Character-driven Game Design Characters Conflict And Gameplay

Petri Lankoski
Staffan Björk

In GDTW2008 Proceedings, Liverpool John Moores University, UK (12th – 13th Nov).

ABSTRACT

Contemporary computer and video games utilize characters in large extent. However, game research literature says only little about how to design gameplay so that it reflects characters' personality; mainly focusing on the narration and graphical presentation of the characters. This paper presents a character-driven game design method, which uses ideas from dramatic character design to include gameplay into the design process. Based upon previous work on NPC design and a new analysis, several design choices regarding gameplay are identified. These choices are described as gameplay design patterns and related to how specific features in a character design can support gameplay. In conjunction with the patterns, the concepts of recognition, alliance, and alignment are used to introduce the method and provide examples. The paper concludes with a discussion on how the method can affect the overall gameplay in games.

Categories and Subject Descriptors: K.8.0 [Personal Computing]: General – Games.

General Terms: Design, Human Factors.

Keywords:Gameplay design, game design, player character, non-player character.

1 INTRODUCTION

Characters are utilized in many, if not in the most, popular, contemporary computer, and there is empirical evidence that characters are important for playing experience [10,26]. However, the existing game design literature discussing game characters has focuses mainly on visual design storytelling or narrational principals of character design [2, 11, 14, 17, 27, 36].

Notably, an important way to affect the game experience is through designing gameplay, which has been defined as "the structures of player interaction with the game system and with the other players in the game." [5] The relation between game characters and gameplay design, in the game design literature mentioned above, can, at most, be said to be discussed implicitly given when the method was developed. Since not all games have characters but it is difficult to imagine games without gameplay, the design of the latter can be said to be more generally applicable for game design. This, however, gives no cause to believe that there is any fundamental conflict between two types of design, although surprisingly little research on the intersection between these has been reported.

Gameplay design patterns (see Björk and Holopainen [5]) have been used in our previous work on non-player character design [20], social networks, and conflicts [21] to describe choices regarding gameplay. This paper builds on those works; therefore, gameplay design patterns are also used in this work in conjunction with other conceptual tools. Gameplay design patterns "are semiformal interdependent descriptions of commonly reoccurring parts of the design of a game that concern gameplay" [5] that support but design and analytical work. Björk and Holopainen [5] model for a pattern consists of a name, definition, general description with game examples, description of using the pattern, description of implications of using the pattern, description of relation to other patterns, and references to other work that relate to the pattern. It is important to note that design patterns are intended to function as part of a larger design language; using them does not require a specific design methodology or choice of aesthetics. Although they may, like any concepts, have associated connotations and biases their principle role is to help chart possible design spaces through providing specific point of design choices. Gameplay design patterns are marked by being in capitalized italics (and should be distinguishable from games by context). Due to space considerations, no full descriptions of patterns are given; existing patterns are referenced to and new patterns can be understood as concepts or game mechanics.

The paper is divided into five parts. First, we describe a general method for character and conflict design. Second, we consider how previous work on designing believable NPCs and social networks pertain to designing social conflicts. Third, we use all the observations so far to make general conclusions on how to design PCs and social networks and conflicts in which the PC is integral part. The concepts of recognition, alignment and alliance are introduced in order to pinpoint junctures between gameplay design and character design. Fourth, we propose a model to support character design by iterating between the different foci. Building on models and game analysis, the main part of the paper focuses upon PCs and the iterative character design model. Last, we give an example of using the presented method.

2 GENERAL ASPECTS OF CHARACTER DESIGN

Egri [9] describes a character as a sum of physiological, sociological, and psychological qualities. Based upon this he describe the bone structure of a character, a checklist of various aspects of the character that will influence behavior of the character. The bone structure is as follows:

Physiology (e.g., sex, age, height, weight, appearance, distinct, and physique);
Sociology (e.g., occupation, education, family life, friends, enemies, and hobbies);
Psychology (e.g., moral standards, goals temperament, obsessions, intelligence).

To give an example, a very short character will use different means to get his hat on a hat rack than a tall one (see Egri [9]). The bone structure is only a tool for design. The character description made using the checklist does not transfer directly to a game; features that convey the bone structure of the character to players need to be designed.

According to Egri, each character has their unique ways to react to an event and behave in a given situation. As Egri emphasizes, the dimensions of the bone structure are not disconnected, but qualities in another category can influence what kind of qualities are possible or believable in the other categories; every aspect of the character should be dealt with light of the character's other aspects [9]. Also, Egri notes that a writer might need more detailed description of a character than what gets into a play. The principals of the character design as presented by Egri is not tied to any particular form; the method has been applied to film writing [3] and to game design [17,36], but, as noted above, the question how gameplay and character design relate is not addresses in detail.

