Andrew Glassner
May 1997

Introduction

So what's a game designer to do? I didn't want to simply copy one of the successful games; that's boring and probably wouldn't work anyway. But I did have a clear feeling that many of the games that I have played have done some things poorly. I can't describe the necessary components of a successful game, but I can identify some of the things that always turn me off. Some are obvious; for example, don't make it easy and attractive for the player to cheat. Other problems are subtler.

In this essay I will talk about the poor principles and techniques that I see repeated time and again in current game design.

I will refrain from using specific games for discussion here. There's no shortage of examples in the current crop of products, but I don't want to single any of them out individually. Most games have their good and bad points, and rather than try to always be fair to each product, I'd rather speak in general terms that apply across many games.

But first, a clarification. By "computer games," I mean games with some resemblance to a story with characters; I am deliberately excluding cartridge games like those played on Nintendo systems; they are very different beasts. The games I'm talking about are generally played on a PC or Mac, and delivered from a CD-ROM or over the Internet, or both. (Note added in 2000: I left this in because it was relevant when I wrote it, but console systems have come a long way since 1997 and now represent some of the best work in the field).

Arcade, Puzzle, Strategy, and Story Game Design

Arcade electronic or computer games began with Pong, and today they've moved up to games like Doom. Real-time response and high-focus playing are the hallmarks of these games. I include games like Tetris in this category, because they depend on real-time responses, even though there's a layer of thinking and planning involved. Arcade games played across the Internet are rare because of the technical difficulty in maintaining high-speed communication between players.

Puzzle games began with Adventure. In these games, the designer creates a series of challenges for you to solve. Like one locked door after another down a long hallway, the only way forward is to solve one of the puzzles in front of you. Sometimes the puzzles are overt: decipher a cryptogram or play a winning game of tic-tac-toe against the computer. Sometimes they are trickier, perhaps requiring you to use an object in an unlikely way (e.g. squirt whipped cream into a frog's ear to make him jump and release a trap door).

Strategy and simulation games typically involve one or more alternating phases of planning and execution. In a machine simulation game, you design an object that implements a success strategy and then watch it perform, perhaps in competition with other machines. In an environment simulator, you make strategic policy decisions that govern the lives of the environment's inhabitants. Explicit strategy games often support many simultaneous players. Each player plans a move (which may be quite complex), and then submits that plan to a central authority. The time allowed for the planning phase may range from minutes to days. Once all the moves have been received, a program combines the player's intentions, the rules of the game, the current situation, and sometimes a bit of randomness to evolve the game to the next state. Each player then gets to see the new situation (or a piece of it), and plans their next move. In extreme cases, the planning time may shrink to zero and the resolution phase may be executed immediately, such as by the application of a rule or the roll of dice.

The most recent trend is what I call story games. These are games that attempt to create some kind of narrative thread in which the player is involved. There's a lot of variety in this category. Because it's the kind of game I worked on, the rest of this essay will be devoted to this sort of game. In today's marketplace, most story games are actually puzzle games with a running story to unify or motivate the puzzles.

The next few sections tackle some of the design flaws that I see appearing in one story game after another. I call these flaws because these are design principles that don't work for me - when I play a game and I find myself offended, bored, or otherwise turned off, I have tried to understand when the problem is simply one of implementation, and when it is a bad structural decision. My comments below are addressed to those design decisions that I find just don't work.

Why Argue With Success?

Many of today's games are successful. That is, they sell a lot of copies, make a lot of money, and keep a lot of people happily entertained for many hours. Why argue with that sort of success?

My answer is completely subjective: I have found that I enjoy these games less and less as I play more and more of them. I see the same strategies repeated again and again, and I get bored by the impoverished design. Although there is a lot of money being spent to raise production values, the design of today's games is still primitive. In the worst games - again, some very popular - I feel diminished as a person, or even that my intelligence or individuality is insulted.

So why do I (and, I believe, many others) play them? Novelty. There is something exciting about the potential of today's games, enfeebled as they are. But novelty is transient. I have already begun to bore of most genres of computer games, and I predict that as time goes one, other players will come to share my apathy. Sales and interest will drop as novelty wears off if we don't create something enduring to take its place. That's why it's important to try to figure out how to build better games now.

