Abstract
Simplistic by today’s standards, the graphical adventure genre has been overlooked in favor of the vast narratives unfolding across more recent three-dimensional virtual worlds and the complex social relationships within online environments. Yet this genre established practices in game construction that allow developers to foreground narrative experiences. Graphical adventure games made with the Script Creation Utility for Maniac Mansion (SCUMM) engine like Ron Gilbert’s The Secret of Monkey Island were among the earliest to produce a sense of a temporal narrativity across the game’s many spaces that is not inscribed explicitly at the level of code. These games are therefore key to understanding the origins of video games as a narrative medium. This essay examines the source code to the SCUMM engine in order to show both how its spatialized data structures were used to produce temporal effects similar to those found in more familiar narrative media and how Monkey Island parodies the engine’s mechanisms.
Keywords
Given the interest in early digital narrative forms like interactive fiction (IF) games and hypertext literature, surprisingly little attention has been paid to graphical adventure games. Simplistic by today’s standards, this genre has also been overlooked in favor of the vast narratives unfolding across more recent three-dimensional virtual worlds and the complex social relationships within online environments. Yet this genre established a narrative aesthetics found in contemporary titles that foreground story, including the Final Fantasy, Half-Life, and Metal Gear Solid series of games. From the perspective of players, the software used to render graphical adventure games—beginning with Sierra Online’s Mystery House in 1980 and popularized by King’s Quest in 1984—marked a sharp contrast to the command line–based genre of IF that dominated the market for digital narrative texts until Infocom’s collapse in 1986 (Figure 1). Despite being marketed in terms of popular narrative forms, IF presents readers with texts that closely resemble the spatial forms first noted in modern literature by Joseph Frank in 1945. 1 Here, players are presented with descriptions of space that disrupt their sense of time—a technique often used in literature to subvert the construction of narrative meaning. 2 While many theorists have observed correctly the spatial construction of interactive media (Jenkins, 2004; Manovich, 2001; Montfort, 2003; Murray, 1997), there have been few close studies on how video games that do produce strong narrative experiences negotiate spatial form. Graphical adventure games like Ron Gilbert’s The Secret of Monkey Island were among the earliest to produce a temporal narrative across the game’s many spaces that is not inscribed explicitly at the level of code. These games are therefore key to understanding the origins of video games as a narrative medium.
A comparison of interactive fiction and graphical adventure game interfaces. From top to bottom: Infocom’s Zork (rendered by Spatterlight), Sierra On-Line’s King’s Quest (rendered by ScummVM’s AGI plugin), and LucasArts’ The Secret of Monkey Island (rendered by ScummVM).
Graphical adventure platforms like Gilbert’s Script Creation Utility for Maniac Mansion (SCUMM) engine incorporate the same design paradigm established by IF. Built to produce gameplay experiences of immersion and exploration, they organize a text’s narrative elements at the level of code within a database of discrete spaces. As if reinterpreting Frank’s theories for use in digital media, Lev Manovich’s (2001) description of databases shares with Frank’s spatial form an opposition to sequential meaning, adding that at the level of code a virtual environment is a database inscribed as “a vacuum containing separate objects defined by rigid boundaries” (p. 225). In other words, encoded data structures, because they are comprised of discrete objects, lack the sort of temporal or causal linkages that players recognize on-screen as part of a story. The act of play therefore becomes, much as literary theorists have suggested about that of reading (Brooks, 1984; Iser, 1989 ), the interpretation of possible algorithmic relationships between objects based upon the conceptual links implied by their on-screen representations. Unlike commercial IF, many graphical adventure games, and particularly those produced for the SCUMM engine by LucasArts, were praised by reviewers and remembered by players for the narratives presented to them on-screen as much as, if not more than, any narrativized account of gameplay. In particular, Gilbert’s The Secret of Monkey Island is a comedic pirate adventure noteworthy for its parody of other games. But there is another kind of parody within The Secret of Monkey Island, evident in its self-conscious tension between form and content—between coded data structures and the game as rendered on-screen. As Espen Aarseth (1997) notes, “cybernetic signs” exhibit a “dual materiality” unique to the computer as a medium (p. 40). Digital media texts are, in other words, written using one set of signifiers—the source code—and transformed through the processes of scripting, compilation, and/or assembly into software objects that renders other sets of signifiers on-screen. Despite acknowledging this unique form of materiality, critics often assume that a text’s code and rendered content are similar. N. Katherine Hayles (2005) comments, for example, that “changes on one level of programming code must be exactly correlated with what is happening at all the other levels” (p. 47). Certainly, this correlation must be present from a functional perspective—so that software functions “correctly”—but not when software is considered as a representational medium. Indeed, the narrative experienced in The Secret of Monkey Island is possible only as a consequence of a representational gap between the signifiers of its code and those rendered on-screen. Considered within the context of this gap, many of the game’s parodic moments explore the theoretical challenges of using a spatialized, interactive medium to produce temporalized, narrative experiences.
