Rethinking Challenges in Games and Stories

Ernest W. Adams

2007 Game Developers' Conference

This is an approximate transcript of the text of my lecture at the 2007 GDC on March 9, 2007. I present it in this form because the nature of the material does not lend itself to the traditional paper format. Also, because the lecture is informal and to some extent ad-libbed, this is not a verbatim document.


I'm bad at a lot of video games. I realize that by admitting this, I've probably destroyed my credibility with some of you, and those people should go ahead and leave. But the truth is, a lot of games are simply too hard for me and I'm not ashamed to admit it. The problem is produced by a combination of factors. First, I don't have all the hand-eye coordination in the world. I'm clumsy to begin with, and I'm getting older, and now need reading glasses, and so on. Second, I don't have that much time to play any more. I can't afford to spend hours trying to beat one particular boss. And third, the nature of the challenges in the games that I'm bad at are such that I don't even enjoy trying. There are certain things I'm prepared to take time over, and others that I'm not.

In the course of the last year I took the book on game design that I wrote with Andrew Rollings—Andrew Rollings and Ernest Adams on Game Design—and I expanded it significantly. In the process of doing that, I realized that I needed to pay more attention to the issue of difficulty. This lecture is the product of my thinking. It's divided into three parts.

The first part is a theory of difficulty that I want to present. It's very pragmatic, and I think that it has direct applications in level design and game balancing.

The second part is some blue-sky thinking I've been doing about challenge-free play; that is, play without gameplay.

The third part is about the effect that challenges have on interactive storytelling, and particularly their emotional effects.

A Theory of Difficulty

Over the last year I revised the book that I wrote with Andrew Rollings, Andrew Rollings and Ernest Adams on Game Design, and in the process I realized that I needed to think some more about the nature of difficulty in video games. The question that faced me was, how can we get from the difficulty that we, as designers, think that we build into a challenge to the difficulty that the player perceives in overcoming that challenge? What factors go into that player's perception, and how are they related to one another?

Csikszentmihalyi's Theory of Flow

ewa_ch11_fig02First I want to introduce a concept called flow, devised, or discovered, by a psychologist named Mihalyi Csikszentmihalyi. [The name is pronounced, approximately, "me-HI chick-SENT-me-hi."] Flow is a pleasurable state of high productivity which can occur either during work or play. When a person's ability to perform a task balances the difficulty of the task that they have to do, then they enter the flow state. When the task is too easy, they become bored. When the task is too hard, they become anxious. When it's just right, they experience flow. I'm we've all felt this when we're playing games. We call it "being in the groove," or "being in the zone." You're really cruising, you're doing well. It's a marvelous feeling.

I believe that this is what we want to achieve for our players, or rather, for our players to achieve for themselves: to enter the flow state. Now, I realize that there are some old-time hardcore game designers who think that the point of game design is to make games as hard as possible, because they equate difficulty with fun. But I think the industry has moved on from that -- at least, I hope so -- and I believe that our goal is to get players into a state where they are enjoying the game enormously simply because they are doing well at it. Not because it's too easy (because if it's too easy, they become bored), but simply because it matches their abilities. In effect, what we want to do is get them to the right level of perceived difficulty. How do we determine the perceived difficulty of a challenge?

Six Factors

In thinking about it, I concluded that there were six factors to take into consideration: four that we can control and measure, and two that we have no control over. Here they are, and I'll discuss them in more detail in a moment:

  • Intrinsic Skill Required to meet a challenge, irrespective of time pressure.
  • Stress, that is, the level of time pressure placed upon the challenge. I separated time pressure from intrinsic skill required because you can usually add or subject time pressure from just about any challenge.
  • Power Provided by the game to help the player meet the challenge. This takes the form of weapons, powerups, available moves, and so on.
  • In-Game Experience, that is, the length of time the player has been playing our game when he encounters the challenge in question. We could actually keep an internal timer that measures how long a player has been playing. We don't usually do that, but we do know that if a player is at level 10 of a game, we have a general feeling for how long they have been playing.

The other two that we can't control or measure are:

  • Native Talent the player brings to the game.
  • Prior Experience at playing similar sorts of games.

Absolute Difficulty

Looking further into this, I asked: How do we measure the absolute difficulty of a given challenge? I concluded that the absolute difficulty consists of two factors: intrinsic skill required and stress. Intrinsic skill required is measured relative to the trivial case, or a baseline case, for the given challenge, again, independently of time pressure. The metrics for every task are going to be different. I'm not trying to claim that you can turn this into a universal formula that enables you to compare the difficulty of heterogeneous tasks to one another. Different tasks will have different kinds of metrics, so, for example, measuring the difficulty of hitting a target with an arrow will be different from measuring the difficulty of solving a logic puzzle such as Sudoku. In hitting a target with an arrow, the factors that contribute to the intrinsic skill required are such things as the distance to the target and the size of the target, whereas with Sudoku, it's how many of the boxes are already filled in.

