[In an in-depth design article, People Can Fly (Painkiller) designer Wesolowski looks at games from Freelancer through Thief and beyond to examine the all-important art of correct pacing in video games.]

Good pacing can be essential to gameplay, just like it's essential to great adventures on the big screen. Game developers often look up to Hollywood for guidance, which seems to work well -- for at least some titles.

But there are major differences between the media, ones that we cannot ignore. Following big screen examples helped us make our baby steps, but ultimately it's a dead end, because it turns good games into bad movies.

Cinematic Experience

Modern media are younger than we realize. My grandmother remembers Life Without Radio; my parents remember Life Without TV; and I remember Life Without Computers. By contrast, both drama and literature were widespread in ancient Greece.

The first computer games were created some fifty years ago, but another two decades had to pass until there were enough players around to talk about them. Some keep complaining about the lack of critical vocabulary to this day. On the scale of cultural maturity, we're toddlers.

Like all children, we haven't really made up our minds yet. Instead, we look up to other media, and film in particular. But cinema is only about a hundred years old itself, so we're toddlers looking up to high schoolers. We think the bigger kid is so much like us, because it relies on sound and animated images like we do.

Films, too, require teamwork, merge art with advanced technology, and have similarly large budgets. And yet, they are so much more successful that we feel a little envious at times. We don't look up to just any film. Our ambition is to be like Star Wars and James Bond when we grow up.

There is an obvious difference between the toddler and the high schooler. Games are interactive. They're meant to let people do things. But we're hoping that we can at least present those things to players the way movies do, and let them respond, somehow, without ruining the cinematic experience. I think we're fooling ourselves. The big kid doesn't even like us.

Pacing in Star Wars. Intensity is an increasing wave of peaks and reliefs. Each peak and slope can be associated with some significant occurence. There are eight peaks on this picture.

There are a number of misleading similarities between computer games and blockbuster films. Most notably, an archetypal story structure called Hero's Journey can be applied directly to game narrative. All of its key concepts translate easily into an interactive setting.

The player is, obviously, the Hero. The combat loop, which constitutes the biggest part of gameplay in most action games, corresponds to confrontation sequences. Missions and levels correspond to trials on the Hero's path, often personified by End of Level Bosses. Cutscenes and briefings serve as exposition sequences and provide relief from intense action.

Bigger enemies, better equipment, and levelling up all correspond to Hero's inner growth, while repetitive tasks and minigames are recurring themes, resemblant of the Voight-Kampff test from Blade Runner.

From these similarities emerges the crucial concept of pacing as an art of governing audience emotions. The hypothesis is that if we arrange events in an "increasing wave" of intensity, we will achieve the cinematic effect of mounting tension. Just like in a movie, when the game ends, the audience shall be left shaken and wanting more.

It would probably work very well -- if only players were an audience.

Plot Twist

A film audience is passive. It just sits and watches. The film may influence them in many ways, but they can only rewind it and watch it again. Their state of mind is affected, but their ability to watch the film isn't.

In contrast, players are bombarded with stimuli which affect their ability to respond to subsequent stimuli. The most obvious case of this is the broadly-defined learning curve. Even if players do notice all your hints and prompts, how do you make sure they have drawn the intended conclusions? When miscommunication happens, a film just goes on at its own pace; a game deviates from intended course due to player interference.

The only exception to this is a cutscene, but cutscenes aren't interactive. So we compromise. We make sure there is only one point of entry and one exit available. We take care not to allow players to look in a "wrong" direction when there is something we feel they have to see. We avoid situations that may take some time to figure out. We use quick time events.

Our efforts are futile: the interactivity is lost, but a truly cinematic experience doesn't appear, because we're unable to achieve a movie-like pacing. Our dubious practices are so limiting we keep making the same few games over and over again. The narrow category of Tower Defense clones displays as much diversity as the whole genre of First Person Shooters.