We have earlier shown that analyzing games using the theories of social network analysis (c.f., Wasserman & Faust [43]) and interaction-oriented actor-network theory (c.f., Latour [23]) can identify gameplay design patterns that expand the possible design space for games. Of the patterns identified Social Gatekeeper, Internal Conflict, Faction, Social Norm, and Social Maintenance merit note since they provide means to translate a social conflict to gameplay [21]. A conflict can arise from the incompatible goals of a player character. This Internal Conflict requires that if one goal of the character is reached another is rendered impossible. The failed goal should also have consequences in relation to gameplay, such as exclusion from a Faction (because of the action of the character violates the Social Norm of the Faction) or change in the Emotional Attachment of a NPC. Joining to Faction can require support of a NPC that function as a Social Gatekeeper.

Notably, social networks behave differently whether the PC is a pivotal character in it or not. The first type of network will break down if the player does not engage it, i.e., if Social Maintenance is required and not performed. The second type of network can be stable with or without the player activity. This implies that when the PC is pivotal character, the gameplay can be built on maintaining the network. On the other hand, social networks in which PC is not pivotal can be used to structuring gameplay, for example, by using patterns Social Norm and Moral Code. The network (as a Faction) provides punishment mechanics if the player chooses to act against the Faction making the character an Outcast or preventing access of resources of the Faction.

In Egri's model, the conflict arises from the characters that are in opposition. Their goals are incompatible, and the characters are not willing to give up their goals. The conflict in the game requires, then, that the NPCs have their Own Agenda (c.f., Lankoski and Björk [20]), that the actions of a NPC are driven by its goals. It some cases, NPCs need to have Goal-Driven Personal Development (c.f., Lankoski and Björk [20]), that they can update their goals when existing goals are completed or blocked.

A believable NPC does have Emotional Attachment (c.f., Lankoski and Björk [20]) which means that the character expresses emotional relation to specific type of game phenomenon such as affection, anger, or fear toward other characters or events. The reactions of the believable NPC are also context depended, which leads to the pattern Context Dependent Reactions (c.f., Lankoski and Björk [21]). This means that, for example, if the NPC gets angry it might not start a fight if a police is near by. Moreover, each character, as argued by Egri [9], reacts to events in its personal way. This can be described by a new pattern Trait Regulated Behavior, meaning that the character dimensions regulate and modify the behavior.

3 CHARACTER ENGAGEMENT

In what follows we discuss about a model of character engagement proposed by Murray Smith [37], which has been used as a base to design model. The model has been refined to be suitable for games [18]. While we present the engagement model, we also link it to games and gameplay using examples of various games.

3.1 Recognition

Following Murray Smith [37], we use the term recognition to refer to process of constructing character. Recognition depends on features such as: perceivable traits of the character (body, face, voice); descriptions of the character (e.g., name, title, profession); actions of the character; and reactions of other character toward the given character.

In case of NPCs, game designers can control all above-mentioned aspects, including actions, of the character. However, PCs in games differ from the NPCs in an important way: a player always (in some extent) controls a PC. Therefore, and somewhat paradoxically the PC as character is seen irrelevant to gameplay and the player (e.g., Aarseth [1]), since, as argument goes, different presentation of the character does not make one play differently. However, character is more than presentation, as Lankoski, Ekman, and Heliö [22] argue. They propose that the recognition of a PC can be guided, in a large extent, by regulating players' actions, and offer the following palette:

providing goals;
providing possible actions (including making actions impossible, and making some choices hard or easy);
defining predefined functions such as voice over narration triggered by event, attack movement as a feedback to the players choice;
cut-scenes and scripted events.

From the design point of view, the design of possible actions and predefined functions are beneficial to connect with each other; in what follows, predefined functions are mainly considered as a feedback to the player's choices. This highlights aspects of gameplay, but neglects, for example cut-scenes, but is motivated by the fact that cut-scene design and their function in games are studied elsewhere in detail (e.g., Dansky [8] and Klevjer [16]). It should be noted that from the perspective of translating character traits to gameplay, the same principles apply whether a character is created by the game designers or players (within framework offered by the game designers).

Choices and feedback, i.e., how the character executes the command is related to recognition and has an impact on the gameplay and character interpretation: In Tomb Raider: The Angel of Darkness [40] a player is required to perform just on time button push to make Lara Croft, the PC, performed necessary acrobatic maneuvers; since one typically fails quite often this makes Lara Croft seem clumsy instead of agile and physically able adventuress. On the other hand, the PC, Altaïr, in Assassin's Creed [41] performs jumps in right moment and jumps have right length, and the player's duty is to plan and make decision on the route of the character; choices in the route can lead to a jump to the death, but the most of the time character performs as a very skilled athlete. To give an another example, the strength of Hulk in Hulk [31] can be seen in possible actions and predefined functions: buildings can be damaged with bare hands and cars can used as weapons.