Then there is the stress, the effect of time pressure on the player. Stress is a factor that discourages logical thinking and planning. The player's play becomes much more tactical, much more reactive rather than proactive when he is under large amounts of stress.

When you put these intrinsic skill required and stress together, if you want to maintain a fixed level of absolute difficulty for a given type of challenge, then the more you have of one, the less you should have of the other and vice versa. If you raise them both, the absolute difficulty goes up.

ewa_ch09_fig03Here's a graph showing the intrinsic skill required versus stress for a variety of types of games. At the extreme left are games of the lowest stress, and all the way up the left hand side we have turn-based games, from tic-tac-toe to golf. Golf requires a very high degree of intrinsic skill, but it's not a high-stress game in the sense that you have to take your shot quickly or someone is going to come and shoot you.

Moving farther out we have action-adventures, which place more stress on the player, real-time strategy games, which require somewhat more intrinsic skill but aren't quite as fast-paced as action-adventures, and first-person shooters, which are very high on the stress side but don't actually require a large amount of intrinsic skill. That is, if you had all the time in the world to aim your shots in an FPS, they wouldn't be very hard. But the stress level is so high in games like Quake Arena that that is the source of most of its difficulty. Finally, I've added a real-world example, cardiac surgery, which isn't as frenzied as a first person shooter (although it does have severe overall time limitations), but which does require an extremely high level of intrinsic skill as well.

Managing Player Perceptions

So absolute difficulty is measured with respect to a trivial case or baseline case, e.g. hitting a target 50 feet away is more difficult than hitting a target one foot away, or in more familiar game terms, defeating a level 5 troll is harder than defeating a level 1 troll.

Then you can include another factor, power provided, and get the relative difficulty of the challenge. This is done by taking the absolute difficulty and subtracting the amount of power provided by the game to the player to help with the task. If you make the player a level 5 knight, then obviously they are considerably more powerful than a level 1 knight is. So a level 5 knight can defeat a level 5 troll almost as easily as a level 1 knight can defeat a level 1 troll. Naturally, role-playing games do this all the time. As the levels up, he gets harder and harder monsters to fight, and you always try to keep the strength of the monsters just a little above the power of the player.

Finally, include one more factor, in-game experience, and you get the perceived difficulty of the task. In-game experience goes up as the player continues to play.


So here's how these factors interact. Absolute difficulty goes up very rapidly. Typically, in something like a role-playing game, if the player is level 1 and gets into a level 3 area, the game is already impossible. In practical terms she simply cannot succeed, so the difficulty goes to infinity. But the game provides the player with growing power to meet its challenges as their difficulty goes up, so the relative difficulty line is a less steep curve. Then, as he gets more experience as well, we create the perceived difficulty, which is what we're really concerned with.

Notice that at the beginning of the game, relative and perceived difficulty are the same because the player has zero experience.

So although absolute difficulty goes up at an insane rate, the perceived difficulty goes up at a much more gentle rate, and that's what we want. We don't want to have terrible spikes in the player's perception of how hard the game is; they complain about that. Players will stop playing if the perceived difficulty of a game goes up much too fast. You need to manage the rate at which the absolute difficulty goes up along with the amount of power that you provide to help meet them. And you need to space them out appropriately so the growing difficulty is also compensated for by the player's growing experience.


If you provide too much, power, look what happens. If you say, "I'm going to make sure that the power provided always exactly matches the absolute difficulty of the challenge in question," that is, it is exactly as easy for a level 5 knight to defeat a level 5 troll as it is for a level 1 knight to defeat a level 1 troll, so relative difficulty remains flat, then the perceived difficulty actually goes down. The player's experience is growing all the time as well, so if it's no harder to beat a level 5 troll as it was a level 1 troll, then the player will begin to be bored. They will move out of the flow state.

So, putting all this together, you get the following equation:

perceived difficulty = (intrinsic skill required + stress) – (power provided + in-game experience) or minus the other two imponderables, native talent and prior experience, which you just don't know about.