The Myth of Interactivity

The first - and perhaps most pernicious - design flaw is a result of a belief that you can hear routinely discussed and solemnly acknowledged almost everywhere people are doing interactive design work. It's dangerous because it can sound right on first hearing, and it can be invoked as a weapon. I call this belief the Myth of Interactivity:

The Myth of Interactivity: Interactivity makes games better, and a game designer should try to make the experience as richly interactive as possible.

Like all myths, it contains a kernel of truth, but it should not be taken literally.

The basic fallacy behind this myth is that it elevates interactivity to a special status above other game elements. In fact, interactivity is simply a quality or attribute of a game (or even moments in the game) like the abstract qualities of genre and mood or the concrete qualities of color and sound.

Interactive entertainment is nothing new. We've had the occasional interactive story since the first campfire, and even interactive literature (where you turn to one of several pages depending on your answer to a question) for many years; I remember reading such a book as a kid. Computer games that respond to players are nothing worth getting worked up about.

Let me state baldly that game quality is not correlated to interaction quantity.

On the other hand, interactivity is certainly important when used well. Unfortunately, discussing these ideas is hard because the word "interactivity" has been used in so many ways that no two people are likely to agree on its meaning. So my first step will be to define participation, which refers to quality interactive experiences.

Interactivity vs. Participation

If there is a portion of a game where the player is not called upon to act in some active way, we say that the game (or experience) is "passive." This label often seems to create anxiety. There is a common view that says that we must avoid passive experiences as much as possible.

The most compelling defense for this view is that in our fast-paced and competitive world, a player must be constantly engaged or she will walk away (or click away, on the net). If the experience becomes passive, goes the argument, engagement is lost, and the player will become bored and leave.

It is certainly true that a bored player will leave a game. But simply barraging her with demands for interaction (e.g. things to click, choices to make) does not make an experience worthwhile. The maze-bound rat must always choose a direction at each intersection, but this does not make its life fun, or even interesting.

And those are the key words. Any interaction in a computer game must be fun, or at the very least, interesting. If it is simply a hoop through which the player must jump, the player will sense the subterfuge and resent it; the result is that the interactive moment will work against the game, rather than for it. I call a quality interactive experience a participatory experience, and define it this way:

A quality interactive experience, also called a participatory experience, is a chance for the player to have an enjoyable or illuminating insight, or an exchange with the game designer or other players: the experience must be fun, interesting, or both.

This may sound obvious, but there are endless examples of games available today, some of which are highly regarded, which are loaded with pointless interactive moments. Working an ATM machine is certainly an interactive experience, but it is neither fun enough nor interesting enough to deserve to be called participatory, and does not belong in a game.

Weak Interaction

I believe that most of the interaction found in today's games is either pointless or actually has a negative result. The discussion below isolates a number of specific interactive forms and their inherent problems.

Needless Demands

Game designers should be just as considerate of their player's time and needs. Unfortunately, many games require you to go through whole sequences of actions that are unnecessary. Sometimes you need to do them more than once, which is particularly infuriating. For example, many games require you to move back and forth between two locations several times. To do this, you must plod step-by-step through all the intervening locations, even though you've been to each one before and have no reason to re-visit them. You accomplish your task, and then repeat the dreary process to return. This is interactive all right, but horrible. Neither fun nor interesting, it doesn't come close to being participatory.

However, this sort of thing persists. One reason is because games sell better if they promise a certain number of hours of play; it helps the player feel she is making a reasonable financial investment since there's going to be a payback in a lot of fun. But there's the rub: you don't get back fun, you get back tedium. Walking through those same rooms and corridors time after time doesn't make me feel like I'm really in the environment; rather, it makes me angry that my time is being wasted and thereby removes me from the game's world.

Sure, we can't teleport in the real world, but we would if we could. In the computer, we can, so we should.

I played a game recently that applied this principle nicely. One of the game's features is that you can select two objects that you are carrying with you and order the computer to "combine" them. If this makes any sense, it does so. For example, combining a light bulb with a lamp means that the bulb is screwed into the lamp's socket. Sometimes very complex things get assembled for you in response to this single command. That's good design.

