This section will examine narrative structure in the context of the video game. This will be done in two stages – firstly, the various aspects of the three basic linear narrative structures described by Robert McKee (the classical structure, the minimalist structure, and the anti-structure (108)) will be outlined briefly and compared to what happens in video games; the purpose of such a comparison being to determine which of these, if any, closest resembles the narrative structure of video games.
Subsequently, we will embark on a detailed exploration of three basic models of pre-designed video game narrative. These three models are the string of pearls model, where the player essentially moves from one pre-designed event to the next, with a greater amount of freedom of action between the events, the branching narrative approach, where the player is occasionally able to affect the narrative by choosing from pre-designed narrative paths, and the amusement park approach, where the player is placed in a world with many possible narrative plots to tackle. Of these three approaches, the string of pearls approach is perhaps the most common; it is also the one best understood, as it is often described in video game design books (109). Somewhat less common is the branching narrative approach, perhaps because games with branching narrative require more cutscene material and thus end up costing more, with Wing Commander IV being the extreme example, at a cost of over US$12 million (110), and a 480 page script (111) – of which, in a single game, the player would only see a certain portion. Finally, the amusement park approach, seems to be tied to the role-playing game (RPG) form, where, although a strong central plot can be present, the emphasis is on the evolution of the player character’s abilities, thus creating the need for a large world with many possibilities (112); this model’s restriction to the RPG (a restriction which, it must be noted, does not imply that RPGs are in turn restricted to this narrative model) means that games offering the player this much freedom within the context of pre-designed narrative are rare (113).
As an aside, there is an alternative explanation for the prevalence of the string of pearls model in games. This explanation is that, as this model is the one most similar to linear narrative, it is the most obvious choice in the early years of the video game, in the same way that many early films were simply recordings of theatre plays (114). At the same time, the question of whether games should or shouldn’t have some kind of linear narrative is one to which various game designers have different answers. As Richard Rouse notes, many writers in the video game industry have backgrounds in (or dreams of) writing screenplays or novels, and regard game design as a day job, necessary for financial support, and often actually dislike the interactive nature of the games they write (115). Jane Jensen, on the other hand, criticises non-linear narratives and particularly the idea of multiple endings, because these multiple endings are red herrings – the game players, according to Jensen, will want to reach the ‘right’ ending (116). Both Jensen and Rouse seem to have a very clear idea of what the correct way of designing game narrative is – what is less clear is whether there is anything more than personal taste behind their arguments. Again, it is worth pointing out that it is more or less futile to look upon games as being a single monolithic medium.
Returning to the three models for pre-designed video game narrative, it seems as though most narrative games do not follow a particular narrative model to the letter. While one approach can be generally seen to dominate within a game, elements of the other approaches will be also be visible. For example, StarCraft: Brood War (1998), follows the string of pearls model, but also features at one point an element of the branching narrative, with the player’s actions in one mission allowing him/her to choose which of two possible missions will be played next. However, regardless of which of the two missions is chosen, at the end of it there will be only one possibility for the subsequent mission, thus returning the game’s narrative to the string of pearls model.
In order to better visualise the possibilities of hybridisation, the three models can be plotted onto a graph. The purest implementations of the three models then form the points of an equilateral triangle. Within the space of this equilateral triangle, games featuring purer implementations of each model would sit closer to their model’s corner, while games with a more hybridised narrative would sit nearer to the centre of the triangle.
Once these three models for pre-designed narrative have been discussed, this section of the thesis will be concluded with a brief discussion of the fourth model for video game narrative, the building blocks model, which functions in games where the designers only create a framework for the narrative, and do not create the narrative itself; the building blocks model thus maps what might be called emergent narrative.
It must be noted that the four models mentioned above are defined by different aspects of narrative than the three structures that we discuss in the context of linear narrative. Whereas linear story structures are defined by diegetic factors, the four models of video game narrative are defined by their interactivity – in the case of the three models of pre-designed video game narrative, this means the player’s ability to influence the outcome of the pre-designed narrative, and in the case of the building blocks model, the player’s ability to create a new narrative. At any rate, the necessity of examining video game narrative from the angle of player’s influence does not reduce the need to examine game narrative from the angle of traditional linear narrative structure. Indeed, a question that needs to be examined here is whether the four models of video game narrative affect a game’s options within the context of traditional narratives – does, for example, branching narrative mean that a game will have to adopt some aspects of the minimalist story structure, characterised by its open-ended nature?
First of all, however, let us return to linear narratives, and examine the structures available there.
Linear narrative structures [back to top]
The oldest and most often used structure in narrative, whether it be a film, book, or a theatre play, is the restorative three or four-act structure. This structure divides the overall plot into three acts, describing the different stages of the so-called Hero’s Journey, during the course of which the hero of the story leaves his ordinary world, goes on a quest into a special world, and following the completion of the quest, returns to the ordinary world. The concept of the ordinary and the special worlds is a metaphor, meaning that a story does not necessarily need to involve travel – the worlds can just as easily be within the hero’s psyche (117), although even then, the exploration of these internal worlds takes place through interaction with the hero’s outside world – indeed, if a story was to take place entirely within the hero’s head (i.e., as though it was a delusion), it would probably take on the appearance of an external story. At any rate, the problem that causes the imbalance within the hero’s world is usually two-fold. There is an external problem that can only be overcome and fixed through external action. Then, there is the internal problem – the flaw within the hero that prevents him/her from fully understanding the external problem and its solution. This internal flaw prevents the hero from resolving the external problem when it is first detected, and needs to be fixed first, before the external problem can be overcome (118).