You can compensate for player talent and prior experience to some extent by offering the player multiple difficulty levels to play at, and I strongly believe you should do this. I consider that a commandment of game design: You should always include multiple difficulty levels in a game if you can, although I recognize that not all games are suited for it. But you should if you can, because it makes your game more accessible to a wider range of people with different levels of talent and experience. And I would add another commandment that someone suggested to me recently: easy mode means EASY, dammit. Easy mode should be so easy that you can win the game by bashing the controller with your forehead. It's no problem to make a game hard, but we can and should work to make games easier. And once again, old-time designers who were mostly interested in making hard games used to say, "But the player will win too soon!" To which I respond, "And that is a problem for you why? Is your ego going to be bruised if they get through your game too soon? If it's too easy on easy mode, then they can put it on a harder mode."

Putting it all together, I think this is a useful way of thinking about difficulty in games, and as we design, we should try to keep all these factors in mind. When a challenge appears to be too hard, what is the reason? Because the game doesn't provide enough power to meet it, or because the player hasn't had enough of a chance to learn how to defeat it? I feel these are valuable concepts in the process of game balancing and level design.

Beyond Challenges

For a long time, I've been saying that, just as Impressionism was a new way of seeing that raised the question of what painting was about and what it could do, so we need new ways of playing that explore the question of what interactive entertainment is about and what it can do. I'm not just talking about sandbox modes in existing games. Sandbox modes are fine, and I think we should have them, but in a lot of sandbox modes the designer just punts. He says, "OK, if you won't play by the rules of the regular game, then all bets are off. Have fun if you can, but don't count on me to help you. I've turned off all the challenges, and you're free to screw around as much as you like, but you're on your own."

But what about if we devoted the same amount of energy to creating challenge-free play spaces as we devoted to creating challenging ones? Spaces that are just as much fun to visit, but whose fun arises from another source than challenge and achievement? This already happens to some degree in MMOGs. Players find things to do that don't have much to do with the game's primary challenges.

Ways that Games Entertain

Let's take a look at some of the ways that games entertain:

  • Gameplay. This is always at the top of the list. I define gameplay as the challenges that a game offers, plus the actions that it provides in order for the player to  meet those challenges. If an interactive entertainment product calls itself a game, then it should entertain through gameplay. But there are many other sources of entertainment that, as designers, we should keep in mind:
  • Aesthetics. The visual and auditory style of the experience.
  • Storytelling, if you choose to include it. Characters to care about and a plot to become involved in.
  • Risk and reward. Gambling games rely almost entirely on risk and reward. If you stop and think about it, most gambling games are really stupid: roulette, blackjack, the lottery. Almost all their entertainment value comes from risk and reward.
  • Novelty, new things to see and do.
  • Learning, gaining understanding and mastery. Raph Koster thinks that learning is a huge part of the source of entertainment in play.
  • Creativity. The ability to build stuff and say, "Look, I made that."
  • Role-playing, being somebody else, putting on a mask and acting a part.
  • Immersion. Going someplace else, entering an alternate reality.
  • Socializing. Interacting with other people.

So gameplay is at the top of this list, but it's far from the only way in which games entertain. There's a lot of other stuff that we could concentrate on more if we chose to.

Games as Systems

There is a common tendency on the part of some designers and theoreticians to think of games primarily as systems – this is the approach taken, for example, in Salen and Zimmerman's Rules of Play. Salen and Zimmerman's work analyzes games in a fairly formal sense, hence the emphasis on rules in the title. Rules are systems. The core mechanics are composed of systems; they are the algorithmic implementation of the rules. Raph Koster also tends to think and write this way. Many game designers who used to be, or are still, programmers, think this way.

The question is, what are they systems for doing? Saying that a game is a system doesn't really tell you much; the more important issue is what the system does to create the player's experience. The systems in most games exist for the purposes of offering gameplay. But suppose that we create new systems that offer all the other sources of entertainment that I mentioned? Again, MMOGs do this to some degree; they include systems for encouraging socializing and systems for encouraging role-playing (not that the players bother, so far as I can tell).

Some designers consider the graphics and the sound, the environment, to be mere window-dressing hung upon the underlying system, and the system is all that really matters. A few even go so far as to regard graphics as nothing more than a necessary evil imposed by the marketing department. This is of course the old graphics versus gameplay argument, and it's now largely a dead issue; any game designer with any brains knows that you have to have a suitable proportion of both, and to execute both well.

But what if we had graphics without gameplay? I don't mean in the sense of the dreadful "interactive movies" of the early 1990s, that offered lots of narration, but gave the player very little to do. Rather, I mean graphics with other kinds of play besides gameplay. In other words, the systems that we create need not be exclusively dedicated to providing challenges, and in some cases I don't believe that we need to provide challenges at all. It's quite possible to make an entertaining experience with all the features on my list except the first one.

