Author: by Nils Pihl

Are players pushing your game in directions you never intended it to go? In this article, Gamasutra explores how games like StarCraft II and Call of Duty: Modern Warfare encourage good and bad player behaviors.

An interesting thing about games is that the player always helps design them. No matter how simple or complex the game is, there is always room for our own creative input. We add new rules, new contexts, new narratives and new measures of success, and we choose which of the original characteristics of the game we want to interact with. Games are much like books in this manner, and we will often find the most interesting things about games between the lines of the author's instructions.

When designing games, this is both a blessing and curse. How the player behaves within the context of the game has an enormous impact on how enjoyable the game will be for the player, and game designers often find themselves struggling with how to encourage the players to play in a way that will be rewarding. Managing the expectations and behaviors of the player is a daunting task, but one of tremendous importance. Games that are well developed in every sense can still fall short to an unhealthy in-game culture. The game is only as good as the players.

So how does one manage, or even anticipate, how players might behave within the game? To understand how, we will first have to gain a rudimentary understanding of behavior itself.

Behavior is the way someone acts in response to a particular situation or stimulus. It's important here that we don't confuse behavior, which is a model for describing someone's actions, with the actions themselves.

"I am going to sleep" is a good example of what an action might be, and "I will go to sleep after this TV show, even if I'm tired now" shows us what a behavior would be. Whereas an action could be described as a data point, behavior is a graph attempting to make sense of the data. If we have a good model for someone's behavior, we can extrapolate, and derive, and experiment.

Some behaviors are particularly successful at achieving things that are good for us, while other behaviors can be wasteful, detrimental, and destructive to us. This helps us rationally choose some of our behaviors, and avoid others. We brush our teeth with toothpaste to keep our teeth white, but only very few will make the leap and attempt to brush their teeth with bleach.

But we aren't that good at avoiding destructive behavior. We routinely engage in behaviors that are bad for us, even when we are very well aware of the negative effects the behavior might have. Consider things like smoking, gambling, unprotected sex, and speeding. How can we explain these irrational behaviors?

The key lies in understanding that the roles we ourselves play in determining what behaviors are prevalent in our culture are fairly limited. Behaviors and ideas seem to have a life of their own.

The famous evolutionary biologist Richard Dawkins coined the term "meme" in his book The Selfish Gene, in an attempt to better explain what generates culture. As a game enthusiast, you may remember that Huizinga thought that our desire to play was what generates culture, but that is not a complete model. Dawkins found a way to explain how the things that play generated got to be so popular -- how they could move from isolated behaviors into the realm of culture.

A meme is a chunk of behavioral code -- a behavioral gene -- that can get copied from one individual to another. Memes are the building blocks of behavior. The words and gestures we use, the phrases we choose, the way we fold our laundry, the way we get our hair cut, these are all memes, and they are ideas that can be observed, copied, and mutated. 

In The Selfish Gene, Dawkins explains to us how natural selection acts on the genes, rather than the individuals. This gene-centric view of evolution has helped shed a lot of light on some of the more peculiar aspects of biology. The theory shows us that certain genes are more successful at being reproduced than others. The more successful genes will outperform the less successful genes and, with time, we will see more of the successful genes than we will of the less successful genes.

This simple process generates organisms that are very well adapted to the environment they live in. The gene-centric view of evolution has helped explain things like diseases and cancer, and is now a more useful model than Charles Darwin's own model of evolution. The individual is a machine built by and for genes, with the sole purpose of replicating genes.

Although they obviously do not exist in any physical sense of the word, imagining that same process of natural selection on ideas helps us understand why bad ideas spread. The survival fitness of a meme is not determined by the effect it has on its host, but rather by how well it propagates to other hosts. Memes seemingly hijack our brains and make us into machines for spreading more memes. Behaviors and ideas have a viral life of their own, just like our genes do, and we are the sometimes-unfortunate hosts of this second replicator. Memes spread through observation (even involuntary observation), and they copy themselves and mutate into new, potentially viral, strains of ideas.

When designing games, we are not completely at the mercy of these memes; there are ways for us to guide the evolutionary process of memes to a place we want. Although we can't choose which particular memes will emerge from selection, we can alter the environment of selection itself. By carefully designing the environment that the memes will populate, we can make some predictions about what will emerge.

A fitness function is a model for evaluating the fitness of an entity. When programming things like genetic algorithms we are in complete control of the fitness function -- we author it ourselves -- but even when we are not the direct authors of the fitness function, we can approach it sideways and try to model how it would work in the environment we created.

In the real world, we can see things like giraffes evolving over time to fill a niche where they have an opportunity to thrive. The victory condition for a giraffe is to survive long enough to reproduce, and this will require a steady source of food, relative safety from predators and a fair chance at competing for a mate.

Their longer necks allow them access to a food source with less competition, and the population grows in response to the improved living conditions.

 Just like the possible emergence of something like giraffes can be predicted by seeing that there are untapped resources in the form of tall, lush trees, we can anticipate the emergence of certain behaviors by examining the victory conditions of the game.