The first act introduces the story’s hero and the world that she ordinarily inhabits. It then describes the events that create an imbalance within the hero’s world, eventually forcing the hero to undertake a quest of some kind in order to restore the balance. This act ends when the hero crosses a threshold, leaving the ordinary world to enter the special world where the quest will take place (119).
The second act then describes the hero’s quest, as he encounters problems of mounting difficulty, up to the point where the hero, having realised his internal flaw, gains a new insight into the nature of the external problem. With this insight comes some success, but the external problem still remains (120). This act is invariably the longest act of the story (121), and consequently is occasionally divided into two acts. However, the resulting four-act structure is not significantly different from the three-act structure, as the two central acts fill the same exact role as the central act of the three-act structure (122).
Finally, the third act is where the story is concluded. Having corrected their internal flaw, the hero can attempt to finally resolve the external problem, and, by extension, make her way back to the ordinary world once the balance has been restored. Overcoming the external conflict, of course, is not easy, and this act usually features various setbacks. However, the hero’s newfound strength overcomes these setbacks, and in a climactic moment, the external forces are overcome. Victorious, the hero returns home to live happily (123) – or, having died in order to achieve this victory, lives forever in the memory of the witnesses, both diegetic and extra-diegetic, (that is, the audience) of her sacrifice. As such, the restorative three-act structure does not necessarily need to end happily – it needs to end in a way that satisfies the story (124).
The restorative three-act structure is the most common and most classical, but not the only structure that appears in traditional narrative media. Robert McKee identifies this kind of structure, as belonging to one of three basic structures, with the other two being the minimalist structure and the anti-structure (125).
Apart from the division into three acts (a division that can be used in all three structures) and the restorative nature of the plot, the classical structure is characterised by a plot of causally connected events, with a closed ending, a sense of linear temporal progression and consistent reality, as well as the emphasis of the external conflict over the hero’s internal conflict; furthermore, one character can usually be identified as the hero of the story, and this hero is an active one, taking action and initiating events. The minimalist structure, on the other hand, while basically similar to the classical structure, leaves the ending open, allowing the audience to determine how the story really ended; it is also characterised by its use of multiple leading characters, their passive and reactive nature, and the focus on internal rather than external conflicts. Finally, the anti-structure, as its name suggests, is characterised by its rejection of the characteristics of traditional structure – the plot is advanced by unrelated coincidences, the story time is blurred to the point where linear temporal progression becomes impossible to discern, and reality is inconsistent; an object may obey the diegetic laws of physics at one point, and disobey them for no reason later on (126).
McKee notes that these three structures are not fixed and exclusive. There is a significant degree of hybridisation occurring between them; a story might mix and match between the characteristics of one structure or another, or it might settle for the middle ground. Thus, for example, a film like Seven Samurai (1954) includes multiple protagonists working together to overcome one external problem, and therefore is a hybrid of the minimalist and classical structures (127).
In general, however, most stories best resemble the classical structure – in film, for example, classically structured films are a staple product that dominates film production worldwide (128). This tendency can also be seen within narrative video games. Bob Bates, writing about game design, describes the restorative three-act structure as the best plot structure available for use within the video game – indeed, he does not even mention the possibility of a different structure (129). Certainly this emphasis on classical structure is reasonable – for example, it hardly seems possible to have a video game with a narrative that uses a passive protagonist, because if the player’s character were passive, there would be no gameplay to speak of. The closest that a game can get to having a passive protagonist, then, is a situation where the player’s character has no active role within the cutscenes, as often happens in strategy games. Thus, in StarCraft (1997), the player’s character is frequently spoken to by the other characters involved in the plot; the manner in which he is spoken to and the player’s role during the gameplay sequences, indicate that the character is certainly an important one – however, in order to maximise the player’s ability to project him/herself into the position of the character, the designers left him/her out of the plot. In any case, the protagonists (plural, as the player switches between three characters during the game) in StarCraft are only passive within cutscenes. In the gameplay, the player is in command of an army, making all the decisions and striving to achieve the victory that will allow the continuation of the plot – a job that hardly defines a passive protagonist.
Games are also limited in terms of the anti-structure elements they can borrow. If a game uses a typically anti-structure element such as inconsistency within the game’s reality, this will generally be regarded as a serious problem – such inconsistencies may be interesting in linear narrative, but for the player, they mean a frustrating inability to tell what effect her actions will have, and why (130). Of course, inconsistency in reality is not a problem in all games – the adventure game The Secret of Monkey Island along with its three sequels, is set in a quirky world where the 17th century Caribbean is liberally sprinkled with elements from the 20th century (131). The Secret of Monkey Island is able to do this without frustration, however, because the game is specifically designed to limit the possibility of death or failure within the game to an absolute minimum (132). Another aspect of anti-structure that doesn’t seem to appear too often in video games is the idea of non-linear time, where the story’s timeline is so blurred that it is difficult or outright impossible to determine which event came first.
At the same time, however, the coincidental causality aspect of the anti-structure is not only tolerated, but indeed embraced by the video game – for example, within The Elder Scrolls III: Morrowind, as the player explores the world, the game randomly places animals in the area surrounding the player. It is thus entirely possible for the player character to complete an important quest, and then die on his way home in a random encounter. Indeed, even if the player only has to deal with characters that were placed non-randomly within the game by the game’s designers, their artificial intelligence will usually result in a slightly different encounter every time the player replays the game. Thus, depending on the type of gameplay involved, a game can utilise more or less pronounced features of the anti-structure. A game’s position in relation to the four models of video game narrative that had been mentioned earlier will also affect the game’s possibilities in regard to choosing the type of linear narrative structure.