Far Cry

Let me talk to you about what happened the first time I played Far Cry. I started it up, and I immediately noticed that the landscape was gorgeous. The trees, the sky... there were even fish in the water. It's full of interesting old ruins. I just wanted to hike around and look at it. And I thought to myself, "Finally I can explore a tropical island without the heat or humidity or poisonous bugs."



FarCryEnemyUnfortunately, when I tried it, every 30 seconds some jerk tried to shoot me! [laugher and applause] How much fun is that? Would you go to the Pyramids, or to Angkor Wat, or Chichen Itza if somebody were trying to blow your head off all the time? So you're supposed to play Far Cry if and only if what you want to do is shoot other people before they shoot you.

And when you think about it, isn't it a shame that our artists spend so much time creating these incredible environments that nobody gets a chance to enjoy properly?

When I introduce the concept of an internal economy in my Fundamental Principles of Game Design workshop, I do it with reference to first-person shooters because they have very simple economies – enemies, hit points, ammo, that's really about it – but I'm always careful to point out that FPS games also have exploration challenges that are secondary to the central challenge of managing your hit points and ammo.

But in fact, as we've begun making rail-shooters or arena games, FPSes have abandoned exploration as a source of entertainment in its own right. If you're on a rail or in an arena, there's not a lot to explore. Exploration is about making choices in an unfamiliar environment. You can't do that when you're on a rail. Now Far Cry isn't exactly a rail-shooter, but it does require you to traverse the landscape in particular ways, in order to force you to confront its challenges in a certain order. In other words, its landscape, beautiful as it is, is optimized for gameplay. That's as it should be, because Far Cry is a game.

Counterstrike sniping areasCounterstrike sniping areas 2Here's another example. This is an analysis one of the Counter-Strike levels. It's a carefully designed space for balanced sniping opportunities between the red and blue teams. Both have an equal chance in this scenario -- they have good positions in which to hide, and certain areas that they can cover, and so on.

Level design for shooters is a particular skill that concentrates on such things as sight lines and choke points. But landscape design for the purposes of exploration itself requires different considerations. And this is what the great landscape designers of the 18th century excelled at. They persuaded the owners of English country houses to tear out their flat, formal, geometrical gardens that had been popular for 300 years, and to replace them with landscapes that were meant to be explored – landscapes full of hills and valleys, hidden lakes and grottoes, and distant vistas. They even built fake ruins, called follies, on the tops of hills, just to make the skyline more interesting. Or they would deliberately allow a small portion of a building to show through the trees, to encourage people to come and find out what was there. In game design terms, they left a clue. You can see that in this image:


So the landscape designers of the 18th century were in effect designing exploration-game landscapes, or rather an exploration -experience landscapes. It wasn't about gameplay, but about exploration.

The landscape gardeners optimized their landscapes for people who were simply taking a walk. Interactive entertainment needs to provide people with more to do than simply take a walk. But I still think there's something to be learned from the principles that they discovered. And I think that both landscape gardening, for outdoor spaces, and architecture, for indoor ones, have a lot to offer us if we want to provide the kinds of reduced-difficulty or reduced-challenge play that I'm talking about. Games for people, who like me, aren't very good at a lot of games.

Earning the Right to Play

The central organizing principle of most video games, and this goes all the way back to our arcade days and has been with us ever since, is that the player must earn the right to play by doing well. If you're bad at the game, too bad. You just don't get to see the rest of it. Now, we also put in cheats to help the bad players, but we label them explicitly as cheats in order to humiliate the player and to remind him that he's no good, as if he needed reminding.

But if you want to make games equally accessible to poor players, and you should, because there are a lot of us and we have money too, don't require the player to earn the right to play. Earning the right to play is a challenge-and-achievement model. It's inappropriate if you want to provide the player with non-gameplay play.

Virtual Tourism

So what about a landscape that's optimized for exploration, and other, non-gameplay-based activities? What about a computerized version of Club Med, where you can do the kinds of things that people like to do when they're on vacation? Here's a brief list:

Visiting interesting places
Horseback riding



    Hang gliders



Recreational sex!

In essence, what I'm talking about here is virtual tourism. And I don't mean to suggest that all these things are easy to do well. You can still include activities that have a learning curve. It takes practice to fly well or to sail well. Rock-climbing and mountaineering are strenuous and difficult activities, and you can, to some degree, mimic that strenuousness and difficulty.  But I'm saying that you don't have to force the player to jump through your hoops in order to earn the right to play. A skier can ski on any level slope she wants to. If she's not very good, she might not do well at the advanced slopes, but since a virtual skier can't hurt herself, why prevent her from trying? Unlockable content is all very well when it's used as a reward for achievement, but that's not appropriate here.