In order to see why this happens, let's take a look at how intensity works in computer games. Simply put, we can either escalate sensory stimuli or build up abstract meaning. The former happens when guns, explosions or enemies get bigger, putting the Hero in a greater danger than before. The latter happens when each part of a narrative means something -- but together they mean something else. "I am your father, Luke" is more than just a paternity acknowledgement.

These two kinds of intensity tickle different parts of our brains. Escalation is visceral and relies on our perception, while meaning buildup is cognitive and relies on our understanding. Escalation is temporary, because it's easy to replace a big gun with a small one.

Meaning lasts. Once we learn to like a character, it takes a lot to convince us to hate them. Do you remember how many Berserkers there were in Gears of War? Do you remember what their main weak point was?

Player Gets in the Way

The problem is, we have much less control over pacing in a game than in a film. Cinematic pacing is completely under the director's control, and the audience's focus is led by the camera. They simply cannot look the other way. Subtle cues work reliably, because the audience is much less focused on acting on their own, and thus more focused on paying attention.

It's possible to make instant switches between action and story, as proven by Luke's Trench Run in Star Wars. He's in the middle of a fight, but there's just enough time for him to take a look at the damaged R2-D2 and listen to Obi-Wan's advice. In a game, such a quick alternation between lower and higher intensity disorients the player. Every time we take the control over the Hero from the player, their ability to keep track of what they've been doing is disrupted.

No game is linear. There are many ways for the player to digress: bonus stages, side quests, user generated content, loadout screens, menus -- even background vistas. A film director can rely on the audience watching the film in a preset order, in a controlled environment. Players are perpetually distracted, usually by the game itself.

Another issue is that most games rely heavily on repetition, which is the bane of traditional narrative. When players enjoy a task, they usually don't mind trying it again, whereas in a film there's no point in watching the same scene five times in a row. The Voigt-Kampff test in Blade Runner is performed more than once, because it means something different on each occasion.

Repetition in games also means it's easier to notice patterns. Simple pacing structures, such as "increasing wave", are very predictable when they have too many iterations. Ironically, the same mechanism that helps create surprises in a film, makes it easier for players to anticipate obstacles and plot twists.

Freelancer is a textbook example of cinematic pacing in a computer game. There are seventeen peaks on this picture. The first major plot twist occurs in Mission 4.

Pacing by escalation is hampered by the game's relative size. Typically, games are between three and twenty times as long as films, which means intensity progression is much slower, and thus harder to notice.

A designer doesn't have much choice: each new obstacle is not more menacing, just different. Average intensity doesn't start to increase significantly until near the end of the game.

There are also more individual action sequences than in a film. In each part of the original Star Wars trilogy, Darth Vader participates in only one duel. In the 2008 edition of Prince of Persia, there are no less than 27 duels between the Prince and one of five major antagonists, and that's a game which tries very hard to make every fight count.

In a blockbuster film, each trial on the Hero's path is unique and significant -- it adds something to the big picture. In an action game, such density of narrative is very uncommon. Enemies are just cannon fodder.

Pacing by meaning buildup is hampered by the failure of most games to create a good connection between the gameplay and the narrative. They are habitually being separated from each other so that player actions don't get in the way of carefully crafted film sequences. There are basically two streams of information running in parallel.

As a result, noninteractive parts are not immediately relevant to the player, and they get pushed aside, because gameplay conquers player's attention as soon as a cutscene ends. Since they tend to remember so little, their comprehension is spotty and meaning fails to build up.

There is a tradeoff with regard to cutscene lengths. Short ones tend to work more efficiently, because they convey simpler and more relevant messages: this is the enemy, this is your gun, this is the exit. But they aren't so good for exposition.

Both escalation and meaning buildup are heavily influenced by player's individual perception. You can give them a bigger gun, but they may still like the previous one better, because it made a nicer sound, or because it fit their style. You can give them a useless but funny sidekick, and they may not appreciate it if they are challenge-oriented.

Audience tastes differ, too, but most filmgoers will agree they watch films for one or more aspects of the narrative. A player may or may not care about other players, the story, the challenge, the opportunity to express themselves, and so on. The game's message needs to be much more robust, because you never know which part the player is paying the most attention to. There's a need for redundancy -- another bane of traditional narrative.