Of course, given that all video games, regardless of which of the models described here they use, are to some degree non-linear, one might ask how the linear narrative structures from non-interactive media can be compared at all to video game narratives. The objective of this comparison, however, is not to seek out the similarities between the four models of video game narrative described here and the three linear narrative structures described by McKee. Rather, McKee’s linear narrative structures become applicable when one looks at the experience of playing a video game. This experience is ultimately a linear one, since even if there are many non-linear possibilities to choose from, the player can only choose one course of action (133); thus, when a player completes a game, she forges a linear story out of the game’s non-linear possibilities. These linear stories will either belong to one of the three linear narrative structure types identified by McKee, or will be a hybrid combining their features.
Thus, it is not the stories themselves, but the possibilities of generating stories using one structure or another differ depending on the video game narrative model that is used. We will discuss these differences shortly, when we proceed to examine each of the four models in detail. In the meantime, it is also worth noting that the idea central to the restorative three-act structure, that of the hero embarking on a quest, is particularly appropriate to video games, especially if taken literally as a travel from one geographic area to another, since, as Jenkins notes, are very often spatial narratives, involving the geographic movement from one obstacle to the next (134). Notably, it is often difficult to perceive any internal problem within the heroes of video games. Thus, Doom (1993) is divided into three acts, and as far as the structure of the external problems is concerned, the game’s structure is identical to that of a typical three-act film – the hero encounters a problem in the ordinary world, enters the special world to solve it, and eventually returns victorious. However, because Doom is a first-person game and features very little in terms of cutscenes, the player never finds out anything about the hero’s psychology – thus, the only psychological problems the hero may have are those that the player projects onto him. Indeed, the intensely subjective nature of the video game means that the player character’s deep psychological problems can often be left unelaborated upon, as the players will project their own problems (such as their unfamiliarity with the game) onto their characters’ performance.
Having examined story structures in linear narrative, we can now move onto the three models of pre-designed video game narrative structure.
The string of pearls model [back to top]
Perhaps the easiest way for a game designer to transfer linear narrative into the non-linear environment, without sacrificing the control over the outcome of the plot, is to conceal the linear story behind an illusion of interactivity – as is very much the case in the rather ubiquitous string of pearls model. In this model, there can ultimately only be one ending. The players are given a varying amount of freedom, but their ability to trigger progression within the narrative is under very rigid control of the game designers (135).
According to game designer Jane Jensen, within this model, each major plot point can be visualised as the beginning part of a pearl; the player then moves on to the wider section in the middle of the pearl. Within this section, the player has the freedom to trigger many different events in any order, thus resulting in a degree of non-linearity. However, as the player continues to proceed through the pearl, it begins to get narrower, as does the player’s range of options, until there finally is only one way for the player to proceed, or (more frequently) a non-interactive cutscene – the string connecting this pearl to the next (136).
This approach to video game narrative, then, is optimised to allow the game designers as much control as possible over the plot and, by extension, to maximise the impact of the story – the game may narrate events that have a high emotional impact, without the player wondering whether or not the event ‘should’ have happened or if it was triggered by the player’s ineptitude. That is not to say that events cannot be triggered by a player’s failure – however, such events become easy for the players to distinguish from the rest of the narrative. If, for example, the player’s character dies during the story, the players will immediately and automatically understand whether this was or wasn’t designed by the designer to occur (137).
The pseudo-linearity present within this model of video game narrative means that it is probably better suited to utilise the restorative three-act structure than the other models are. Indeed, it is perhaps unsurprising that this is the narrative model most often used for video game adaptations of stories from other media. For example, the adventure game Indiana Jones and the Last Crusade (1990), an adaptation of the 1989 film of the same name, uses this pseudo-linear model for most of its gameplay. The game is divided into a number of locations (New York, Venice, Iskanderun and multiple locations in Germany). In each location, the player is free to move around and in many places can solve puzzles in a non-linear order. However, within each location, there are several tasks that must be performed before the player can move on to the next location. At the same time, Indiana Jones and the Last Crusade demonstrates the fluidity of these three models of video game narrative. While mostly sticking with the string of pearls model, the game cleverly adopts some elements of the branching narrative in order to allow the player to customise the style of gameplay – thus, for example, the player, guiding Indiana Jones and his father as they escape from Germany, can choose to either steal a biplane or sneak onboard a zeppelin. However, both paths join up soon afterwards, leaving the player once again with just one way to proceed, and thus re-establishing the string of pearls model as the dominant model within the game. Notably, in a pure implementation of this model, there is no alternative conclusion to the storyline that can be displayed if the player fails. This is indeed the case in Oddworld: Munch’s Oddysee (2001), where the death of the player’s character has very little meaning indeed – the only consequence is the need to restart the level, and this can be done an infinite number of times.
In some ways, it is difficult to discuss this model’s narrative limitations, as they seem to be practically non-existent, at least once the innate limitations of video games in general have been taken into account – any narrative structure and narrative technique that can be used in a video game can also be used in a string of pearls game.
Thus, model’s suitability for classically structured narratives does not close it off from other structures that can be identified within linear narrative. The Secret of Monkey Island, which, as was discussed above, borrows the anti-structure aspect of inconsistent realities, also happens to follow the string of pearls approach. Indeed, virtually any aspect of minimalist structure or anti-structure, insofar as it can actually appear at all in a video game, could be found in string of pearls games. This does not mean, of course, that they will be found. As Rouse notes, game protagonists usually do not have strong personalities (138); this would suggest that games focusing on internal rather than external conflict are likely to be quite rare. Similarly, even in games that seem to utilise minimalist-style multiple protagonists, one character usually takes a more prominent role. Thus, in Oddworld: Munch’s Oddysee, the player is given control of two characters, Abe and Munch, and is usually free to switch back and forth between one and the other. Furthermore, the gameplay here is designed to require the player’s use of both characters. Ultimately, however, this partnership is more like what one might find in a classically structured story than a minimalist story with multiple protagonists working independently of each other.