Second Life

Naturally, one of the first things that comes to mind when we talk about virtual tourism is Second Life. Second Life's conceptual ancestor was an online environment for – I kid you not – the Commodore 64, called Club Caribe, which was named in deliberate imitation of Club Med.

Gamers kind of upset by all the mainstream publicity that Second Life gets, because it has a fraction of the number of participants that World of Warcraft has. And Second Life keeps on attracting attention and getting mainstream press, as when presidential candidates open offices in Second Life. Why do you think this is? The reason is that Second Life is about money and sex, while World of Warcraft is about killing imaginary monsters. In other words, Second Life offers an experience that the mainstream press is interested in, partly because it bears a relationship to real life, which it's the mainstream press's job to cover. When the Swedish government decides to open an embassy in Second Life, that's mainstream news. It's about a real-world entity stepping into cyberspace for a real-world purpose. World of Warcraft is about a world that it is not the mainstream press's job to cover. From a numbers standpoint it would make much more sense to have an "Obama for president" office in World of Warcraft than in Second Life, but that would be a fantasy-killing element and everyone would hate it. World of Warcraft is a game; Second Life is a place.

So Second Life approximates what I'm talking about, but it places much more emphasis on social and commercial activity than it does on virtual tourism. I was exaggerating when I said that Second Life was about money and sex; it's about money and creativity. Sex is just a by-product that you tend to find in these sorts of environments. But its emphasis on player creativity is both a strength and a weakness. It's a strength in that it's very Web 2.0, and it counts on letting the contributors do the work of providing the content, which is certainly cheaper for Linden Lab than doing it themselves.


The weakness, however, is that the content of Second Life is extremely surreal. It's the product of a very large number of competing visions. As a result, Second Life is both aesthetically and culturally incoherent. It feels like walking around Disneyland on acid. [laughter] One minute you're in Tomorrowland, but the next minute you're in Frontierland and a woman with a blue cat head is trying to talk to you. It's a sensory overload of the strange. Because it's not the product of a single guiding mind, it doesn't convey a harmonious sense of place. Rather, it's an endless series of discords, so I don't think it's the answer if we really want to provide virtual tourism. Personally, I place a high value on creative vision; that's why I like looking at the English country-house landscapes. I can say to myself, "Ah, this is a Capability Brown landscape, and I can feel his deft touch at work here in the placement of these lakes and little rivers and so on."  But I do think that Second Life is a step in the the right direction: entertainment that offers things to do, without forcing you to overcome challenges to earn the right to do them.

S.T.A.L.K.E.R.: Shadow of Chernobyl

I want to talk for a minute about S.T.A.L.K.E.R.: Shadow of Chernobyl, which does something pretty amazing with space as well. S.T.A.L.K.E.R. has been vaporware for several years, but it just went gold about a week ago. The title is somewhat unfortunate, for those of you who have not heard of it, and it represents what I think is a mistranslation from the Russian. It has nothing whatsoever to do with stalking in the familiar connotation of obsessively pursuing an innocent person. A better translation would be "hunter," but at this point the marketing has been done, so THQ is stuck with the name.

I was asked to do some design consulting on S.T.A.L.K.E.R., and I flew over to Kiev to meet with the development team, so I got a good look at it. S.T.A.L.K.E.R. is a first-person shooter set in the 30-kilometer exclusion zone around Chernobyl. It's pretty much the opposite of a rail shooter – it's mostly outdoors in the open,  and you can go anywhere you want, although there are certain choke points to prevent you from getting into areas before you're ready for them.

But what really struck me about Stalker is the extent to which they've modeled the real place. I was in their offices and I happened to look out the window and noticed an interesting and distinctive pine tree of a type that I had never seen before. Then I looked at one of their screens and there were those pine trees. And they have modeled the dead city of Pripyat, which was right next to the power plant, with extraordinary accuracy. Here are two pictures of it, the top one a photograph of the real place and the bottom one, the modeled one in the game.

Pripyat - Real
Stalker - Pripyat 4

Pripyat was once home to 45,000 citizens, who were all evacuated in the space of a few days by a group of heroic bus drivers who went back in again and again, ferrying people out. All those bus drivers are dead now, and the whole city stands empty and decaying. But GSC Game World, the developers, have reproduced every building. The species of trees are right, the abandoned vehicles are right. What they've done is stalkerminito create a memorial in computer game form.