A game should offer the fastest and easiest possible way to do everything unless there is some entertaining or informative reason to prevent it.

Deception

Bad deceptions are objectionable to everyone. Imagine a child who calls a parent on the phone, crying from a terrible injury; when the parent reacts in shock, the child laughs and reveals that it was just a prank. This isn't funny to anyone. Few game players enjoy similar pranks, particularly since they've paid time and money for the pleasure.

Now consider a computer game where a player finds a scuba outfit lying by the side of a mysterious but inviting lake. The player has heard stories of a sunken ship and a secret it carries; clearly the right thing to do here is to put on the suit and head into the lake. Checking the air gauge (which reads full), the player dons the gear and enters the water. A few minutes later, far from shore and deep in the water, the air runs out. As the player fights her way back to safety, she realizes that the dial was defective and the air tanks were almost empty when she found them. Had she known this, she never would have gone in.

When this happens in a book or film, it can be very exciting. In the best examples, we identify with the character and feel her panic, and we feel angry with the villain who placed our hero in this predicament. We learn about the hero by how she responds to the problem, and are carried along by our fear and our trust. We trust the writer of the story to make it satisfying for us, and worth our time and concern.

In today's story games, we do not have a given narrative. We are constantly reminded that we are the hero of the story, and the things we do determine what will happen. It is not our hero who has been tricked, it's us. It's not the evil villain who left the broken scuba gear lying around, it's the game designer. We are in constant conversation with the game creator, more than we are with almost any author or screenwriter. Bad things done to the hero in fact happen to us, personally.

Sometimes you hear this defended with an appeal to verisimilitude, claiming that we run across deceptive information all the time in the real world. True enough. And in the real world, we also get false promises from politicians and lies from drug companies. The appeal is weak - just because deception appears in the real world doesn't mean it belongs in our entertainments.

Mistakes aren't always bad. Part of the fun of learning about a new environment is learning a new set of rules. And part of that experience means making mistakes by misinterpreting instructions, or mistaking a new object for a familiar old one and using it incorrectly. That's learning.

But a deliberate deception is an artificial roadblock, created solely to make life harder for the player. That's not fun, that's annoying.

The game designer should not deliberately deceive the player unless it leads to insight.

Player Profiling

But the designer of such a game is faced with a difficult problem: how to characterize the subtlety of a player's psychology? It's hard enough for people to understand each other, much less design a computerized analyst.

Often the result is a crude series of choices - in effect, replicating the standard multiple-choice diagnostic personality inventories in the disguise of making game decisions (this is why the technique is so easily applied in branching narratives). Even children see right through the disguised choices. Rather than deciding what to do in a situation based on their own feelings of right and wrong, children make choices for the protagonist they are controlling because they want to "make him mean" or "see what happens if you make her really stupid." They game the game - that is, they attempt to influence the underlying mechanism of the game. In other words, they are not playing the game itself. This marks failure for a game designer. By analogy, when a filmgoer starts thinking about the special effects or makeup, the film director has failed.

Adults see through these personality tests as easily as children, but because they are more aware of the fact that they are being labeled and pigeonholed they are more likely to resent it. And having paid for the game and invested valuable time, they are apt to be offended by this attempt to characterize the complexity of their personality by a trivial scoring system. In effect, the player is being told that subtlety of expression and depth of understanding are irrelevant. The player is clearly aware of the need to think about how the game will interpret her choices, which removes her from immersion in the game itself.

Don't trick players into providing a personality description.

Randomness

Theme and variation is a basic building block of literature, but random variations are not thoughtful ones. Nobody would accuse Bach simply employing random variations on his themes; the delight in hearing his melodies transform and re-arrange comes from the quality of the intention behind the changes. In other words, theme and variation is a structure for intention. Random variation is unappealing; otherwise we would have countless musical and visual computer-mediated works which simply created endless random variations on a few human-created building blocks. They'd be out there if anyone cared to own them. But who would want such a thing? There isn't enough time to absorb the carefully crafted works by talented people (whether those works be deep or simply entertaining); why would anyone want to instead subject themselves to a computerized creation? Computers have nothing to say.