It is important to note that the all aspects of the PC's personality need not be fixed. It might not even be possible to fix all the traits because in games the player can make choices that have impact on the personality of the character. On the other hand, designers are always fixing some traits of the PC when they define the game system (e.g., by setting up the goals). As a general rule, the more choices that are given to the player, the fewer traits will be fixed. Some games even let players create their characters. This can be described as the new gameplay design pattern Player-Designed Character, i.e. that the character has influence of the design of a character, and is present in The Elders Scrolls IV: Oblivion [4] and to a lesser degree in Deus Ex [13] and The Witcher [7]. Conversely, the PC in these games can be consider to be more fixed if the character is evaluated only in the terms of gameplay; the player sets up the many traits of the character in character creation, and the traits set up then fixes many character traits. One should be aware that no character is completely fixed, not even literature characters such as Anna Karenina, as a character are always interpreted from narration or perceived qualities. On the other hand, as argued above, the limits and possibilities the game offers have relation to character recognition. Hence, the character cannot be totally open.

As argued above, the player character design can be seen as designing action possibilities for the character based on character and conflict sketch that are important for that conflict. Importantly, also making some action possibilities impossible is import part of design, e.g., a pacifist should have different action possibilities to reach a goal that a bully. This said, it is also important that players do not perceive these limitations ungrounded. Moreover, how the character executes the choices made by is important in gameplay. To continue the example in blatant way, the pacifist might be slow to react to commands at fight and have ineffective attacks, whereas the bully executes attack straightforwardly and the attacks are effective (a more elegant solution might be letting the pacifist to dodge fight by talking oneself out of the situation).

3.2 Alignment

The term alignment, first described in Smith's theory on film [37], refers to how the characters and player are related in terms of control and access to information that enables recognition. Besides how a character is controlled, alignment relates to what and kind of access is offered to a character's thoughts, affects, and actions. The access can be very minimal and provided from the first person point of view. An example of this is Half-Life [42]. In the game, players do have very little access to characters thoughts and feelings. Implied goals give some information about the character. Mostly the game uses the reactions toward the player character and actions of NPCs to feed information about the player character. Thief Deadly Shadows [12] extends how the information is on the character is given to players. The access is still very minimalist: The game utilizes voice over narration triggered by certain events; the role of voice over narration is, in addition to feed information to recognition, is to give hints to players how to play. Short cut-scenes are also used. Mainly the character and gameplay is structure using explicated goals.

Character goals can be stated explicitly or implicitly (see Lankoski [19]). One game that uses explicit goals is Thief Deadly Shadows [12], in which goals are listed (and players can consult the list when they please), and the goal list shows also if the goal has been reached or not. On the other hand, games, such as Ico [38], use more implicit goals: the player is not given direct access to any goals of the game. In Ico, goals are not explicated to the player (with the sole expectation of escaping from the castle, which is stated in the user manual), but enforced by the game system: for example, a failure to protect Yorda, the main NPC in the game, leads to game being over.

Part of a character design is defining its social relations. Advancing in the social groups in the game world provides one avenue for achieving alignment based upon this, and is possible in games such as Elders Scrolls IV: Oblivion [4], Canis Canem Edit (aka Bully) [33], and Fable [24]. The pattern Character Social Status encapsulates this characteristic of a game, but should not to be confused with the pattern Social Status (c.f., Björk and Holopainen [5]) which deals with players and not characters. Uses of the pattern can be related to Progress Indicators [5] but can require additional Social Maintenance, impose a stricter Social Norm and cause Internal Conflict.

In relation to progression structure, such as goal structure, a game alignment can be described as patterns, e.g., Detective Structure and Melodramatic Structure. These patterns can also be combined. The patterns can be found in many already available games, and a creator of a game can assume that the assumed player will be familiar with them and thereby have a preconception of what the experience of playing the game will be. In Detective Structure the player is controlling a single character, and the players' information restricted to the player character, and examples of this include Thief Deadly Shadows [12], Deus Ex [13], Half-Life [42], and The Elders Scrolls IV: Oblivion [4]. In Melodramatic Structure the players' are controlling more than one character and they know more than any single character (c.f., Lankoski [18] and Smith [37]) Fahrenheit [30] exemplifies Melodramatic Structure through letting players control three characters, one in time, depending on what goals have been reached.