While I do not want to reignite earlier debates about whether ludology or narratology should be the focus of video game criticism, I argue that the overlapping yet distinct histories of the SCUMM engine, considered as a platform, and The Secret of Monkey Island, considered as an implementation of that platform, demonstrate that this tension can exist within the very structure of particular digital media texts. Following the example of Nick Montfort and Ian Bogost’s 2009 platform study of the Atari VCS, I first trace the history of the SCUMM engine in order to demonstrate its origins in and movement away from the structural standards of IF. The SCUMM engine is a platform because it establishes a “particular standard or specification” that functions as a common base for development of other digital media texts but which has a history that precedes “any particular implementation” (Montfort & Bogost, 2009, p. 2). I then examine how The Secret of Monkey Island’s data files enact a representational gap, producing a distinctly temporalized narrative using a spatial form designed to produce dynamic, emergent, and potentially volatile narratives experiences through gameplay. The self-conscious parody of other games in Gilbert’s The Secret of Monkey Island are particularly valuable in demonstrating various algorithmic strategies spatialized, interactive media use to produce narrative. When considered within the context of SCUMM’s algorithms, these parodic moments comment on the game’s negotiation of space and time at the levels of code and screen.
A platform study is valuable as a method, in this regard, because it accounts for a problem Matthew Kirschenbaum refers to as “screenism.” Kirschenbaum (2008) notes that many close studies of digital media still exhibit “a medial ideology” in which “many of the plain truths about the fundamental nature of electronic writing” remain “unknown at a simple factual level, or else overlooked or their significance obscured” (p. 45). 3 I contend that this problem is not the fault of scholars, but rather unique to the copyright protections framing access to source code. Because platform studies must be technically rigorous, I contend that they are possible only with software that has been open sourced, either intentionally by their developers or through reverse engineering by hackers. 4 The SCUMM engine was open sourced by a small community of hackers in 2001 in the form of the ScummVM Project. SCUMM games themselves, however, remain closed. A platform study of SCUMM must therefore supplement a discussion of the engine as visible in ScummVM’s source code with what Kirschenbaum (2008) calls a “forensic analysis” of a game’s data files. In order to document the structure of The Secret of Monkey Island’s data files, therefore, I use some of the tools developed by ScummVM’s programmers to explore SCUMM data files, most notably Descumm and version 2 of SCUMM Revisited. 5
The History of SCUMM as a Platform
Many games, including The Secret of Monkey Island, are authored and rendered through reusable software machines. Referred to within the industry as “engines,” these machines act as platforms by establishing a shared algorithmic grammar and structure among the texts they support. From the mid-1980s to the late 1990s, most graphical adventure games were written using one of two engines: LucasArts’ SCUMM or Sierra On-Line’s Adventure Game Interpreter, later renamed the Creative Interpreter (AGI/SCI). 6 The SCUMM engine was produced in its first form in 1987 by Ron Gilbert and Aric Wilmunder, the then employees of LucasArts, for the initial Commodore64 version of the game Maniac Mansion. During its commercial lifetime, the SCUMM engine was revised 8 times and was used to produce thirteen games at LucasArts. Originally based on the spatial form established by Infocom’s IF games in the early 1980s, SCUMM games would distinguish themselves from others in the graphical adventure genre more for the narratives presented on-screen than for any aspect of their gameplay after the release of The Secret of Monkey Island in 1990.