Sometimes randomness is used to introduce surprise into a game. A lake may be peaceful the first ten times you row across it, and then suddenly turn into a whitecapped mess the eleventh time. There are small variations in nature, and randomness can cover that, but large randomness just makes things different for the sake of being different, not because there's any purpose or reason behind the change.

In a game, the player is in dialog with the game designer. And when there is no message in that dialog, the conversation stops. Randomness is a substitute for deliberate intention, and without intention, the value of the conversation drops precipitously.

Avoid large-scale randomness.

Multiple-Choice Conversations

Today's technology is a very long way from creating programs that can understand an arbitrary question (whether typed or spoken), and then automatically generate an appropriate response. So everything must be pre-scripted. When in a conversation with another character in a game, you are given a list of questions and statements to choose from. Your character utters the appropriate one, the other character responds to the tone, style, and substance, and then terminates the conversation or waits for you to choose another thing to say; the list may change in response to your previous choice.

There are a few problems with this scenario.

Most importantly, the choice you want may not be on the list. Suppose that you are stranded at a roadside, and someone stops to offer you a lift. You may either: politely thank him and get in, politely decline, ask his name, or ask how far he is going. If one of these fits your desires, great.

But this is a game, not the SATs. There's no right answer, and no best answer. Most games go to great lengths to encourage you to act as you see fit, so that you feel immersed in the game.

Suppose that you think you recognize the driver as the rude man you argued with in the coffeeshop earlier. You really want to confirm his identity, but that option's not available. The basic principle of the game has been violated: you can't do what you want. The only response is to either choose at random, or else game the game, meaning slipping out of the game's world and trying to figure out what the different courses of action will lead you to and picking the one that's most attractive (or least unattractive).

This is a disaster for a story-based game; the player is ripped right out of the game's world and all the production values in the world won't bring her back in, at least not for a while.

The apparent ways to save the multiple-choice scenario are to include an enormous number of choices, or get the choices exactly right so every player can find just what they want.

Both of these approaches just make a bad problem worse. Suppose that through extensive testing the game designer realizes the necessity of including a polite question, asking the man where he had lunch. Close, but no cigar. You're still trying to emulate human conversation, which is subtle and complex. Suppose you want to ask the question in an innocent way, or a slightly brusque way, or even an accusing way. There's no way to script for every possible nuance. So even a huge list of choices is too small.

So in every case you must leave the game's world and try to guess what the right answer should be. If you, as a player, are talking to a robot or a Coke machine, this is what you'd expect. But not from a person. The fact that the game masquerades these robots as people is ridiculous; it's obviously untrue.

But what's worse is that you, the player, have become just as much a robot as the story's characters. They have scripted, pre-authored questions and answers; so do you. They are limited to what the game designer invented for them; so are you. They have no autonomy, no individuality, no creativity, no imagination; in this game, neither do you.

Avoid multiple-choice conversations.

Don't Keep 'em Guessing

This is pointless interaction. If you're not acting intentionally, then why are you acting at all? If you don't understand the choice, your value as a participant is very small; as with other poor modes of interaction, the player is diminished by this situation.

Some designers claim that this is a reflection of the real world, where we often don't understand the nature of our choices. True enough. But entertainment is not the same as real life. Players do not spend time, money, and energy to be faced with vague situations and feelings of lack of control. We get enough of that every day.

In art, we are willing to tolerate a certain amount of deliberate ambiguity. Carefully controlled ambiguity allows an audience to enter into the work and interpret it personally. But too much ambiguity means the artist has been too lazy to properly shape the work. Similarly in entertainment, we want clarity of character and story. And when choice is added to the environment, we want clarity of choice. Options that are poorly understood are false options.

Describe choices clearly.

Repeat Painlessly

In one game I played recently you assemble some pieces in what you think is the right order, and then push a big button to submit your answer. Pushing the button initiates a sequence of visual and audio effects, simulating some big machine "examining" your answer. Eventually it might tell you that some of the clips are in the wrong order, and then you hear some audio encouraging you to keep trying. All of this takes about 15 seconds, but it feels like a half-hour. By the third or fourth time I submitted an answer I was resentful that I was forced to waste my time waiting for this now-boring effect to repeat. There was no way to hurry it up or skip over it. By the tenth time I went through the process I was ready to climb the walls.