Now don't get me wrong. This is still a shooter, and it's full of zombies and hostile soldiers, and all kinds of other nasty stuff that's constantly trying to kill you. And I freely admit that S.T.A.L.K.E.R. has the same problem that Far Cry has, in that you can't just explore freely, because there's always a risk of being eaten by a mutant pig. [laughter] But this isn't yet another fantasy world. It's the game developers' own country, that got poisoned and can never be inhabited again. Imagine if, instead of being flooded with water, New Orleans had been flooded with radioactive waste. That's what happened to northern Ukraine. And they deliberately chose to remind their players of that, which I think is an interesting and moving thing to do in a video game.

America's Army

ArmyOps1Now  let's look for a moment at America's Army. This is one of the most peculiar games you can imagine the US Army producing, because the Army is the last place you would expect to find moral relativism. Moral relativism is the idea that right and wrong are simply a matter of perspective. But because of the strange politics of making this game, they had to build moral relativism in. It was politically impossible for the Army to make a game in which players could shoot at US soldiers, and obviously they didn't want to have US soldiers shooting at each other either (which happens by accident with distressing frequency).

The Army could not do that, so they had to make a game in which everybody thinks that they're a US soldier, and everybody thinks that everybody else is a terrorist. The graphics display the world that way. So in this image [below] here we are, we're US soldiers. We've got the drop on this terrorist. He's surrendering. We're standing there with our M-16s, and he's holding up his AK-47 in surrender. That's how we see the world. If we were to flip it around and see it from his point of view, he would see himself as a US soldier and us as terrorists. The graphics engine would render us as terrorists. He would see himself holding an M-16, and us holding AK-47s. It gets even stranger still. If we pick up his AK-47, he sees us pick up his M-16. Completely bizarre. Not only is the representation of the people relative, but even the weapons. Total moral relativism! Everybody thinks they're the good guy.


I think this idea that you can display multiple perspectives on the same place is extremely interesting. In America's Army, there is no underlying reality. There's no data bit that says, "This is the correct answer, you really are a terrorist." It's completely morally relative. And I think that this could be put to use in other kinds of games as well.


peacemaker03Here's a screen shot from another game that recently came out, PeaceMaker. PeaceMaker is about trying to create a successful two-state solution in Palestine. You take the role of either the Israeli Prime Minister or the President of the Palestinian Authority, and you try to manage the situation and negotiate with the other side in order to arrive at a two-state solution. At the same time, you have to deal with various militants on your own side, whom you have to try to keep in check. And it's exactly the right sort of game for showing the same situation from different points of view. In PeaceMaker it's not done with a 3D engine, it's done mostly through reports of events, and it chooses to report events that are of interest from different points of view. So, for example, "18 Palestinians killed and 40 wounded by Israeli tank fire" is an event that directly affects the Palestinian Authority President's ability to negotiate with the Israelis, because of course his own people will be very upset by this and demand a strong response. This is the kind of game in which showing the same landscape and circumstances, but from different points of view, would be particularly advantageous. In this case it's political points of view, and in the case of America's Army it's graphical points of view, but I think there's a lot to be taken from this. If we stop and think about these kinds of things, we can find opportunities to present places for people to experiment with play in situations that seem different to each other, but are in reality the same. They may even learn to bridge their differences. The really interesting thing about PeaceMaker, which incidentally is being released simultaneously in English, Hebrew, and Arabic, is what happens when people from one side play the other side -- the one they're not familiar with. They get a chance to see what the world looks like from a different point of view. I had a similar experience when I played the Russian side in Chris Crawford's Balance of Power, many years ago. I had never really thought about what it would be like to be the Russians during the Cold War, and the game really brought it home to me. All of America's friends were extremely rich and powerful and armed with nuclear weapons, like Britain and France, and all of Russia's friends were extremely poor, like Cuba. And the Soviet Union was surrounded by rings of steel in the form of NATO and other treaties. So I had this really unfamiliar and enlightening experience of what it was like to be the Russians during the Cold War, and I think this capacity to present people with different points of view, whether graphical or otherwise, offers them an experience that has very little to do with challenges. Now it's true that as I played Balance of Power I was trying to accomplish certain tasks in this context, and that helped to make the point. But in any case, it was an experience that I don't get when I'm just trying to blast aliens.