Notably, a game can allow players to influence what kind of person PC is, which proposes challenges to alignment design. Games use patterns of Player Constructed Worlds, Planned Character Development (c.f., Björk and Holopainen [5]), and Character Defining Actions to enable player to define PC. Actions performed by characters define how their skills and abilities change, leaving much of the PC design to the player during gameplay. This functionality is present in many role-playing games, e.g., The Elders Scrolls IV: Oblivion [4], Deus Ex [13], and Fable [24] and in some action games, e.g., Canis Canem Edit (aka Bully) [33]. Another way for players to design their character is to voluntarily not use available actions and advantages which can be codified as the new gameplay design pattern, Code of Honor, restricting behavior in the game due to the character designer (whether it is a player or game designer) choosing the moral view of the PC. This is one part of Roleplaying (see Björk and Holopainen [5]). Examples of player interest in this include Fallout 2 [6] for which the Wikipedia [44] entry discusses whether it is possible to complete the game without killing any NPCs (or killing all), showing that exploring the design space of PCs is deemed one of the noteworthy features of the game. Another example in making self-imposed restrictions can be found in the use of rule-enforced advantages and disadvantages in GURPS [15] and insanity rules in Call of Cthulhu [28]). That these are tabletop roleplaying games show that the concept is not limited to computer-based games. However, when games provide game mechanics for limiting characters this can lead to the Enforced Character Behavior pattern, that the game takes control over the character to maintain the design personality. This pattern, which has a clear relation to Cut Scenes, can challenge or support Roleplaying depending on how consistently the character design is transferred to game.

An important part of designing aligment is to structure gameplay and different parts of narration to support each other. In this designers need to consider what aspects of the character are fixed in gameplay and narration, and trying to minimize the potential conflicts. This means that if designers want to give players construct some aspects of the character in gameplay, these aspects should not be later fixed in narration, as there is a potential conflict there (see above example of a pacifist character and possible conflict between traits proposed by different quest, i.e., goal structures in The Elder Scrolls IV: Oblivion [4]).

3.3 Allegiance

Allegiance is about how the players evaluate a character; to be allied with a character, the players typically need to perceive the PC having desirable or preferable positive traits or qualities (such as beautiful, skilled, similar moral or ethical values). The positive traits are relational to the standards the game set via the other characters. [18, 37] Notable techniques that can be used to prompt positive evaluation of the character are as follows:

physical beauty (beauty relates to symmetry, v-shaped body of male, hourglass body of female) [32];
expressions of affection or fear [37];
moral evaluation in relation to standards set in game world (i.e., character needs to have morally better than other characters) [37];
via gameplay (we return to this below).

For sympathy the player does not need to evaluate the character positively in every aspects, but find something positive. Allegiance is not only relevant in relation to a PC, but some games might require strong allegiance with a NPC, not the PC. An example of such a game is Ico [38], in which the player needs to find Yorda, NPCs that needs to be protected, sympatric in order to engage the game.
In games, design of controls, as the perception if controls have two-folded role: Hard controls can make the game unplayable. It is easier to evaluate easily controllable and, thus, able character positively (see above the comparison of Tomb Raider: The Angel of Darkness [40] and Assassin's Creed [41]). Related patter, Character Defining Actions, makes possible to shape PC within gameplay: skills used within gameplay will increase making using them later on more favorable. The games using Character Defining Actions pattern leaves much of the PC design to the player by offering a framework within which the player needs make design choice in order to be able to play. This functionality is present in most role-playing games (e.g., Deus Ex [13] and Fable [24]). The pattern can be used to strengthen allegiance as it enables players to choose a mode of play they prefer.

Another relating strategy to create allegiance is exemplified in The Witcher [7]. In this game the player takes on the role of Geralt of Rivia, who makes a living as a witch hunter. The PC has amnesia in the beginning of the game. This design has been used more or less explicitly in many other games; MobyGames lists 49 games as being amnesia-themed, i.e., having it as the primary game plot. A similar plot device, simply not explaining the back story of the PC (the pattern has also been used in, for example, The Elder Scrolls IV: Oblivion [4]. Having a PC with amnesia fixes the personality and backstory less. This abstraction of the PC is, again, a strategy for making players more easily sympathize with the PC. The design choice to have a PC without a backstory, either by amnesia or simply omission, can be described as the gameplay pattern Tabula Rasa: the PC's personality is formed by the player during gameplay, so that the character can be differ each time the game is played—although the thematic aspects and the game system set limitations to how different the personalities can be.

Successful allegiance makes players feel that what they are doing in the game is the right course of action since they buy into the goals of the PCs. But on could ask, if the behavior of the character is non-problematic to the player, does the character become an unnecessary construct for understanding the gameplay and the story? Interesting choices can still be available but are then either system related or related to moral questions that can make players notice or question aspects of themselves. Typically sympathy with the player character requires that the player evaluate the character having positive or desirable qualities in relation to standards set by other character.