I suspect that designers do this because they work hard on these entertaining segments and want to get a lot of mileage out of them. But too much mileage and their charm is lost, ending in self-defeat. A far better approach provides the elaborate and time-consuming feedback the first time or two, and then switches to a highly abbreviated version that ends almost immediately. The player will remember the production piece but will not become angry or bored by being forced to sit through it endlessly.

Do not repeat unless the player requests it.

Arbitrary Complexity

These puzzles often depend on magic items; these are objects that you find lying around the environment, often physically and temporally far from where they're needed. For example, near the start of a game you might find a sharp pencil on your office desk in San Francisco, and at the end of the game you need that pencil to defuse a bomb while in a spaceship orbiting Mars. Nothing else will do - a hatpin that you found in Los Angeles will not work. Two conventions of puzzle and story games are that you need to pick up and save all magic items when you run across them, and that their ultimate use is generally impossible to guess when you first encounter them.

There are two general approaches to helping the player locate and identify magic items. The first is to distinguish them from the environment; in the example above, a pulsing red spotlight might draw your attention to the pencil on your desk. This is the height of artificiality, even when the attractant is subtler, such as a thick black outline around the object. Drawing attention to mundane but ultimately necessary objects has no correlation to the real world; it's just an artifice used by game designers.

The other approach is to make you search and test for magic items, so that you just "accidentally" turn them up in the course of playing the game. For example, you might move your arrow-shaped cursor all over the desktop, and when it moves on top of the pencil your cursor turns into a hand, indicating that you can pick up the pencil. The testing might be more explicit - you might have to actually click on the pencil to discover you can pick it up. Why, you might wonder to yourself, can't you pick up the scissors? Or the paper? Or the phone? The answer, of course, is that they aren't magic objects, and aren't needed later in the game. If you could pick up everything, everywhere, then you would, and you would quickly have an inventory far too large to manage. It would also make programming the game enormously more difficult, because the implementers would need to accommodate your defusing the bomb with any pointy thing, from a paper clip to a toothpick.

Whether magic items are indicated explicitly, or are revealed when you test them, finding magic items requires you to perform an exhaustive search. And this is indeed exhausting. When I purchase a game for pleasure, the last thing I want is to be forced to move at a snail's pace through each and every room, running my cursor over every visible surface, clicking on every visible object just in case it's magical. I recall one game where I stood in the midst of a long tunnel built of bricks. There were thousands of bricks. A clue earlier on told me that there was a secret "behind a brick". I had to spend about an hour moving through each section of the tunnel, punctiliously clicking on every single brick until I eventually found one that moved a little bit; clicking a few more times worked it loose and revealed a key behind it. This key was essential to continuing the game. I can't imagine that this boring, tedious process could really be much fun for anybody, and finding the key at the end did not make things all better.

Avoid arbitrary complexity and magic items.

Spackled Interaction

You can spot spackled interaction by asking yourself whether the game designer and the interaction designer ever needed to speak to one another.

The remote control for my home VCR is a terrific example of this. The bulk of the unit (made by a famous manufacturer who should know better) contains ten numbered buttons. To choose a channel for viewing, I simply press the one or two digits that make up the channel number, and the machine jumps to that channel. However, when I want to program the VCR to record a program later, I must use a different set of buttons along the top of the unit. To program the VCR, I go into "programming mode", and set the time of the show. To specify the channel to record, I press the same little numbered buttons, right? Wrong. The ten buttons on the keypad do nothing at all. Instead, incredibly enough, I must use two buttons near the top of the remote control, marked "+" and "(". If I want to record something on channel 39, I move my cursor to the channel field (which defaults to 1), and then press the "+" button 38 times. No kidding. Whoever designed the programming hardware never spoke to the person designing the channel-switching hardware, and neither of them thought through or field-tested the integrated product.

This kind of disconnected interaction is common in many games. The person who designs the heart of the game and the way it works leaves the interaction tasks for someone else. The game is not conceptually whole, and the player feels that she is manipulating it from a distance.