The string of pearls model is also not restricted in any way, beyond the restrictions imposed by the nature of video games in general (as discussed in the earlier parts of this thesis), when it comes to subjectivity, the style of narration, and other aspects of narrative. In terms of subjectivity, for example, games following this model can range from the strict first-person subjectivity of Doom to the rather anti-subjective tongue-in-cheek attitude of The Curse of Monkey Island and the other games from this series, where the player is just a guide, and the protagonist frequently acts as though he was aware that his fate is in the hands of somebody else, occasionally going as far as to speak directly to the player. The same is the case for narration – on one end of the scale, there are games like Max Payne with its complex diegetic narration, and on the other end, there are first-person shooters like Doom, where the narration is almost exclusively mimetic. At the same time, games using this model have the potential to be less limited in diegetic narration within cutscenes than games using other models, because the narration can refer back to the events within previous cutscenes, whereas in other models, the game designers have much less knowledge of which scenes the player had seen at any given point in the game.
Nor is the string of pearls model is not limited to a particular form of game, although it is especially well suited for adventure games (139), where the emphasis is on the storyline. For example, Doom divides its narrative into three episodes, each divided into nine levels. Within a level, the player has freedom to tackle different areas of the level in any order she chooses. However, there is only one way to finish each level, so there is only a single line of progression within the game’s narrative. It is, in fact, difficult to conceive a type of narrative game that couldn’t operate within the string of pearls model, although at the same time, games already designed to use a different model couldn’t be adapted into this model without significant alterations.
To summarise, the string of pearls model is perhaps the simplest form of video game narrative, in the sense that it is the closest to the pure linearity of non-interactive narrative. This often leads to criticism, as the model can easily be used to create games where a strongly linear storyline is prioritised in the development process, leading to rather weak gameplay (140). The model’s theoretical simplicity and potential of linearity, however, should not be taken to suggest that video games following this model are somehow less interesting than others. Indeed, it is a rather ironic fact that some of the most gameplay-rich and story-poor video games operate within this extremely versatile model. Most games are divided into a series of sub-games (141), called levels or missions depending on the nature of the game. Thus, this model is a natural fit for video games, with the levels forming the pearl, rich in possibilities of interaction with the game world, while the cutscenes placed between levels form the string, providing space for narrative sequences.
Using this approach, however, it is dangerously easy, if not enough effort is put into the game’s development, to end up with a storyline that has been tacked onto the game without actually becoming an integral part of the game. In such a situation, the gameplay and narrative clash rather than compliment each other and the conflict between them can detract from the overall experience. This goes back to Poole’s point – a sports player would find it very strange indeed to have to read another chapter of a book between one game and the next (142).
This is a problem for which game designers have come up with any number of solutions. One solution is to make the cutscenes pseudo-interactive, an example being the mission briefings in Damage Incorporated (1998) where the player is able to flip back and forth between text segments, pause, rewind and fast-forward through the spoken text, but ultimately still had no impact on the actual flow of the scene (143). Other solutions, however, allow the player to directly influence the storyline, choosing which action their character should take in a cutscene, or which strand of the story they wish to pursue; the branching narrative and amusement park models can be seen as the end results of this quest for greater integration of the gameplay and narrative. We will therefore now move on to an examination of the branching narrative model.
The branching narrative model [back to top]
One interesting, albeit perhaps somewhat unpopular method to improve the connection between the game and the narrative has been to give the player a say in how the narrative will unfold. This is what happens in the branching narrative model, where the story, rather than remaining linear (as is the case in the string of pearls model), is designed to branch off into different directions. This branching can be achieved in many different ways, both in gameplay and cutscene sequences, and it need not always be immediately obvious to the player that he’s faced such a choice. A player may, for example, fail to defeat a particular opponent at a particular time, and consequently ends up having to confront the same opponent again later in the game, perhaps in tougher circumstances (144). On the other hand, the choice and its results may be immediately obvious, as is the case in Wing Commander IV, when the player, in the midst of battle, is given the option of changing sides. Of course, even if the immediate consequences are obvious, there may be other long-term consequences that for the time being remain concealed.
The result of such branches is a tree or, in more complex cases, a network of possibilities. Sometimes, the branches join up later, leaving the player with only one way to successfully complete the story, although a second way to end the game, reachable by death or extreme failure, is generally also available. Whether this second ending it is more than a “you have died” message, however, depends both on the game and the situation within the game. Thus, in the space fighter simulation Wing Commander, if the player’s fighter is destroyed (and the player does not eject in time), the game ends with a brief funeral scene for the player’s character. However, if the player fails a certain number of missions without dying in the process, her failure has a more interesting result – the player’s side in the war, the Terran Confederation, suffers a crippling defeat, with the remaining Confederation forces, including the player’s character, shown retreating for a last-ditch defence effort. On the other hand, in the afore-mentioned Indiana Jones and the Last Crusade, the player occasionally faces the possibility of defeat, but the result, regardless of which point the player has reached in the game, is always the same – a brief, one screen text summary informing the player of what happened to her character, Indiana Jones.
The multiple outcomes of the game’s story can also go beyond simple victory and failure. Several good examples of this are found in the third and fourth games within the Wing Commander series. In Wing Commander III, which tells the story of the conclusion of the Terran Confederation’s war against the alien Kilrathi Empire, there are two main outcomes – either the player achieves victory for the Confederation, or, if the player fails, the Confederation loses the war and the human race is exterminated. While the outcome of failure is always identical, the victorious ending of the game comes in three versions. During the course of the game, the player’s character, Blair, has two opportunities to be romantically involved with two other characters, Rachel and Flint. Consequently, while the grand narrative of the Confederation’s victory remains the same in all three winning game endings, the conclusion of Blair’s personal storyline is different depending on whether he got romantically involved with either Rachel or Flint, or if he is alone (either because the player chose not to pursue the romances, or because the character he chose died along the way). A more explicit (and perhaps disturbing for the player) difference exists in Wing Commander IV, where the player, once again controlling Blair, can successfully complete the game only to find that Blair, as a consequence of the player’s choices during the game, has turned bad.