Philip Zimbardo describes an interesting effect, "blind obedience of authority" based on psychological experiments, of rule systems that relates to allegiance: "[Rules] work by externalizing regulations, by establishing what is necessary, acceptable, and rewarded and what is unacceptable and therefore punished. Over time, rules come to have an arbitrary life of their own and the force of legal authority even when they are no longer relevant, are vague, or change with the whims of the enforcers." [45]. Deciding on following the rules may create a tension between what a player's private beliefs are and his or her public behavior. The typical reaction to this conflict, called cognitive dissonance, is to justify the actions by rationalizing them. [45] Although Zimbardo's experiments are not games, the mechanics game systems work. The Ultima VI [29] makes use of this kind of approach to modify allegiance setup; the game allies the player in the beginning with the PC, but later on prompts antipathy to the PC by bringing attention to questions of racial tolerance. In the game the player is given the quest to end a threat from the race of gargoyles that initially are understood as strange creatures taking hostile actions against the PC and the PC's home for unknown motives. As gameplay progresses one learns about the gargoyle culture, and to understand even their language which is presented as another language in the game interface, only to realize that from their perspective they have just cause for their behavior and that the PC's actions have been prophesized as those by a "False Prophet" that will destroy their entire civilization.

Similar structure is used in Silent Hill 3 [39]. In the beginning, the game tries to ensure that the player sympathizes with Heather, the player character. Near the end, one NPC (in a cut-scene) tells Heather that the monsters she has been killing are real people. This is used to create antipathy with PC and players choices in order to amplify horror atmosphere. Use of cut-scene in Silent Hill 3 can be seen as an example Enforced Character Behavior pattern, in which a single piece of input from the player can set in motion a sequence of PC actions where the player has time to reconsider what to do but cannot interrupt the action (another example of the pattern Enforced Character Behavior would be jumps of Altaïr in Assassin's Creed [41]).

Summarizing, for players to have emotional attachment their PCs they need to have to make meaningful choices; here implying that the choice has consequences which have an impact to gameplay or the moral evaluation of the action. Specifically, moral dilemmas working only the representational level are not usually effective dilemmas, as the player does not need to live with the consequences of the choice. For players to keep the interest in the choices they need to not know when the effect of their choices will be revealed but they must have the perception that the will be revealed. In another phrasing, if the choices do not affect the character or game stats they are not interesting but the effects cannot be immediately revealed.

4 ITERATIVE GAMEPLAY DESIGN

With the explorations of character and gameplay design finished, the results necessary exist to propose an iterative character-driven process that highlights aspects of social conflict. The proposed method is based upon Egri's original proposal [9] and changed and expanded to be applicable to games. A design needs to have focus and Egri suggests a premise, a proposition that the work tries to prove. In Romeo and Julia, according to Egri, the premise is "great love defies even death" [9].

Given a focus, the next part of design process is orchestration, which is about selecting (and creating) well-defined characters in opposition. The seed of a conflict is in the qualities of selected characters [9]. The bone structure approach to creating character provides a basis for this given that the traits are selected to support the premise; in one sense this means focusing more at this stage on aspects of the characters' social network than their believability. The theme naturally influences the choices if already decided; otherwise the choice of character traits can prompt certain themes. The choice of trying to instantiate certain gameplay design patterns, for example, Player-Designed Character, may be necessary already at this stage since it can restrict the design choices available regarding character traits. Likewise, initial ideas for how to provide alignment with the player and character needs to be considered here. Overarching patterns such as Detective Structure and Melodramatic Structure can be identified as feasible candidates here since they provide frameworks for how conflicts play out, and the premise can provide enough information to support design decisions regarding this.

Given a first draft of the characters that can drive the premise, the next step is to operationalize the ambitions or goals of the designed characters. What players will be trying to achieve in the game and what means they can use in this? This requires designers to identify possible gameplay actions the PC supports and what task should be easy or hard to perform. In addition to listing possible action and impossible actions, designer can analyzing the characters through gameplay design patterns as shown in the Romeo and Juliet example. The identification of patterns such as Internal Conflict, Social Maintenance, Code of Honor, and Character Social Status provide states that want to change or maintain during gameplay and this can start as initial suggestions for goals and goal structures in the game. The suitable actions to try and achieve these goals form a basis for core gameplay activity and provide additional means of creating alignment with player and character. These goals also give first ideas to how the player can have a feeling of alliance with the character. This step may focus upon PCs, but identifying their range of actions and related patterns typically led to solidifying aspects of NPCs as well.

Given the initial NPC traits and intended player goals, points of conflict can be identified through analyzing how NPCs will intentionally or unintentionally cause conflict by resisting these goals. From a character narration point of view this requires ensuring that the NPCs social behavior is consistent and believable, which may require focus upon the patterns of Own Agenda, Goal-Driven Personal Development, Emotional Attachment, and Context Dependent Reactions. Sketching the responses of NPCs to PCs' actions and vice versa provide a framework to create a cause and effect mapping. By creating subsections in this mapping both a general narration structure and a gameplay partitioning (e.g., through Levels) can be created. This mapping can also inform about the feasibility or necessity of patterns such as Emotional Attachment, Tabula Rasa or Character Defining Decisions as well show where and if it is interesting to consider manipulating the player based upon their alliance to the character for purposes of narration and emotional impact.