Alternating Cut-Scenes and Interactive Periods

In the best games, the puzzles are tightly integrated into the story, relying on your knowledge of the characters and the world to find a solution. Typically, when you're in "interactive mode" you have as much time as you like; the world of the game is suspended while you experiment with the puzzles, and wander around the environment collecting other information (e.g. clues and magic items) to help you solve the puzzles. After key puzzles have been completed, the game goes into "film mode" and plays for you a pre-produced cut-scene that shows the next few steps in the plot unfolding. Lots and lots of games work this way.

I find this terribly offensive.

The problem is that the game is sending you, the player, mixed messages.

When you are working on puzzles, the game is telling you that your skill and imagination matter. Your understanding of the world and its characters through your creative insight are important. You are encouraged to build up an internal representation of the world. In the best games, you need to understand the psychology of the other characters in the story, and take actions that will lead them to respond in appropriate ways. I find this very exciting, and I enjoy immersing myself in the game's world. My time is my own, my character is me, and I act (within the game's limits) as I wish.

Of course, each puzzle has a correct answer, just like a crossword puzzle has a correct answer. My individuality is largely an illusion, since I will end up at the same place as every other solver of the puzzle. But the enjoyment of the game can be strong enough so this isn't a problem - consider the legions of people who love doing crosswords. When the game is good, the fun of working it is its own reward, and there's nothing to break your personal style of involvement.

Then the cut-scene takes over, and this illusion is utterly annihilated. My character starts acting in ways that I would never dream of. I cannot stop things from going wrong, I cannot control what goes on - I am irrelevant. The cut-scene plays at its own speed, hurtling the story forward and carrying me along with it. Virtually always these cut-scenes are the cinematically interesting ones, which means that they have lots of action. Character is revealed by action; what the hero does on the screen tells the viewer much of who that character is.

But wait a second, that character is me!

Well, only during "interactive mode", when I am in total control. My speed, my choices. Then I am abruptly jerked into "cut-scene mode", where I am a totally passive spectator as the action plays out at its own speed, and my character does things I would never do. When the cut-scene ends, I am thrown back into interactive mode, and suddenly I am left with the problem of cleaning up the new situation. A situation that is result of my actions, which I never made!

And that's why I get offended. I'm told that I need to exercise my imagination and involve myself in the world and its characters, and then I am shown that I got it wrong. Even your own character acts in ways that disagree with your intentions, and this is the worst: every now and then control of "yourself" is ripped out of your hands, and you the player can only sit passively while the onscreen "you" acts incorrectly. This must be what it feels like to be schizophrenic.

The game repeatedly and alternately requires you to be creative and inhabit your character, and then denies your creativity and yanks the character away from you. This is bad design.

Never take over control of the player's character.

Interactive Fiction

Which is not to say that it isn't popular. If you browse the shelves or web pages for today's games, you'll find lots of enticements along the lines of "Ten Different Gameplays!", "Six Different Endings!", and so on. Then there's the more direct kind of involvement: "You Control the Action!", "A Customized Story Based on Your Choices!".

Well, sure, we can write programs that respond to input. And we can probably use that facility to create new and old kinds of art that say things about our selves and our culture. But it's a big leap to think that this is a good way to tell a story.

The Myth of Interactivity Redux

It's the Myth of Interactivity again - recall that this myth tells us: Interactivity makes games better, and a game designer should try to make the experience as richly interactive as possible. And what goes for regular games goes for story games. This belief in the universal power of interactivity is what leads people to try to marry interaction and storytelling.

Story games that are based on some form of "interactive fiction" offer you two incentives: they can be replayed (thereby increasing the value of your purchase), and you can control the development and resolution of the story.

The first appeal is largely illusion. Why would you want to play the game a second time? Much of the fun of interactive games is learning about the world and the characters and how things work. Once you've played all the way through, you know these things. In the best case, you've seen a satisfying story and had a good time. Why play through it again with just a slightly different twist on things? Most of what you'll encounter you'll have seen before, though there will be some differences. But you've already seen the game and reached a satisfying resolution. In my experience, most players investigate some of the alternative endings simply to see what they're like, not because they want to play the game again.