Following the model used by Infocom for its Z-Interpreter and Sierra for its AGI engine, Gilbert and Wilmunder designed SCUMM for LucasArts as a tool to simplify game authoring. The SCUMM engine acts as an intermediary between software developers and computer systems or game consoles. Developers, in short, write games in a form of proprietary source code—a scripting language—that is read and executed directly by the SCUMM engine. LucasArts’s decision to adopt this model was made both for financial and practical reasons. Infocom’s domination of the IF market during the 1980s was not solely due to the quality of its games but to the fact that the basic functions required to execute a game did not need to be rewritten for each new game made for the Z-Interpreter (Montfort, 2003). A modular, reusable system therefore not only shortened the production period for games but also permitted LucasArts to market individual games across a variety of hardware configurations. Following Maniac Mansion’s release on the Commodore64, SCUMM was ported to the Apple II later that year, the DOS operating system in 1988, the Commodore Amiga and Atari ST computer systems in 1989, and the Nintendo Entertainment System in 1990, with a rerelease of the original Maniac Mansion data files accompanying each new port of the engine. Most games written for the SCUMM engine throughout its commercial lifetime featured a similarly wide release across a variety of hardware platforms. More importantly, the decision to implement a reusable interpreter also established the SCUMM engine as a platform in its own right because it extended its development history beyond its initial implementation for Manic Mansion.
In addition to IF’s commercial strategies, SCUMMs design also incorporates IF’s primary gameplay mechanic of exploring discrete spaces. IF games present players with static descriptions of rooms that they must read and interpret for clues on how to proceed (Figure 1). Any sense of narrative becomes fragmented and embedded within these spatial descriptions, subject to reconstruction and/or misinterpretation by players. IF’s sublimation of narrative is not unique in this regard. Critics agree that interactive media present players with immersive spaces that they narrativize through their actions. Janet Murray (1997) comments, for example, that virtual spaces allow for the “dramatic enactment of [a text’s] plot;” narratives in digital media are experienced as a “graceful choreography of navigation” that “lure[s]” players through “expressive landscapes” (p. 83). Henry Jenkins (2004) likens authors of interactive media to “architects” because they “embed” narrative elements within the spaces they design (p. 121). Designers create spaces with player experiences in mind, allowing a space to become a narrative through the players’ responses to it. As an alternate aesthetic, this spatial form does not preclude narrative; rather, it shifts responsibility onto players in the sense that they must look for and pick up on a designer’s narrative cues within a game’s spaces, resulting in a wide range of narrative experiences, depending on players’ skill. The SCUMM engine was modeled after this style of play, and therefore it similarly presents players with spaces—two dimensional, color images—that they must read for navigational cues. This spatial form of narrative is problematic, however, if players’ thoughts and actions do not match those anticipated by designers. Unable to find the correct cue, players may not be able to discover how to advance to the next space: their narrative of play ending in frustration while the narrative presented to them on-screen remains unresolved. IF’s limitation to text, however, further privileges space to the extent that its textual strategies subvert narrative form. The sparse and often unchanging descriptions of discreet spaces ensure “the time-flow of the narrative is halted” and the reader’s “attention is fixed on the interplay of relationships within the limited time-area” (Frank, 1945, p. 231). Many early graphical adventure games, including some designed for use with SCUMM, merely replace text with static images, forgoing any sense of narrative apart from players’ actions. The players’ ability to move within time and space—to experience a game space as narrative—in IF and other early game genres therefore becomes dependent upon their ability to anticipate the intended experiences embedded by designers within that space.
Games written for the SCUMM engine also shared a mode of interaction with IF that requires players to imagine causal relationships between objects encountered in the environment. As with IF and early AGI titles, players interact with SCUMM games by constructing verbal commands. Infocom’s text-based IF games implement this mode of interaction through a text parser. Natural language processing algorithms allow the software to “react to [textual] input meaningfully,” establishing natural language as the “normal framework for interaction” (Montfort, 2003, pp. vii–viii). As they would during exploration, players have to read the description of a time-limited, discrete space for narrative cues to determine the proper commands to issue to objects, combining or otherwise activating them in ways intended by designers. Early graphical adventure games like King’s Quest (1984) relied heavily on this mode of interaction. Yet as shown in Figure 1, the SCUMM engine incorporated this mechanic differently from other early graphical adventure engines in the sense that it did not require players to type commands. In place of a parser system, SCUMM games provided players with a list of verbs that, when activated by clicking on them, established a simple imperative sentence. Direct and indirect objects could be added through subsequent clicks on objects visible on the screen. Additionally, the verbs required for a successful command were chosen by default in later SCUMM games, beginning with The Secret of Monkey Island, even though players were still free to experiment with incorrect ones, often resulting in comical protest from the games’ characters. From a commercial perspective, SCUMM games were built upon the popular interactive style of IF while at the same time supporting a wider variety of interfaces and player skill levels. Gilbert and Wilmunder’s design decision made SCUMM games available to a larger audience not only by allowing games to be ported to hardware platforms that did not use keyboards but also by presenting a simplified, less intimidating interface than the text-based command prompt.