Science Museums and Science Software

I think there are other ways to use this kind of power as well. Why are science museums cool -- the Exploratorium here in San Francisco is cool, the Museum of Science and Industry in Chicago is cool, the Ontario Science Centre is cool, the National Air and Space museum is the most-visited museum in the United States -- but science educational software stinks? Why is this? The software is hands-on, you get to do things, but even so, it's terrible. Well, I think it's because we're not making an effort to create a sense of place. The Exploratorium is a cool place to be in, it's a vast building full of blinking, flashing things where all kinds of exciting things are happening, and you can feel it around you. But most science education software just presents little flat worlds in which you pour one test tube into another test tube and the mixture turns blue and that's about it. We could use some of our power to present interesting phenomena, in a non-gameplay context, to make that software more exciting and fun to play with.

We can now show every leaf on every tree. Every blade of grass, every petal on every flower. We can show the quality of the sunlight at dawn as it breaks through the storm clouds of the night before. We can even display alternative versions of the same reality to different players. So it's time that we used this power to entertain in its own right. To provide environments that are more than simply places in which to shoot things. Our nouns are amazing. Let's get some new verbs. [applause]

The Effects of Challenges on Interactive Stories

What does the author owe the reader in a traditional story?

  • Credibility -- the story has to be believable. Even in fantasy and science fiction stories there's a credibility budget; I talked about that last year. Ken Perlin puts it this way: "You can be the last human being left alive on a remote asteroid and the rest of the human race has died out, or you can invent a time machine, but not both."
  • Coherence -- the story has to hang together as a harmonious unified whole without any contradictions, unless they're intentional and presented to create ambiguity or to represent different points of view. But in general, coherence is one of the things we owe a reader.
  • Dramatic meaningfulness -- the story has to be about things we actually care about.
  • Technical and aesthetic proficiency -- it has to be competently crafted.

Those are the things that we owe a reader in a non-interactive story. That's about it. But in interactive storytelling, the player places additional expectations upon us. First and most obviously, they expect to be given something interesting to do. And generally, they expect to be a hero rather than a sidekick, to be the prime mover in a story rather than in a secondary role. Most of the time, what we give them are challenges. But there are issues with this.

If we offer the player storylines that can change, they typically change for one of three reasons:

  • Randomly, or in response to internal computed mechanisms that the player has no control over.
  • In response to the player's choices.
  • In response to the player's ability to meet challenges.

If the story branches based on internal mechanics, then it can be frustrating to the player, but as long as the story is credible and coherent , it doesn't matter that much. If the story branches on player choices, then it reflects the player's own wishes, and it gives him an opportunity to participate more fully in the game world. If it branches based on the player's ability to meet challenges, then certain problems arise.

Changing the Plot via Challenges

If we change the plot via challenges, then the player's degree of success or failure produces different outcomes. The emotional significance of this is that the player expects to be rewarded with positive dramatic consequences for meeting those challenges. The player will get annoyed, and be unhappy with you and with your story, if he is told that his efforts have been in vain ("Oh, did we ask you to bring back the magic whatsit, and you sweat blood and died fifteen times to get it? Sorry, it turns out we don't need it after all.") or in the wrong direction, i.e. you lied to him ("Thanks for sweating blood and dying fifteen times to get the magic whatsit... what you didn't know is that it is the final piece I needed to assemble the Doomsday Machine and destroy the world!").

So if the game is about achievement, then the plot must reward achievement.

Changing the Plot via Choices

By contrast, if the plot changes on the basis of player choices, then the player is free to choose among options or behaviors that affect the plot. The emotional significance of this is that dramatically significant actions, that is, those that do affect the plot, must be apparent, not obscure or trivial. ("Oh, you walked out of your house wearing the Manolos instead of the Jimmy Choos, so I've decided to destroy the world today." "Wait a minute. It depended on what kind of shoes I put on? How was I supposed to know?" "You didn't! [evil laughter].") That's not acceptable.

Secondly, the player expects the progress of the plot to meaningfully reflect her choices. If you tell the player that her choices matter, then they damn well have to matter. Telling the player that it's vitally important that you make a choice, and then she discovers later that it didn't matter at all, is not acceptable either.

But changing the plot based on choices is a different thing from basing it on challenges, because with challenges the player actually has to accomplish something, and if she fails, then the plot goes a different way, and she has to go back and work at it some more in order to get it to go the way she wants it to.

So changing the plot via choices is very useful for setting up situations such as moral dilemmas, or social or political decisions to make. But the greatest advantage of this is that you don't have to earn the right to play! You make the choice and on you go. You get to see what the consequences of that choice was, and then if you want to go back and play the game again with different choices, you can do so. Whereas, if you base the plot branching on challenges, what happens if the player is really good? He zooms through the game and does really well, but if he wants to go back and see what other storylines there might have been, he has to play deliberately badly in order to see other branches. And that seems kind of weird. Counterintuitive to what people actually expect.