The branching narrative model is currently rather unpopular with game developers, and is also occasionally criticised by game scholars. One reason for this is the development cost of such games – the alternative storylines must be presented to the player, and the additional voiceovers or cutscenes needed for this purpose raise the overall price of the game’s development (145). Such costs are especially noticeable in the case of games that use cutscenes with live actors – in Wing Commander IV, despite the game’s $12 million development cost, the graphics within the gameplay sequences remained virtually identical to Wing Commander III, indicating that most of the budget had been spent on the cutscenes. Equally telling is the fact that in the 1980s, branching narrative was a standard feature in text adventure games, where additional storylines required only additional text (146). As the text adventure game was replaced by the graphical adventure game, costs of development forced a reduction in branching narrative (147).
Another problem with branching narrative especially when it ends with multiple endings, according to game designer Jane Jensen, is that players immediately sense that only one of the narrative paths is ‘correct’, and are not interested in experiencing multiple conclusions to the game – they only want the ‘right’ ending (148). This, however, is a rather peculiar line of reasoning coming from someone working in an industry devoted to non-linearity. As Rouse notes, players are not looking for a linear experience, and they expect to fail and die in the process of playing a game (149). More notable is Poole’s criticism that, once the concept of branching narrative has been sufficiently reduced in scope to become feasible, it ends up losing much of its appeal, because it reduces the story’s possibilities of referring to events that had occurred earlier in the game – in branching narrative, the game designers cannot always be sure exactly what events the player had gone through to reach a particular point in the game (150).
Like games using the string of pearls model, games with branching narrative tend towards classically structured narrative. However, here the situation is necessarily more complicated – while there is no reason why it would be impossible to design a branching story where every single branch still results in a classic ‘hero’s journey’ story arc, this has not thus far been implemented. Such an extensive branching structure would presumably increase even further the already high costs of developing a branching narrative. Consequently, games with branching narrative sit within the nexus between classical, minimalist and anti-structures. The player might encounter a classical structure while playing along one branch, and particularly the game’s ‘winning path’; on the other hand, completing a different branch can give the player a more minimalist structure, and the failing conclusion of a game almost invariably bears some resemblance to the anti-structure in its refutation of storytelling logic. An interesting example of this exists within the action game Prince of Persia (1989). The game is a hybrid of the string of pearls and branching narrative models – in a rather linear fashion, there is only one exit from each of the game’s twelve levels, and there is no variation in the scene resulting from the victory. However, the player only has 60 minutes to complete the entire game, with failure to meet this time limit resulting in a different ending for the story. The story is thus a classically structured narrative when the player’s character successfully completes his goals, while failure results in a structure that, in its focus on futility and unrewarded effort, better resembles minimalist structure or even anti-structure.
Several more complex examples of narrative structure in games with branching narrative can be seen in Wing Commander IV. This game deals with the story of a society that has survived three decades of war, and must adapt to peace; the same conflict is reflected in Blair, the player’s character, who spent his entire adult life fighting in the war, and must also adapt. It is a classically structured story, with the player’s adversary, Tolwyn, offering one solution (further militarisation to prepare for future conflicts) to society, and the player’s allies offering the other solution (demilitarisation, harmony rather than order). The winning path concludes this story exactly as one might expect in a classically structured film – Blair stops Tolwyn and the militarisation of society, and in doing so discovers inner peace, allowing him to sort out his own future. However, as was mentioned earlier, this story allows for an alternative form of victory – Blair might stop Tolwyn, but in the process of doing so adopt Tolwyn’s ideology, ultimately defeating Tolwyn only to replace him. It is a bitter ending that has more in common with minimalist and anti-structure stories than with classically structured stories where, even in a tragic ending, the hero at least understands his failure, though at the same time, it remains classically structured to a certain degree – on this narrative path, Blair still has a coherent character arc, with this ending being its logical conclusion. More anti-structuralist are the failing endings, where (depending on the particular circumstances) Blair is either killed in combat, executed as a traitor, or dismissed from the military.
In general, then, branching narrative offers as many possibilities for classically structured narrative as it does for minimalist structures and anti-structures. However, it is difficult to think of a game with branching narrative where the winning path is anything other than classically structured. This may perhaps be the reason why, as we saw earlier on, players find it easy to distinguish a game’s primary winning path from the other paths through the game. It may indeed be a two-way connection, with game designers like Jensen keeping the main (or, in some cases, only) path through the game classically structured in order to allow the players to recognise more easily which path is the correct one. Yet, this does not mean that narrative branches are unnecessary or undesirable from a player’s point of view – as Rouse notes, branching narrative is highly rewarding for players and developers, improving the gameplay and the popularity of the game (151).
Depending on the nature of the branching, games with branching narrative can achieve an even higher level of subjectivity than usual. This is particularly true for games such as Wing Commander III and Wing Commander IV, where much of the storyline branching takes place within cutscenes; although the player’s character is visible on the screen (unlike their gameplay sequences, which are experienced from a first-person view), the fact that the player chooses Blair’s reactions in so many situations allows the player to identify further with Blair – and indeed, in Wing Commander IV, where the player’s actions have such a great impact on the outcome of Blair’s personal story, it could be argued that the player to a certain degree creates Blair. Although the player is in fact merely choosing from possibilities provided by the designers, there is still a high degree of customisation here. However, although this tendency towards greater subjectivity exists, it is not a necessity in branching narrative games. An example of a lower than usual level of subjectivity in such a game would be Indiana Jones and the Last Crusade. In this game, a third-person view, combined with a strongly-defined character that frequently comments on his situation in a manner not dissimilar to that seen in The Secret of Monkey Island (although without the implicit awareness of the player’s existence) result in a relatively low level of subjectivity – rather than actually being Indiana Jones, the player merely controls him.