After these steps are completed, a very basic structure of the gameplay and character design exists. To continue making the design more detailed, iterating can be started by beginning to add more detail to characters taking into account the additional information about what is required to make them socially believable. The added characterization can provide additional or refined goals for the player which in turn requires addition thought about how the actions of PCs and NPCs trigger other actions. Iterating in this fashion can continued until the wanted level of detail and complexity is achieved. New iteration can also take input from play testing with various types of play testers (c.f., Rouse [34] and Schuessler and Jackson [35]).

5 CASE: LIES AND SEDUCTIONS

Lies and Seductions is a game designed using the methodology presented in this paper, and it used to illustrate the method. The game is inspired by a classical story Les Liaisons dangereuses (Dangerous Liaisons) by Pierre Choderlos de Laclos (first published in 1782). The designer took the main characters and their relations as a starting point. In the story Marquise de Merteuil and Vicomte de Valmont make a wager on if de Valmont can seduce beautiful and chaste Madame de Tourvel. In the following first the character and conflict design is described followed by how the gameplay design choices were influenced by these.

5.1 Character and Conflict Design

Character and conflict design of the game went through many iterations, which we are not discussing in detail—only important choices are depicted. Designed team chose to inverse the roles in the conflict. The player character is Abby (counterpart is de Valmont) who makes a wager with her friend Becca (counterpart is Marquise de Merteuil) to seduce Chris (counterpart is Madame de Tourvel) the lead of a Christian rock back, publicly promised to stay virgin until marriage. Abby, Becca, and Chris and are on holyday cruise. Chris is traveling with Emma and Ed, the bass guitarist in Chris's band.

Emma and Ed were introduced as the designers wanted to complicate the conflict, and make it possible to give players more choices. Seducing Chris should be impossible without some help from his fiends, but getting Ed or Emma to help should not be straightforward either. Designers wanted to make Abby blackmail Ed or Emma. To help forming positive allegiance with Abby, Lord James was added to design. His main role is to lower moral standards of the game world.

The next step was to deepen the characters based on Egri's dimensions to fill these functions (Egri's dimensions were used in the design even before the following description, for the sake of simplicity this is not presented). For Emma and Ed, the designers needed sociological and psychological traits enabling blackmailing. They both like Chris and want to be in good terms with him, so blackmailing is needed both to motivate their behavior and to give sub goals to the players. For Ed the design solution was to make him a gambler with alcoholic tendencies. Just before the cruise he has lost considerable sum of money in poker tournament, and is broke although he will not refuse to play if asked. The overall design choice that every character was meant to be seducible by Abby required Emma to be a lesbian (or bisexual). This was decided to be key to blackmailing Chris's sister Emma; as Chris is religious and not approves premarital sex, it was obvious extension that Chris thinks that sex between the same sexes is unnatural. Emma, again, knows that and is afraid what happens with their good relationship if Chris finds out that she is lesbian. This serves as a hook for blackmailing Emma.

To give depth to Ed, he was decided to be the songwriter of the band. Also, to make him even less stereotypical, the designers made him doctor of philosophy. As Lord James needs to set low moral standards, he needs some negative psychological traits: he is pushy, chauvinist, womanizer who likes to boast about his exploits. Chris, naturally, is a devoted Christian with high moral standards. To contrast this he was made to be charismatic rock star. Figure 1 shows concept art for the characters. The concept art was based on character descriptions.

Figure 1: Character concept art: Emma, Chris, Abby, Becca, Lord James, and Ed (from left to right).

Becca was only minimally designed, as it was not possible add her to the game due the lack of animator. She appears only in dialogue, intro, and end cut-scenes.
Abby is charismatic, social, extrovert single who is good with seducing men, but bad to keep up relationships. She is also bisexual. She likes clubbing and partying, and time to time, she enjoys playing poker with her girl friends. She an editor in fashion magazine, in which Becca is managing editor. To provide grounds for positive allegiance despite the exploitive bet, she was made beautiful and, as noted above, Lord James was introduced to lower moral standards in the game.

5.2 Gameplay Design

Many aspects of Abby that relate to gameplay, is fixed in character design presented above.
The goal of Abby is to seduce Chris. As this is not possible without help of Emma or Ed, additional goal is, either, seduce Emma and blackmail her, or steal Ed's money in poker. This also means that the game utilizes Social Gatekeeper pattern.