So that's something to keep in mind about all this stuff. Now I'm not saying that challenges are wrong in stories, or that you shouldn't use them. I'm saying that there are emotional consequences, and these consequences are not always beneficial. Understand what they are as you decide how to design your game.

The Great Debate

This leads me to my final mini-rant. Last year I pointed out some of the difficulties with branching storylines, and I presented a new idea as an alternative approach. This may have led the less thoughtful among you to conclude that I was condemning branching storylines. That is not the case. There's this huge debate going on in the academic literature and on the Internet forums about what interactive storytelling is supposed to be, and there are various factions.

There's the anti-storytelling crowd, the people who believe that all storytelling in games is a waste of time; they're the people who button through the movies right away. When id Software made Doom, the called story "the s-word," and consequently when somebody made a movie out of Doom, my God, it was an s-word. [laughter]

Then there's the pro-storytelling crowd, and this is a whole bunch of different people. It's the fans of adventure games, who are still out there and getting what they want, exploration and puzzles. There are practical researchers and people who are trying to sell engines; that's the category I put Chris Crawford into. And the interactive fiction developers who are working on text-based games and so on. There are the narratologists and people in the academy. There are the wanna-be film directors from within the game industry, who are positive that they know all about what interactive storytelling is supposed to be, and so on.

And there are various arguments that you get on the boards and other places:

  • "Linear stories are bad! It's only true interactive storytelling if the player can change the outcome."
  • "Branching stories are bad! The combinatorial explosion of branches makes them cost too much."
  • "Foldback stories are bad! The freedom that they offer the player is false. They're fraudulent!" That's what Chris Crawford said in his book. I'm going to go sue LucasArts.
  • "Emergent stories are bad! Only humans can tell decent stories."

They're all wrong. And where they're all going wrong is that there's way too much emphasis on structure. The big error that these theorists make is to concentrate on structure and delivery and organizational mechanisms. That is like taking a class in creative writing and spending the whole time studying grammar. What matters is the player's experience, not the mechanism that delivers it. All this stuff about "this is the right way to do it," and "this is the wrong way to do it,"  is a waste of time. The only thing that matters is how the player perceives it in the end. You don't create art by prescription about technique.

So think for a moment about the huge number of types of non-interactive stories in the world. Here's a list:

  • Jokes
  • Urban legends
  • Fan fiction and Slash
  • Television advertising
  • Children's cartoons & TV shows
  • Sitcoms & satire
  • Plays & short dramas
  • Drama series
    • Unlimited, e.g.
      Law and Order
    • Limited, e.g. 24
  • Soap operas
  • Movies


  • Children's books
  • Comics
    • Single panel, multi-panel, and books
  • Airport fiction
    • Chick-lit
    • Thrillers
  • Genre fiction
  • Short fiction
    • Women's magazines
    • Literary magazines (New Yorker, Atlantic Monthly)
  • Highbrow literature

No one theory of storytelling can cover all of these. Aristotle does not tell you how to write urban legends. Joseph Campbell does not tell you how to write for the New Yorker. So why would anybody think that one theory of interactive storytelling can possible cover all the forms of interactive stories?

The Meta-Approach

So here's a meta-approach that cuts through all this crap. Forget all the debates. Forget all the people who say, "This is the one right way to do interactive storytelling." Don't let yourself be bullied. Don't let anybody tell you that linear stories are no good because the player can't change the outcome. That's OK with many players. Don't let anybody tell you that branching stories are no good because they cost too much. You can make it work if you keep the number of branches down. Don't let anybody tell you that foldback stories are no good because the player cannot change certain inescapable events. Sometimes events just are inescapable. The burning of Atlanta in Gone With the Wind is an inescapable event. If you insist that in your storytelling game Scarlett O'Hara be able to prevent the burning of Atlanta, then she's not Scarlett O'Hara, she's Wonder Woman.

Do what works for your player and your product. The meta-approach is, write a requirements spec for what you want. Ask yourself what you want interactive storytelling to do for you. Then choose an approach that meets your needs. Only you can answer the important questions about narrative immersion, depth of characterization, coherence, credibility, if and how the player influences the plot, multiple endings, and sequels and later exploitation opportunities. Only you can answer this for yourself. No argument on a message board can provide you with the answers to this. Let your answers, not other people's arguments, help you to determine what structure and mechanism you need. [applause ]

OK, folks, thank you very much again. Please fill out your feedback forms!