Games with branching narrative have certain limitations imposed on narration, and particularly diegetic narration. These limitations can be more or less obvious depending on the mimetic techniques used to present the narrative. Returning again to Wing Commander IV, this game’s use of expensive live-action video meant that it would have been logistically impossible for each cutscene to be affected by the player’s choices at all the previous branching points in the narrative. Consequently, most scenes are almost completely independent of the player’s narrative choices – Blair might offend one of his friends in one scene, and there will be no evidence that this incident had even occurred in the subsequent scene. The game solves this by ensuring that the effect of such decisions on the player’s part is felt within the gameplay sections – each of the main characters has a morale level, and if a pilot’s morale reaches a sufficiently low point, their mission performance is markedly reduced. On the other hand, games utilising cheaper techniques to present the narrative can be much more sophisticated in that regard. In Maniac Mansion (1987), where a combination of text and cartoon-style graphics put together on the screen in real-time was used, the player was able to choose the characters that she would use to complete the game. Two characters had to be chosen (in addition to a the team’s leader, who was always the same character) out of a total of six, and thus there was a significant number of combinations. Each character had different skills, and consequently, even though the choice of characters at the start of the game was the only really significant branching point in the game, there were nonetheless many different paths through the game available to the player, with somewhat different endings each time.
Like the string of pearls model, the branching narrative model appears to be useable in virtually all types of narrative games, ranging from strategy games like StarCraft: Brood War to adventure games like Maniac Mansion and flight simulations like Wing Commander. However, the financial drawbacks of the model also mean that this model, at least for the time being, is essentially impossible to implement in a pure form. Consequently, there are few games of any type that use features of this model in any extensive manner. More often, narrative branching is limited to a few branches sprinkled throughout the game, that (as is the case in StarCraft: Brood War) have no impact on the conclusion of the story. Another interesting problem for the branching narrative model emerges when looking at game series that, over time, form a grand narrative consisting of a number of independent but diegetically linked games. In such a situation, the game ending that had only implicitly been the ‘ideal’ ending that the player should be striving for, must become explicitly official as the correct ending in order to allow the sequel to carry the story onwards without the weight of the previous game’s choices bearing down on it. In the WarCraft series of real-time strategy games, for example, WarCraft (1994) allowed the player to select one of two sides, the Orcs and the humans, in a war. The end of the game was thus dramatically different depending on which side the player had chosen. The sequel, WarCraft II (1995), also allowed the player to choose one of two sides (once again, the Orcs and the human-led alliance), but in order to do so it had to assume that the Orcs had won the war presented in the first game, and were now invading a different landmass. Finally, WarCraft III (2002), begun with the assumption that parts, but not the entirety, of both of the previous game’s campaigns had occurred. WarCraft III also went on to abandon the branching structure the previous games had offered – although it still allowed the player to play with multiple factions, the four campaigns this time were presented in a linear fashion, each continuing the previous one. As each game had included a more pronounced narrative than the previous game, the WarCraft series seems to support the argument that strong narrative is easier to implement in a more linear game.
To summarise, the branching narrative model is one that seeks to break up the linearity of the string of pearls style of narrative by allowing the player to actively influence the flow of the story and gameplay. Sometimes, the player can make the game easier or harder by choosing one path over another; at other times, the choice of path, rather than affecting difficulty, affects the story more directly, even going as far as to allow the player to conclude the game in more than one way. This added narrative complexity is an expensive proposition, with the development of a branching story costing more than it would cost to develop the same story in a linear form. The story itself is also somewhat weakened by the limits imposed on diegetic narration, and temporal continuity between a game and its sequel can also be problematic. Consequently, a game that would more fully implement the branching narrative model, offering the player choices at every significant point of the story, is not likely to be possible, and most games only use narrative branching to a small degree.
Yet, even if narrative branching can only be used in a limited capacity, it still adds another dimension to the interactions present within games. At the same time, narrative branching renders narration somewhat problematic, with the difficulty of continuous diegetic narration sometimes reducing the impact that the narrative can have on the player. The branching narrative model does not, as Rouse seems to imply, only improve a game’s storytelling capabilities. However, contrary to Jensen, this model also does not result solely in frustration and wasted time. Just like the other models of video game narrative, the branching narrative model carries with it both benefits and problems.
The branching narrative model is just one way of developing more complex narrative that offers the player more control. Another option is the amusement park model, which we will now discuss.
The amusement park model [back to top]
If the branching narrative model is about narrative complexity unfolding over time, the amusement park model is about narrative complexity unfolding spatially. Rather than having new narrative branches appear as the player proceeds through the game, this model allows the player to access different narrative strands by exploring the game world and, basically, finding the right places. Games within this model often place so much emphasis on the player’s choices that the theoretical boundary between the amusement park and the building blocks model is a rather vague one – after all, in both cases, the game’s narrative is in some way generated by the player’s choices (indeed, as Rouse argues, providing the player with some authorial abilities is the main reason for having non-linearity in games in the first place (152)). However, the narrative of the amusement park model contains a much larger element authored by the game designers, whereas the building blocks model describes games where the narrative is almost entirely generated by the player’s actions. At any rate, we will explore the precise difference between these two models in the next section of the thesis. For now, the focus will be on the amusement park model.