The second appeal of interactive fiction is your participation. It's easy to observe that in our culture only the occasional curiosity has involved the listener in the telling of the story. This has nothing to do with production values or technology, but goes right to the heart of storytelling. It comes down to this: good stories are carefully constructed by skilled storytellers. Whether constructed intuitively or explicitly, a storyteller uses craft to engage an audience, keep them, and then satisfy them at the end.

Creating a work of fiction is not easy. That's why we have books and college classes on the subject. A skilled author understands her characters, and can place them in situations where they reveal themselves through action. The audience learns about the character from the author; if the audience selects how that character should react to a situation then the character will not be acting from a central core, but erratically. We call such behavior psychotic, and notwithstanding a few exceptions, psychotics are not very interesting to watch and don't make for good stories.

So traditionally we trust our storytellers to create interesting and appealing characters whom we follow through their interesting and revealing story. If we can simply cause characters to react in any old way, we've lost both the character and the plot.

The exception to this is where the characters are deliberately stand-ins without personality. For example, the little computerized people on an electronic football field aren't anyone we care about personally, and in most football games we can control their actions without worrying about consistency (some games try to model the players on an individual basis, which is a different situation). Similarly, adrenaline games can get away with unmotivated characters, who are really just surrogates for ourselves in pressure situations.

But to create a successful work of "interactive fiction" with meaningful characters, every possible storyline through the narrative has to be as satisfying as the single storyline in a successful work of traditional fiction. Otherwise one gets a substandard story, and then has the penalty of having participated in the creation of this lackluster beast. That's hardly satisfying. And given the significant effort required by talented storytellers to create a single good storyline, imagine the difficulty of making something that will work across multiple possibilities. How could you handle foreshadowing? Growing tension? Theme and repetition? Plants and payoffs? Pacing? The list goes on and on - all the traditional tools of storytelling become far more difficult to apply.

The successful author of a piece of interactive fiction will end up investing as much effort in the work as would be required by several traditional stories. What justifies this effort? The fact that the reader has the opportunity to interact with the story is the only incentive. But I argued above that people don't want this: they've not wanted in the past when it was possible, and simply adding cool technology to the equation doesn't make story interaction any more rewarding. It's novel, and has some limited, short-term appeal, but novelty fades.

To summarize, the fundamental problems of "interactive fiction" are twofold. First, the author must do much more work to create a piece of comparable quality to a non-interactive work. Second, when a listener/viewer/player controls characters in a story, she can interfere with the character's personality, which fatally injures the development of the character and leads to a psychotic personality and uninteresting story.

Note that exploration games, primarily aimed at children, don't suffer from this problem because they are largely designed in such a way that the story line is stopped, the player explores the environment of the current "page", and then the story is resumed when the page is turned. The player isn't really interacting with the story itself, just exploring the environment in which it develops.

Computerized Narratives

This strikes me as a remarkably fruitless endeavor. There is little enough time for any of us to read, view, hear, and otherwise absorb the carefully crafted stories and works that other people have created for us. Whether simple diversions or deep philosophical studies, they are all the result of a creative person who is striving to communicate to us. And they all have much more to say than any computer. Computer-assisted media are just fine; computer-generated content is inherently empty.

Computerized Characters

One popular model for creating synthetic characters is to build a core personality into a computerized character. Careful design can create a character with both consistent traits and idiosyncrasies. Bringing out these traits in a book or play is difficult enough; trying to display them within the context of an unpredictable game is very difficult. Today's trend seems to be the simulation of a complex character by placing a layer of random behavior over the character's programmed behavior. They may be speaking to you calmly, but when some unseen roll of the dice comes up larger than a pre-defined number, the character becomes more hostile.

Randomized procedural behavior creates psychotic robots.

Sane people do not act randomly. They act in ways that may appear random, if we don't know them well enough, but their actions are integrated into their personality. Even when a person holds simultaneous, opposing beliefs, the expression of those beliefs is a result of their history and personality. In short, there is reason and justification behind the strangest acts.