This second implication is important because it marks the first change Gilbert introduced to IF’s design paradigm in order to use SCUMM to produce texts which emphasized a singular temporal narrative over narrativized gameplay. While Montfort’s history of IF software can be read as a description of how the exploration of space can produce narrative experiences, Gilbert’s work with SCUMM poses a challenge to the ability of IF software to produce engaging stories. According to Montfort (2003), early text-based IF games incorporated puzzles “to control the revelation of the narrative;” their inclusion is what distinguishes IF from texts that are merely typed using a word processor or hypertext software because the player’s role in solving in them makes the experience an “interactive process that generates narrative” (p. 3). Both the player’s actions and the program’s responses take the form of natural language, thereby allowing the player’s commands to be incorporated into the spatial descriptions stored in a game’s data files. Players’ actions and the designer’s descriptions are presented on-screen as equivalents. In this sense, the on-screen narrative and the spatial narratives players constructed through gameplay are one and the same. Contributing to the sense of coauthorship, a player’s decisions could produce multiple endings in the form of some “state of victory or a state in which the player character is dead” (Montfort, 2003, p. 31). While certainly the IF genre did invite players to interpret their experience in terms of print narratives, SCUMM’s developers recognized that IF’s text parser system more often than not constrained the storytelling ability of designers. In a recent interview, Gilbert comments that he was glad to abandon the text parser because it transformed what was presented as a coauthored narrative into a constant effort to “second-guess the parser” (as cited in Barnes, 2010). Players, he felt, did not see themselves as contributing to the production of a narrative; rather, they were feeling around blindly for verbal algorithms marked in a game’s data files by designers as solutions to puzzles. Gilbert’s decision to abandon the parser system reflects a belief that players’ narritivization of gameplay should be separated from the narrative presented on-screen. To enact this distinction, designers would need to find ways to use the spatial form of video games in a way that allowed them to present a singular, largely uninterrupted temporal sequence without stripping players of the ability to explore and interact with the spaces presented to them.
The same graphical enhancements that allowed SCUMM to abandon the text parser would also help Gilbert to separate players’ narrativization of the game’s spaces from the on-screen narrative. The text-based nature of IF conflated the roles of player and protagonist: it was always the players who acted through the commands they entered into the parser. The use of computer graphics, on the other hand, allowed designers to implement a third-person perspective. The result is a representational gap, a shift in players’ awareness from the algorithmic mazes they interact with and to the story presented on-screen. Players still issued commands, but the resulting algorithms were hidden beneath the actions characters took on-screen. Additionally, the inclusion of graphics also provided players with a sense of time flow even if their own movement through the game’s spaces had been halted. Whereas the earliest graphical adventure games present static images to players, later SCUMM games provide a sense that time was advancing while waiting for input: fires burn, lights flash, water flows, animals scurry about, and shifty characters glance nervously from side to side. Yet SCUMM’s origin in IF’s spatial form still limits the engine’s ability to render time independent of players’ actions. Everything is contained within a discrete space: animations within a space loop endlessly and actions taken in one area seem to have no effect on others. Time does indeed appear to flow independently of players, but it goes nowhere until players solve the correct puzzles.