Possible and impossible actions were designed by looking what actions should and need to be possible. Based on character and conflict description summarized above, following list of action was chosen:

having conversations and flirting
proposing sex
blackmailing
dancing
playing poker
giving gifts
drinking (beer and drinks)
eating
walking
eavesdropping
sitting

Predefined functions of Abby were designed by an animator and a dialogue writer based on character descriptions and possible actions. Notably, all the choices made by animators and dialogue writer also define details of Abby. Actions possible by the other character was defined based on the possible actions of Abby, as they need to interact with Abby in the game. Cut-scenes are used as setting up the conflict and ending the game. Abby having sex with a NPC is presented in cut-scene as the designer did not want to add a gameplay mode for that. In addition, the blackmailing scene is presented as a cut-scene. This decision was made rather late. The change was needed because if Chris appears to witness the blackmail it would ruin Abby's chance to win the bet. Thus, to prevent failure just by bad luck, after the player chooses to blackmail Emma or Ed, a cut-scene is started.

The design was iterated using gameplay design pattern presented by us earlier [20,21]. This iteration introduced the idea of getting information about NPCs referencing by Eavesdropping. Moreover, some ideas that where implicitly in the character and conflict design were explicated as gameplay design. To give some examples: Pattern Information Passing made designers think that actions of Abby should not be directly available to all NPCs. Rather, they should transfer their opinions and knowledge to each other when they talk. The idea of Emma and Ed working as Social Gatekeepers was explicated it this phase. As the above-presented goal structure (blackmail Emma or Ed) can lead to very linear gameplay, patterns Actions Have Social Consequences, Emotional Attachment, and Context Dependent Reactions were used to counteract that. Chris's (and other NPCs) Emotional Attachment toward Abby changes based on Abby's actions, i.e. Actions Have Social Consequences. In addition there are designed to change based on characters Emotional Attachment toward Abby. Context Depended Reactions introduces variation, for example, can be used to create variation on how one can seduce Chris after getting Emma or Ed to help, as how the blackmail goal is reached has influence the game flow. Designed to be rather short game evolving around goal to seduce Chris, natural choice for alignment was the Detective Structure, as the Melodramatic Structure would easily give too much information about NPCs to the player.

6 CONCLUDING REMARKS

We do not suggest that the method presented above is the only method that should be used for game design. The designers should choose their method based on the design goals of the game. Our method is meant to highlight the potential that relations between PC–NPC and NPC–NPC have for designing gameplay. NPCs and their social networks is a tool that can be used to provide the sense of the game world as emotionally believable, and therefore to provide a basis for emotional attachment to the setting. Moreover, social networks and character-driven social conflict can provide alternatives for violent combat-driven gameplay.

Character design and logical connection between gameplay and representations are needed for certain kinds of effects, and the suggested model is intended to make this easier to design. However, for other effects or interpretations, one might need to ignore what is presented in this paper. For example, dream-likeness and surrealism requires inconsistencies in characters or in sequences of events [18]; intentionally breaking the design rules presented here may be needed so such effect.

Naturally, the method described is a simplification of how design processes really work. Creating a game consists of work in many other design fields, often simultaneously, which can affect each other and cause iterations to be restarted in various ways (see Björk and Lundgren [25] for a model of such multidisciplinary design work). However, we argue that having models for design processes help in initial planning and support on identifying problem sources and how to update plans for ongoing processes.

Concluding, the method described here offers game designers a process for creating games where the focus is on having a strong link between gameplay the character design. It thus provides an alternative to existing methods and opens up for novel game designs. The concepts used, primarily alliance and alignment, provide a slightly finer level of detail when considering designs and the gameplay design patterns even more so. Together, these provide designers with tools are several different levels of granularity that can be used individually or together to support working on new games.

7 ACKNOWLEDGEMENTS

Lies and Seduction concept art courtesy of Niklas Gustafsson.