The Uroboric Cycle

For every game, there are strategies (behaviors) for winning, and given enough time the behaviors we observe will drift towards strategies that are better at winning. Simply put, people will get better and better at winning the game with time. The rules that the designer of the game puts in place (the internal rules) are the first building blocks for the fitness function for behaviors, so we must take great care to make sure that the internal rules create a problem that is solved by behavior we want to observe. We need to create the game in a way that ensures that the optimal winning strategy is something that we want our players to do. When the desired behavior of a player is not aligned with the optimal winning strategy you can end up with a product that is unenjoyable for many, hard to manage, and difficult to scale.

In any game environment with multiple players, the situation gets complicated further. The way that players behave will change the game, and thereby the fitness function, and give rise to a new generation of memes in response to the changed environment.

How to best play a game with several players depends greatly on the other players, and the prevalent behaviors that we can expect from them act as a new set of external rules that will alter the fitness function even further. This iterative process of uroboric balancing will continue until one meme is successful enough to dominate the memepool.

The internal rules of the game act as an initial first-generation fitness function for player behaviors. As different strategies are tried, the internal rules will help determine which strategies are more successful.

An example of this could be how a slow-closing reticle in a tactical shooter will favor a slower, more methodical player movement meme. Player strategies in this environment will have to strike a balance between accuracy and mobility, and with time, players will intuitively play the game in an ever closer-to-optimal way.

In a multiplayer environment, however, the players will have to compete with each other for limited resources. There are only so many kills to go around, and strategies can quickly develop that are more competitive. When this happens, all players will have to respond to this new competitive environment -- an environment that was not initially designed by the game's creators (although it might well have been anticipated).

The players themselves become a part of the environment, constantly shaping the in-game culture towards better winning strategies.

StarCraft II and the Artificial Selection of Memes

The more complex the game is, the longer it will take for the uroboric balancing to reach a final equilibrium. For some games, the in-game culture keeps changing for several years, while other games reach an equilibrium in a matter of minutes.

Blizzard Entertainment takes advantage of the dynamic nature of the uroboric balancing cycle in its competitive game StarCraft II. To ensure that the game stays interesting and dynamic, Blizzard slightly alters the internal rules whenever a certain meme is getting too dominant, or if the developers want to encourage the growth of another. Whenever the internal rules are changed, Blizzard injects a great deal of momentum into the uroboric cycle, and the in-game culture (or the meta-game, in StarCraft II lingo) changes dramatically.

What's more is that StarCraft II's internal rules have been deliberately designed in a way that allows for a flourishing and diverse meta-game. Blizzard has made it so that for every situation that you find yourself in, there are several strategies for proceeding, each with its own special trade-off.

Some games have rather strict, straightforward rules that dictate how you should respond to a certain situation -- and your prowess at the game is measured in how well you carry out that specific action -- but in StarCraft II, the emphasis is not necessarily on how well you execute an action, but rather how well you make decisions. When a talented player responds to a situation in the game, he or she will make a decision based not only what he or she knows the opponent to be doing, but also on what the current in-game culture predicts is likely for the opponent to do.

StarCraft II is a complex real-time strategy game that pits two warring factions against each other in a struggle for resources and dominance. The art of the game is balancing your investments in economy with your investments in military might, allowing you to reach the game's ultimate goal: the complete destruction of the opponent's base.

Although the game is tremendously complex, the internal rules are surprisingly simple and linear. Buried in the internal victory condition of destroying the enemy base are several necessary milestones. To destroy the enemy base, you must first defeat the enemy's forces, which will require you to have built forces of your own, which will in turn require you to have augmented your base with the infrastructure to allow for troops to be built, the cost of which will have required you to gather additional resources. Failing to do any of these is a guaranteed way to lose the game.

Although destroying the enemy base is pretty straightforward, the steps along the way leave a lot of room for personal style -- and for mistakes. How much time do I have to improve my economy before I must invest in military infrastructure? What kind of military units should I produce, and how many, and how should I deploy them to defeat the enemy in combat? The answers to these questions are not coded into the internal rules, but are a part of the in-game culture. 

Gifted and curious StarCraft players quickly start arriving at closer-to-optimal build orders over time, and these new build orders will then face tough competition in the in-game culture. When the in-game culture reaches something close to a consensus on what the optimal build order is, Blizzard makes a change to the internal rules that will have the entire community (both subconsciously and through deliberate effort) reevaluate that consensus.

Blizzard Entertainment has carefully and meticulously guided the in-game culture of StarCraft through artificial selection to have as much entertainment value as possible for both players and spectators. What makes StarCraft II into such a popular and entertaining game is the in-game culture that it has spawned.

But in-game culture can have tremendously detrimental effects on a product as well. I mentioned earlier the possibility that the desired behavior and the optimal winning strategy might become unaligned, and I would now like to give you an example of this.

Modern Warfare and the Tragedy of the Commons

Call of Duty 4: Modern Warfare is a tremendously popular first person shooter -- fast-paced and nerve-wracking. Being good at it required great reflexes, fantastic hand-eye-coordination, and a fair deal of strategic thinking. Being good at Call of Duty is hard, and not everyone can be competitive on the leaderboard...At least, that was the intent. The map Crossfire in Modern Warfare illustrates beautifully how a detrimental meme can gain a foothold because of the game's internal rules.

