, believe it or not.

That eventually leads you into the material about fun and learning in psychology. I hate to tell you this, but the only major book on fun is still mine. I don’t know why that is. It has been ten years. Somebody else should have written one by now.

That will take you further down this line.

There is actually shockingly little research on meditation—almost no scientifically rigorous research at all.

For flow, Csikszentmihalyi is the key place to look. For deliberate practice, K. Anders Ericsson.

Yeah. That is a decent starter library. Sure.