8 REFERENCES

Aarseth, E. 2004. Genre Trouble: Narrativism and the Art of Simulation. In: Noah Wardrip-Fruin, N. Harrigan, P (eds.) First Person: New Media as Story, Performance, and Game. The MIT Press, Cambridge, 45–55.
Bateman, C. (ed.) 2007. Game Writing Narrative Skills for Videogames. Charles River Media, Boston.
Berman, R. A. 1997. Fade In: The Screenwriting Process. 2nd edn. Michael Wiese Productions, Studio City.
Bethesda Softworks LLC 2006. The Elder Scrolls IV: Oblivion.
Björk, S., and Holopainen, J. 2005. Patterns in Game Design. Charles River Media, Hingham.
Black Isle Studios 1998. Fallout 2.
CD Projekt RED STUDIO 2007. The Witcher.
Dansky, R. 2007. Cut scenes and scripted events. In Bateman, C. (ed.) Game Writing Narrative Skills for Videogames. Charles River Media, Boston, 127–147.
Egri, L. 1960. The Art of Dramatic Writing. Simon & Schuster, New York.
Ermi, L., and Mäyrä, F. 2005. Fundamental Components of the Gameplay Experience: Analysing immersion. In Castell, S., Jenson, J. (eds.): Changing Views: Worlds in Play – Selected Papers of the 2005 Digital Games Research Association's Second International Conference. Simon Frasier University, Vancouver, 15–27.
Gard, T. 2000. Building Character. Gamasutra. URL=http://www.gamasutra.com/features/20000720/gard_01.htm.
Ion Storm 2005. Thief Deadly Shadows.
Ion Storm 2002. Deus Ex.
Isbister, K. 2006. Better Game Characters by Design. Morgan Kaufmann Publishers, Amsterdam.
Jackson, S., Punch, S., Pulver, D. 2008. GURPS Basic Set: Characters. 4th ed. Steve Jackson Games, Austin.
Klevjer, R. 2002. In Defense of Cutscenes. In Mäyrä, F (ed.) CGDC Conference Proceedings. Tampere University Press, Tampere, 191–202.
Krawczyk, M., and Novak, J. 2006. Game Story & Character Development. Thompson Delmar Learning, Clifton Park.
Lankoski, P.: Empathic Engagement with a Player Character in Computer Games. Submitted to review.
Lankoski, P. 2007. Goals, Affects, and Empathy in Games. In Philosophy of Computer Games Conference (Reggio Emilia, Italy Jan. 25–27, 2007).
Lankoski, P., and Björk, S. 2007. Gameplay Design Patterns for Believable Non-Player Characters. In Proceedings of Situated Play Conference (Tokyo, Japan, Sep. 24–28). Digra Japan, Tokyo, 416–423 (2007)
Lankoski, P., and Björk, S. 2007. Gameplay Design Patterns for Social Networks and Conflicts. In Proceedings of GDTW2007 (Liverpool, UK, Nov. 14–15, 2007). John Moores University, Liverpool, 76–85.
Lankoski, P., Ekman, I., and Heliö, S. 2007. "A Metod for Designing Multifaceted Player Character in Computer Games", 装饰 (Art & Design), 72–75.
Latour, B. 2005. Reassembling the Social an Introduction to Actor-Network-Theory. Oxford University Press, Oxford.
Lionhead Studios 2004. Fable.
Lundgren, S., & Björk, S. 2005. Forces, Clashes and Remnants: A Model for Event-Driven Iterative Multidisciplinary Game Design. In Proceedings of Advances in Computer Entertainment Technology (Valencia, Spain, Jun 15–17, 2005), ACM Press, New York, 326–329. DOI= http://doi.acm.org/10.1145/1178477.1178538.
Mallon, B. 2007. "Towards a Taxonomy of Perceived Agency in Narrative Game-Play", Computers in Entertainment, 5:1,1–15. URL=http://doi.acm.org/10.1145/1324198.1324202.
Meretzky, S. 2001. Building Character: An Analysis of Character Creation. Gamasutra. URL= http://www.gamasutra.com/resource_guide/20011119/meretzky_01.htm.
Monroe, B., and Petersen, S. 2004. Quick Start Cthulhu. Chaosium Inc., Hayward. URL=http://www.chaosium.com/forms/coc_quick_start_bw.pdf.
ORIGIN Systems, I. 1990. Ultima VI: The False Prophet.
Quantic Dream 2003. Fahrenheit.
Radical Entertainment Inc. 2003. Hulk.
Reber, R., Schwarz, N., Winkielman, P. 2004. "Processing Fluency and Aesthetic Pleasure: Is Beauty in the Perceiver's Processing Experience?", Personality and Social Psychology Review, vol 8, 364–382. DOI=10.1207/s15327957pspr0804_3.
Rockstar Vancouver 2006. Canis Canem Edit (Aka Bully).
Rouse, R. 2005. Game Design: Theory and Practice. 2nd edn. Wordware, Plano.
Schuessler, N., & Jackson, S. 1981. Game Design – Volume 1: Theory and Practice. SJ Games, Austin.
Sheldon, L. 2004. Character Development and Storytelling for Games. Thomson Course Technology, Boston.
Smith, M. 1995. Engaging characters: Fiction, Emotion, and the Cinema. Oxford University Press, New York.
Sony Computer Entertainment 2002. Ico.
Team Silent 2003. Silent Hill 3.
Ubisoft Entertainment 2003. Lara Croft Tomb Raider: The Angel of Darkness.
Ubisoft Montreal Studios 2007. Assassin's Creed.
Valve Software 2001. Half-Life.
Wasserman, S., & Faust, K. 1994. Social Network Analysis. Cambridge, Cambridge University Press.
Wikipedia contributors 2008. Fallout 2. URL=http://en.wikipedia.org/wiki/Fallout_2.
Zimbardo, P. 2007. The Lucifer Effect: Understanding How Good People Turn Evil. Random House Inc., New York.