There seems to be very little theoretical work done on games of this type. This may be because this model seems to be linked to role-playing games – a form of game that often requires much larger amounts of time for completion than most other narrative games. Not all games that classify into the role-playing category necessarily belong to the amusement park model, but they do nonetheless fit better into this model than most other games, due to their emphasis on the development of individual characters within huge worlds offering innumerable possibilities of interaction (153). Most other narrative game forms fit this model much less comfortably. For example, it is difficult to see how adventure games, with their emphasis on a strong (often linear) narrative, could work in the amusement park model – considering the expensive nature of games with branching narrative, one might speculate that the costs of developing a game where the player has not one but many strong narratives to choose from would be very high indeed. On the other hand, strategy games such as Civilization, insofar as they have narrative, will usually better fit within the building blocks model. At the same time, the role-playing element can often be tacked onto other game forms, resulting in something more capable of utilising this model – Wing Commander: Privateer (1993), for example, is a space/flight simulation containing role-playing elements, with the player being free to explore the universe in search for opportunities to earn money and property, rather than having to play a series of missions.
Richard Rouse, who has the advantage of having actually designed such a game
(Odyssey: The Legend of Nemesis (1995)), writes that the gameplay within
this type of games is about selection – the player might at a given time have
multiple challenges to choose from, and does not ever need to complete all of
them (154). The emphasis, then, is on
freedom to explore the game space, and the freedom to choose one’s activities.
Indeed, this emphasis is to a certain degree reflected by the additional
material that one receives when purchasing the game. Where normally one might
get a manual describing the story and the opponents, The Elder Scrolls III:
Morrowind provides the player only with the most cursory description of the
game world and the gameplay – but the player also gets a poster-sized map,
offering a great deal of information on the geography of the game world.
In general, the amusement park model is not usually fully implemented, because, as Poole notes, total non-linearity can lead to non-urgency – if every subplot is entirely optional, there is no necessity to pursue any particular plot (156). For this reason, games of this type, apart from having a number of optional sub-plots, usually also have one main plot that, depending on the game, may be more or less linear. Thus, in Ultima VI (1990), the player’s alter ego is summoned into the world of Britannia to combat the gargoyles that have apparently invaded this world. The player may choose to pursue a number of different sub-plots, but all of them are in some way related to the plot – Ultima VI is a meticulously designed network of sub-plots that together form the overarching story. The world of Ultima VI also offers enough space and possibilities to de-emphasise the pursuit of the overarching story, but ultimately it is virtually a necessity to start pursuing this story, suggesting that Ultima VI is a hybrid combining features of both the string of pearls and amusement park models, with a significant amount of emphasis on the characteristics of the former – except that here, there’s not much string, and the pearls are quite large.
Meanwhile, one example of a game that would be closer to the amusement park model, is The Elder Scrolls III: Morrowind. Here, the sheer number of different sub-plots available to the player means that the pursuit of the main story is utterly unnecessary – it is possible to play this game for tens of hours without even commencing the main story. Indeed, where Ultima VI disables the save-game feature if the player ever does anything that would make the completion of the main plot impossible, The Elder Scrolls III: Morrowind merely warns the player, and then allows him/her to continue playing.
Even closer to a pure implementation of the amusement park model, and indeed hovering somewhere between this model and the building blocks rather than the string of pearls model, is Pirates! (1987). Set in the 16th and 17th century Caribbean (with the player being able to choose from several more specific time periods within these two centuries), the game’s main plot (to locate the player character’s lost family members) is so thoroughly optional that it is indeed unclear whether it can be even considered the main plot. Certainly, the other options that the player can pursue (to rise in rank by serving a particular nation, to gain wealth, and even to get married), combined with an imposed time limit (where in other games, the player’s character improves, in Pirates! his skills are gradually worn down by age and old wounds) force the player to make hard choices about what he will do in any Pirates! career.
How, then, do linear story structures work within the amusement park model? In general, it seems as though the classical structure that can work within both the string of pearls and branching narrative models at least partially fails within the amusement park model. Some of its aspects, such as the focus on external conflict, linear time and consistent reality, as well as the presence of an active (and usually single) protagonist, continue to function within amusement park games. Thus, even if the player’s character is different every time she plays Pirates! or Ultima VI, the character is still an active protagonist, and even if she leads a group of adventurers, one character can usually be identified as the main protagonist. Similarly, even though the player character The Elder Scrolls III: Morrowind changes dramatically over time, this change does not necessarily come with internal conflict, and at any rate is not likely to generate sufficient internal conflict for it to become more significant than the external conflicts prevalent within the game.
However, beyond these characteristics, the classical structure gives way to the minimalist structure and the anti-structure. Thus, even though much of the story in The Elder Scrolls III: Morrowind is causally linked, coincidence also plays a large part. As described earlier, many of the most dangerous situations that the player character might face are random encounters, with the opponent being generated by the game when the player enters a particular part of the game world. The idea of the closed ending is also questioned within the amusement park model, albeit not in the same way that occurs in linear media utilising the minimalist structure – whereas a film’s ending may be open in the sense that it can be interpreted by the audience in many ways, the ending of The Elder Scrolls III: Morrowind avoids closure by not being set – the player can simply play the game ad infinitum. Therefore, it is the player that determines when the game’s story ends. This lack of closure is not always the case, however – games in the Ultima series end when the player resolves the primary plot, and Pirates! ends when the player’s character retires. In the latter case, the player may choose to retire at any point during the game, but sooner or later it becomes inevitable, with the game becoming increasingly difficult as the player’s character weakens with age. Thus, amusement park games may, but do not need to, adhere to the concept of the closed ending. In this aspect, then, games following this model much closer resemble the emergent narratives of the building blocks model than the structured narratives of linear media.