As a result, pacing in games is no longer capable of serving its original purpose, which is to keep the audience interested. It does, however, fill a very important role: it helps the player stay in "the flow".

The clear rythm it gives to the game helps learn to anticipate challenges and get ready right on time. Providing relief in regular intervals prevents players from getting tired too quickly. Sudden surges of intensity supply them with adrenaline rushes, and the increasing wave structure makes sure they have time to recover afterwards. But we don't really need the increasing wave in an action game. We just need those surges.

Each of twenty regular levels in the 2008 edition of Prince of Persia has the same intensity pattern. The two duels are intense, but much shorter than the other two sections. The last section is partly optional.

So, if we don't really need, and cannot really use cinematic pacing, then why does everybody stick with it? Actually, that's not true. Many games -- some of them very successful -- don't even try to retain a movie-like pattern. Of all the numerous examples, four seem particularly interesting.

Pacing by Player Skill

Consider the original 1989 release of Prince of Persia. Typically for that time, it was short and intended for many playthroughs. The player had an infinite number of lives, but had to beat the game within an hour. Lethal mistakes resulted in moving the player back to the beginning of the level.

The game felt different to novices and veterans. The former experienced a desparate race against all odds, while the latter stepped in the role of an unbeatable hero. It wasn't just a question of difficulty.

Each mistake or delay meant the hostage princess was more likely to die. The gameplay, while simple, allowed for tricks and tactics. Would you rather try a risky running jump through snapping metal jaws, or stop and wait to catch their rythm?

Prince of Persia is an example of pacing by player skill. The faster the player is, the more skill is required. But the more skill, the less need for haste. Both novices and experts were affected by some kind of tension, either from time constraint or danger.

Skillful play was rewarded mainly with freedom to choose between the two. Pacing served as obstacle, reward, and punishment at the same time.

In a film, the audience is not in control, therefore they can suspend their disbelief and pretend it's possible for the Hero not to make it on time. In a game, the player is in control, and they know, rather than believe, whether or not the time constraint is real. If it's not, it won't work. In Prince of Persia, it was always real, even though you could finish the game in under fiteen minutes.

Great Middle Section

Fallout 3 is a crossbreed between a non-linear Role Playing Game and a First Person Shooter. Previous parts of this famous series used a similar story structure, but were not action games. Note how the transition from tactical turn-based combat to action doesn't seem to have brought harm to either the narrative or the gameplay.

In all three parts, players are only forced to play through a short introductory section, and then they are free to go where they like, and do what they like. The main plot doesn't impose itself upon them. Essentially, this means no pacing at all, if players take their liberties to the extreme. However, they are not left clueless, as landmarks and quests draw their attention to specific events and locations.

Free from pressure to move onward, the player can pay more attention to details and subtle relationships. Even though the story is made up of predefined blocks, players create the context of each block by themselves. Films are doubly incapable of this, because all scenes always arrive in the same order, and because they are time-constrained in that they end at director's, rather than the audience's, discretion.

Experimental novels, such as Hopscotch, achieve a similar effect, but to a much smaller degree, because they cannot create events procedurally. It's only a pity that most games try very hard to avoid such complex topics as those explored by Julio Cortazar.

In any case, the great middle section does not prevent Fallout from being interesting to a large number of people. Its main advantage is that the definition of "interesting" is now under player's discretion. Players become their own directors and no longer need us in this role.

Pacing by Setting

Gothic is an old PC game by a small developer who had trouble finding publisher in some countries, which is a shame, because it's also one of very few action-RPG hybrids to ever manage to turn backtracking into fun.

Most of the action in Gothic takes place in a single nonlinear area. Red dots are hubs. Blue dots are mission-specific locations. All features are to scale.