Nonetheless, Gilbert’s work with The Secret of Monkey Island began to position the SCUMM engine as a tool for temporal narrative experiences. For this game, Gilbert made a number of changes to the mechanics SCUMM had inherited from IF. Most notably, Gilbert removed the engine’s death state, allowing for only a single ending. 7 During the 1980s, both IF and early graphical adventure games earned a reputation for being extremely difficult. Puzzles were complex and subtle, with the result that players often died with little warning if they did not follow exactly designers’ narrative cues in the intended order, turning games into long, punitive processes of trial and error. Gilbert includes a parody of these puzzles in The Secret of Monkey Island (Figure 2). If the player directs the game’s protagonist to stand on a certain cliff, the ground suddenly collapses and the protagonist falls. A menu appears resembling the death state in AGI/SCI games, teasing players with the possibility of having to start the game from the beginning. After a short delay, the protagonist is flung back onto the cliff, explaining that he was saved by a “rubber tree,” and the game continues without penalty. Gilbert implies with this parody that while dying makes sense in an arcade-style genre where gameplay sessions are brief—or where companies want players to insert money at regular intervals—it is disruptive in a narrative context that requires players’ engagement over long periods. Mechanics that work well to increase drama or heighten conflict during gameplay as part of players’ narrativization of space can hinder their experience of the temporal narratives presented to them on-screen.
Parody of AGI/SCI engine death state screens in The Secret of Monkey Island.
Even though Gilbert’s changes to the gameplay standards established by IF arguably narrowed the range of actions available to players, they were warmly received by reviewers and were later emphasized explicitly in the presentation of SCUMM games. Commenting frequently on the game’s narrative—particularly the uniqueness of the game’s characters and the humor of its plot—the brief reviews of The Secret of Monkey Island that appeared in newspapers and personal computer publications praised the simplified interface, noting that it allowed the game to remain challenging without becoming frustrating (deCoster & Crook, 1990; Lynch, 1990; Smith, 1994; Trivette, 1991). 8 In a move both to take advantage of the positive reviews and to distinguish its games from those of its competitors, LucasArts began to include Gilbert’s criticism of IF explicitly in the instruction manuals to all SCUMM games. Beginning with Gilbert’s next SCUMM release, Monkey Island 2: LeChuck’s Revenge (1991), manuals explained to players that the removal of the text parser would allow them to “spend [their] time following the story” and not worry about “typing in synonyms until [they] stumble upon the computer’s word for a certain object.” LucasArts further emphasized narrative by characterizing gameplay as something which should contribute to but not interfere with the unfolding of a story. Under the heading of “Our Game Design Philosophy,” LucasArts announced its belief that players “prefer to solve the game’s mysteries by exploring and discovering, not dying a thousand deaths.” Despite being designed according to the spatial form popularized by IF, SCUMM became explicitly associated with temporal narrative experiences following the release of Gilbert’s The Secret of Monkey Island.
The Secrets of Monkey Island
Because the SCUMM engine was constructed around a spatialized database similar to those used in IF, Gilbert’s desire to emphasize a temporal narrative in The Secret of Monkey Island was constrained by the very tools he had earlier created for LucasArts. The SCUMM engine’s algorithms inscribe and read data files according to a hierarchal structure similar to those found in object-oriented programming. Within the database’s hierarchy, space is the primary organizational principle in the sense that those structures which define space encapsulate all other structures. This style of hierarchy is still frequently used today because most games are organized around levels or other types of discrete spaces. While critics agree that this spatial form allows players to narrativize their gameplay, it remains to be seen how designers have been able to produce more traditional narrative experiences using this alternative aesthetic. In SCUMM games, even prewritten sequences of events—commonly understood to be the primary carriers of meaning in narrative texts—are subsumed within data structures called Rooms that define portions of the narrative’s setting. Turning to 1992 CD-ROM edition of The Secret of Monkey Island, we can see in detail how Gilbert manipulates the spatial data structures of the SCUMM engine to produce a representational gap between the signifiers of code and screen that serves his goal of using the platform to emphasize a temporal narrative experience. 9 When read through the context of the game’s data files and the engine’s source code, the parody on screen of pirate tales and other games becomes a commentary on Gilbert’s manipulation of SCUMM’s spatial form.