The amusement park model also serves to strengthen the subjectivity of the gaming experience. This subjectivity may be strengthened through the use of perceptual subjectivity, like in The Elder Scrolls III: Morrowind. Perhaps more important, however, is the fact that the player so often gets to define her character, or at least determine the character’s development during the game. This control over the character’s creation and development helps the player to further identify with the game character as an extension of him/herself, even if the character is always looked upon from a third-person point of view (as is the case, for example, in Ultima VI). Diegetic narration, on the other hand, in some ways becomes much harder to implement in games of this type – as the player explores the world of the game, there is little room for such diegetic devices as voiceovers or textual descriptions of the game world. This can vary a great deal depending on the period when the game was developed – Pirates!, having been developed in 1987, still uses text descriptions a great deal, and thus it relies on diegetic narration a lot more than later games. Indeed, even modern games of this type often use text a lot more than games using the other models – in The Elder Scrolls III: Morrowind, characters greet the player character vocally when the player moves within a certain distance of them, but once the player chooses to begin a conversation, the rest of the interaction is done entirely in text, substituting diegetic narration for mimetic narration. The use of text allows the game to reduce the amount of expensive materials that need to be developed in order to create the game – as we had discussed earlier, games using this complex model of narrative can become very expensive. Beyond the use of text as a cost-cutting measure, however, textual diegetic narration is also occasionally used in a more unusual fashion – in Ultima VI, whenever the player found a book within the game world, he was able to open it and read it (although these books rarely had more than few short pages of text). While this kind of narration could be used in almost any type of game, it becomes especially useful in the context of amusement park games, where diegetic books serve as a cost-effective fashion of presenting the player with information about the game world.
To summarise, then, the amusement park model of video game narrative allows a way of complicating the narrative possibilities of a game in a manner entirely different to that of the branching narrative model, by emphasising the player’s selection of the narrative strand to pursue rather than the selection of a course of action within a particular narrative plot. This model of video game narrative is useful primarily in games that contain at least some elements of the role-playing game form. Furthermore, the stronger a game’s narrative becomes, the more aspects the game needs to borrow from the string of pearls model. It is also well worth noting that this can be a three-way mixture, with elements of all three models being used in one game – The Elder Scrolls III: Morrowind, for example, features not only selectable sub-plots but also narrative branches within these sub-plots. However, much like the branching narrative model, the amusement park model carries with it higher costs than exist in linear games. This forces games within this model to occasionally resort to the simplification of both diegetic and mimetic narration, while simultaneously compensating for this simplification by using narrative techniques such as the presentation of textual sources from within the game world. The end result, then, is narration that is no less complex than the narration in other types of games.
There is one final model of video game narrative that needs to be briefly discussed before this thesis is concluded; this is the building blocks model.
The building blocks model [back to top]
If the three models discussed above describe video games as storytelling systems, the building blocks model of video game narrative describes video games as story creation systems. Narrative video games can be divided into the fully-fledged narrative games described above, and pseudo-narrative games where the game designers, rather than implementing a narrative for the player to experience, implement a system of parts that come together to form a story in the hands of the player.
This thesis has focussed on an examination of pre-designed video game narrative, and a detailed examination of games where the player generates the story will not fit in here. However, a few words need to be said here about this model of narrative, partially to simply acknowledge its existence, and partially to define the border between the player’s narrative and the author’s narrative. This is an issue that had briefly surfaced in the previous section, as some games within the amusement park model of game narrative come quite close to allowing the player to generate their own story, blurring the boundaries between that model and the building blocks model.
What, then, is the building blocks model of narrative? This is a model that functions within games where a story exists primarily within the mind of the player and (if the player so chooses) in discussion after wards – unlike purely non-narrative games, however, some kind of narrative is implicit within the game. Game designer Sid Meier, describing his game Civilization, notes that the basic difference between the story in Civilization and the stories in more narrative games is that whereas usually the game designers make all the important decisions and leave the player to make small decisions as he plays the game, in Civilization and other games like it, it is the designers that make the small decisions, and the player that makes all the important decisions (157). This kind of game thus resembles a set of building blocks that the player plays with, putting them together in whatever fashion seems interesting.
Another example of a game using the building blocks would be The Sims (2000), where the player creates and controls a family of people, in the process creating their story (158). The common characteristic between Civilization and The Sims appears to be the fact that in both cases, the player doesn’t play from a subjective point of view, instead presiding over the game world as a kind of god.
In both cases, then, the fact that the player is not taking on the role of a character within the story, and has control over the story, makes it quite clear that it is the player’s story that is being created, rather than the designer’s story or even a common effort between them. In other cases, however, the distinction between the player’s narrative and the game designer’s narrative becomes much fuzzier – particularly so when it comes to massively multiplayer online role-playing games. This type of game is in many ways identical to the games that use the amusement park model – indeed, Ultima Online (1997) is an example of a MMORPG adaptation of the Ultima series of games that operate within the amusement park model. The difference, however, is that where there is only one player in a game like Ultima VI, there are thousands of players in Ultima Online, and the game’s story – or rather, stories – are the result primarily of the interaction between the players.
It thus becomes difficult to tell who exactly the author of the story is. Given that the game designers remain involved in the daily running of the game, are they still creating the story? Or is it a communal effort? These are questions that are well worth exploring. Indeed, there is much still to be learned about the characteristics of building blocks games in general. Such investigations, however, are beyond the scope of this thesis – for now, it must suffice to say that building blocks games go far beyond the degree of co-authorship offered to the player in ordinary games.
Footnotes [back to top]
108. Robert McKee, op. cit., pp. 44-47. [back
Copyright 2003 Jakub Majewski