Let's take a look at the map of the game's main area. Note the complex network of roads and rivers. Water is swimmable. Falling from heights is lethal, but many cliffs are climbable. Forests are dangerous, but provide shortcuts. The player spends a lot of time travelling between the three hubs in one of following ways, starting with the slowest:

The fundamental notion of pacing by setting is that there is literally a landmark around every corner. Locations in Gothic are small and nonuniform: the biggest forest is some 200 meters across, and no two ruins are identical. Every feature plays a unique role in the game's world, as they pose varying challenges and offer different rewards.

They also interact with each other in subtle ways. For instance, vistas are even more fun to look at, because they show places player has already been to or can potentially access. A cave near a forest allows the player to lure goblins from the cave and watch them fight wolves who live in the forest.

In other words, landmarks act as interdependent parts of a larger system. The game works consistently, because each landmark has a clearly defined role that makes sense within the setting.

This is a very efficient way to create content. Mission specific locations act as unique levels, but so do the routes in between them. The amount of effort needed on part of development team is proportional to the much lower number of landmarks. Nonlinearity in Gothic actually saves work.

Emergent Pacing

Thief: the Dark Project is brilliant in many ways, but its most incredible feat is how it manages to retain a strong narrative in an environment that gives the player strong initiative. There is no combat loop and almost no predefined scenario within each mission, and yet there is elaborate exposition, tons of dialogue, and a lot of tension.

Pacing in Thief is governed by looped, overlapping patrol routes of guards, who are equipped with autonomous AI. They make simple observations, perform searches, have memory, and communicate with each other.

The gameplay is highly asymmetric: the player can outsmart enemies easily, but combat, while possible, is best avoided.

There's a lot going on, and it's easy to make a mistake, but these are rarely lethal. Instead, player actions propagate throughout their surroundings, influencing nearby guards, who in turn influence the player.

Emotions oscillate between sneaking by an unsuspecting guard, and hiding from an angry mob in furious search of the trespasser. Thus intensity is a complex function of level layout and player decisions.

Tension and pacing are also governed by player's chosen style. The player literally creates their own gameplay within a flexible set of rules. Emergent gameplay adapts automatically to player preferences and level design. Hence, pacing is emergent, too.

A few of the potential outcomes of the simplest possible setup in Thief. Typical situations involve two or more guards.

Bits of story, usually presented as either diaries or conversations, act as reward for succesful infiltration. If you manage to sneak upon people in the middle of a talk, you can attack them, and they will defend themselves.

But most players choose to stay put and listen, partly because it's useful, partly because the writing is good, and partly because the game always encourages them to take their time.

And the best thing is, Thief is a game based mostly on sounds, not visuals. Whenever there's a relevant change in the environment, the player literally hears about it. Even in a game like this it's technically impossible to listen in a "wrong" direction.

Emancipated Audience

There are five lessons in humility we can draw from above examples.

Prince of Persia shows that we don't have to fool the player, which is great, because we can't anyway. In a film, we can, and we should.

Fallout series shows that we don't have to decide for the player on everything, because the player knows better. A film director's role is the opposite of ours.

Gothic shows that the setting will take care of itself, if we take care of underlying principles. A film setting doesn't have this capability.

Thief shows that interactivity and narrative are not mutually exclusive in a computer game. Cinameatic narrative is only possible because of film's complete lack of any interactivity whatsoever.

Most importantly, in all these games pacing is governed almost entirely by their design, and it's still good. It's a complex function of player actions, parametrized by game's dynamics and aesthetics.

It's a dynamic, rather than preset feature, hence it's difficult, and sometimes impossible to specify it upfront and then design a game to it. We can't just take the increasing wave, and slap it onto the gameplay.

We often see ourselves as all-powerful creators of worlds. Our job is to present, and our players' job is to admire. But in reality, our job is to create an interesting, consistent, and interactive system.

All of its components, including decorations, and the narrative, are interdependent. If the underlying process is interesting, rewarding and entertaining, then it will create a compelling intensity pattern dynamically. We don't have to predefine the pace.

Another word for "pacing" is "storytelling". We never really tell stories to players; we just put them in games. Then players tell our stories to themselves.