Unless there is truly no justification at all. Then the person is acting crazy, or psychotic. That's the label we use when people act in ways that are really random. And that's what behavioral control with a layer of randomness simulates. This may be a useful technique for simulating a psychotic character, but is no good for people that are supposed to be helpful, dependable, or sane.

Interaction and Stories

Every one of us can make up and receive stories all the time. We talk to friends, read books, watch films and television, and attend to stories in the world because they allow us to see into the hearts and minds of other people, and into worlds with which we are unfamiliar. When receiving a story, we trust the author to make the experience fulfilling and worth our time and effort. When we take over some of the storyteller's job, we are interfering with that process, and severing the fundamental relationship between storyteller and audience.

There might be new ways to bring these roles together, but they won't be fiction, and we ought not to waste time trying to stick close to fictive forms. At best, we will create sterile hybrids that simultaneously break the author-reader bond while pretending it's still there.

Participatory Design

At the heart of every good design will be respect for the player. And that means that the player's time will never be wasted, and the game will never insult or deny her individuality, intelligence, or creativity. The player will always be engaged in fun or interesting events, whether she is receiving them passively or interacting with them.

The powers of the new technologies appear on two fronts, and they both derive from the idea behind the participatory experience, rather than the merely interactive one. I believe that each of these fronts offers a promise of finding the enduring qualities of computer-mediated games that will last.

Person to Person

On the net, there are always people available. As long as people have even a sporting chance of finding a compatible teammate or opponent, I believe they will take advantage of the opportunity. Two people who meet online in the midst of a game already have a common interest and a ready topic, which is already more than you usually find with a stranger at a party.

Person to Culture

But a game can elevate this, even if only by a small amount. The player still is sitting in a chair, typing and clicking. But because the player is immersed in a game, he or she will be exposed to the environment of the game. And just as with all entertainments, this can carry along an exposure to a new idea, place, or culture. Playing a game set in a rainforest canopy is not the same as "experiencing" being in a rainforest canopy, no matter how sophisticated the technology, but it's a lot better than knowing nothing about rainforest canopies at all. The same goes for games set in 1940's Poland, Ancient Rome, an anthill, or downtown Seattle. Executed well, games can all offer something more than simple diversion.

Less Is More

Consider interaction with computer-controlled characters. In the discussion above I denigrated the use of multiple-choice menus for carrying on conversations. We don't have the ability to carry on general conversations with computer characters, and just adding on randomness makes them appear dim-witted or crazy. But we can create very limited personalities.

Simulating a human by using a weak personality model leads to trouble, because we can't deliver on the expectations. But we can dress up a small appliance in a tiny personality, and it's going to do fine. Who's going to get upset because a responsive desk lamp misunderstood the nuance of a request? We don't expect too much brilliance from worms or birds, so we won't be too disappointed when they fail to deliver it. By scaling back to a simple character we can use our best techniques and they will fit. Players don't expect much from only slightly self-aware toasters.

Similarly, we can accept the need to talk with the environment in very limited ways. Right now, I can only express myself to my blender through ten push-button switches. If I were given a very limited syntax and grammar, it would be a step up. I might ask the blender whether the bananas are too squishy, and it could give me back a five-word sentence. It's not great conversation, but it's interaction that doesn't fail expectations.

This general rule applies to all aspects of game design. If we can't do something well, then we should redesign its use until we are applying the technique at a level we can pull off smoothly. Pushing technology is fine, and game designers hardly need to be encouraged to do so. But I think many of today's problems are a result of game designers shooting high (which is fine), but then failing to acknowledge where they actually ended up.

In game design, we should embrace the limits of our understanding and technology, and strive for the best work within (or slightly beyond) those limits. All of the design problems discussed in the majority of this essay are a result of designers implicitly promising more than they could achieve, and thereby delivering disappointment. Promise less, and you can deliver beautifully, delighting and entertaining players.

Wrapping Up

It would be a mistake to place too much importance on what a game can deliver - it is, after all, just a game - but it would be equally unwise to miss the opportunity to give the player a chance to expand her world, even in a limited way.

Games can be much more than simple diversions. Successful board and card games have combined simplicity, psychology, and respect for the player to provide an enduring challenge. The best computer and network games will do the same.