The Crossfire map was very popular because it supported a range of different play styles, but there was a crucial design element of the map that gave rise to some of the most frustrating and prevalent behaviors in Modern Warfare. The opposing teams would start on either end of the map, separated by various buildings and other obstacles. To get to the enemy, you would have to leave the relative safety of your starting location and move into the more dangerous warzones in the middle of the map. To be a successful player on the Crossfire map, you had to strike a difficult balance between forward momentum and tactical retreats -- it was a difficult map with a lot of room for improvisation. Topping the leaderboards on the Crossfire map could prove very difficult.

But players quickly found a way to get cheap points. One of the walls that separated the two teams early in the game was low enough for a grenade to be blindly thrown to the other side. There was a good statistical likelihood that an enemy would be on the other side, and chance would determine if you got a "free" point or not. The combination of predetermined starting points and insufficient obstacles between the teams had allowed one creative player to get a stylish kill.

The very first time it happened it was undoubtedly impressive -- to have invented the technique, you needed to have a very intimate knowledge of the game and the map. The problem was that the meme was easy to copy, without requiring any particular skill at all.

More and more people started blindly throwing grenades over that wall, which decreased the chances of each individual thrower getting a point -- but increased the chances of a hapless opponent being unfairly killed early on. The "nade spamming" meme was the perfect storm of unfortunate circumstances.

The strategy was attractive to players because it was easy to use. Very little skill or exertion was needed to have a chance at a free point or two, so a lot of people gravitated towards it. The cost of trying it was also exceedingly low -- in the fast-paced environment of the Crossfire map, it would be hard to use the grenades later on, so you might as well throw a few off early in the game and hope for the best.

And in hoping for the best lay the second rub. Random interval reward schedules can be incredibly addictive. The fact that you didn't know if you would hit an enemy or not turned throwing the grenade into highly addictive gambling scenario. All you had to wager was a cheap grenade, and you could win very desirable points in exchange. It was a meme that was easy to copy because it was so easy to observe and understand the required actions, it was fairly successful at a low cost, and it was very addictive.

The more people copied the meme, the faster it spread, and after a while the meme was so prevalent that it was expected behavior. Both teams would throw grenades over the wall in the first 20 seconds of the game, and chance would determine how many players from either side would be taken out of combat before the battle even begun. Needless to say, being taken out in the first 20 seconds of the game is incredibly frustrating -- but if you chose to avoid the situation where you could get killed early, you would forfeit an opportunity to stay competitive with your teammates.

Once the meme had become popular enough, no one really stood to gain anything from it anymore. Too many people were nade spamming, and the overall fun of the game was taking a severe hit. The problem was that the meme proved very difficult to eradicate -- the cost of changing the map itself was prohibitive, and for game hosts to kick offenders proved to be a task of Sisyphean dimensions. The viral meme was spreading too fast to be contained. The internal rules of the game made it fantastically easy for the meme to spread; KillCams would replay the moments before your avatar's death, and the victims quickly caught on and reciprocated with more nade spamming. Teammates on either side observed the behavior and joined in.

Although the behavior had been perfectly rational, beneficial, and entertaining before the meme become widely adopted, the situation changed with scale. There was only room for so many people to nade spam, but there was no mechanism from preventing everyone from having a go at it.

In game theory, what happened to Modern Warfare is referred to as the Tragedy of the Commons, and it is a surprisingly common occurrence in all kinds of everyday situations. But the problem of containing the spread, and preventing the Tragedy of the Commons, is better explained with another game theory classic: the Prisoner's Dilemma.

Because the meme was so easily copied, players would have to reach an agreement (explicit or otherwise) to cooperate in maintaining its spread. If everyone agrees not to nade spam, the game will be more enjoyable for everyone. The problem is that once people have stopped nade spamming, the winning potential for someone trying it gets very high -- and it only takes one defector to nudge the uroboric cycle back towards the Tragedy of the Commons. Richard Dawkins, the father of meme theory, explains the phenomena of both the Tragedy of the Commons and the Prisoner's Dilemma very well in his 1987 documentary Nice Guys Finish First.

Interesting to note is that the KillCam was designed as a way of deterring people from engaging in negative behaviors like nade spamming and spawn camping. Although the KillCam introduced an element of penalizing the bad behaviors, it also created a new and very effective vector for the memes to spread. Measures like the KillCam are double-edged swords: on one hand, they punish wrongdoers, but recruit more wrongdoers on the other. At the end of the day, Modern Warfare's internal rules gave birth to an optimal winning strategy that was not in line with the desired behavior of players.

In Conclusion

Designers should take great care when creating the internal rules of their products, as the in-game culture that the game will inevitably have can have a greater impact on the product than your own direct design input. The meme can be a very powerful friend or foe, and although the meme cannot be tamed, we are ultimately the architects of the environments in which they spawn.