From the perspective of players, The Secret of Monkey Island is a fairly straightforward adventure narrative, explicitly divided into four chapters, that incorporates a substantial comedic element through its dialogue and puzzles. Gameplay is divided between periods of nonlinear exploration and linear cut scenes—something not found in IF—that resemble short, single camera film or television sequences. Players issue commands to Guybrush Threepwood, a young man who dreams of becoming a pirate but has no knowledge of what being a pirate entails beyond archetypes drawn from popular films and literature. The game begins on Mêlée Island, where Guybrush is tasked with demonstrating his skill at sword fighting, thievery, and treasure hunting before the island’s Pirate Leaders will recognize him as one of their own. Gameplay involves multiple puzzle threads, one for each skill, that may overlap spatially and temporally but never algorithmically. Each thread is a series of puzzles that take the form of small tasks. Within each thread, these puzzle tasks must be completed in a specific order; yet because these puzzles are embedded throughout the game’s various spaces, players are free to move between each thread without penalty. Players must, for example, feed drugged meat to guard dogs in order to sneak into the governor’s mansion and steal a statue. The flowers that the player must mix into the meat can be found at the entrance to the maze that leads to the island’s buried treasure. Players, in short, are relatively free to move through the games spaces but can only move forward temporally under certain conditions. As players navigate the game’s spaces and complete tasks, the environment itself remains unchanged: it is always night, and everyone remains in their positions. Cut scenes, on the other hand, provide a sense of a continued progress through time. Following his completion of the three trials, for example, cut scenes show the narrative’s antagonist, the Ghost Pirate LeChuck, kidnapping Mêlée Island’s governor and Guybrush’s love interest, Elaine Marley. When LeChuck kidnaps Elaine, the island’s spaces are changed and new puzzle threads are activated that must be solved before Guybrush sets off to rescue the governor. Clearly perceived on the screen by players, this narrative becomes visible only in fragments when one examines its coded inscription in the game’s data files.
The 1992 CD-ROM version of The Secret of Monkey Island is stored in two data files written in encrypted binary code and a series of audio tracks in CD Audio format. These two data files, MONKEY.000 and MONKEY.001, are of drastically different sizes: 8955 and 4779713 bytes, respectively. This difference reflects the Version 5 convention of storing the game’s resources in a single file, MONKEY.001, and using a separate index file for navigation. 10 The index file, MONKEY.000, can be divided into six blocks, each identified by its first four bytes: RNAM, MAXS, DROO, DSCR, DSOU, DCOS, DCHR, and DOBJ. RNAM serves little function outside of debugging and is not accessed after the engine first reads MONKEY.000 during a normal play session (engines/scumm/resource.cpp, 447). 11 The values stored in MAXS are used by the engine to initialize the arrays that handle gameplay related variables (engines/scumm/resource.cpp, 470). Each of the directory blocks contains in this order following the four-byte identification header: four bytes defining the size in bytes of the block, two bytes defining the number of objects described in the block, and a two-part list of data identifying the location of resources in MONKEY.001. 12
The game’s implementation of SCUMM’s spatial form is visible in the directory blocks’ description of the resources stored in MONKEY.001. When processing the directory blocks, the engine understands the stored descriptions only in terms of locations within MONKEY.001. With the exception of DROO and DOBJ—the directories of rooms and objects, respectively—the two-part list within a directory block is comprised by a series of index numbers, followed by a series of byte offsets. 13 For DROO, the first part of this list is a series of one-byte segments in big-endian notation which are set to 0 or 1. Read in order, these segments identify whether a given room should be defined in the array of all rooms (engines/scumm/resource.cpp, 494). The presence of this first part of the list reveals that the size of the initialized array may not reflect the number of defined objects stored within it. There are, for example, 100 rooms according to the size value of DROO but only 84 segments in the list are set to 1, and only 84 are defined in MONKEY.001. Presumably, these gaps indicate rooms removed by the game’s authors during the later stages of development. For DSCR, SCOU, DCOS, and DCHR—the directories of Scripts, Sounds, Costumes, and Charsets—the index numbers refer to locations within the array used to store the list of Rooms. The lists of byte offsets serve as pointers in the sense that they indicate where in a Room a given resource can be found. A resource’s location, in other words, is defined according to space in two ways: the Room block it is stored within and the location of its first byte within that block. The script which displays the game’s introductory sequence, for example, is stored in Room 10 and has an offset value of 56852. Within MONKEY.001, this script begins 56852 bytes after the first byte of Room 10’s header. Room 10 has an offset value of 356120. Therefore, when the engine needs to load the introduction sequence, it adds the script’s offset to Room 10’s and then loads the script that begins at byte 412472 in MONKEY.001. This index system permits the engine to move directly to a specific address within MONKEY.001’s database without needing to process data unrelated to the particular segment of the game it is rendering. Addresses are therefore only meaningful relative to Rooms, establishing a data structure which defines space as the game’s primary organizational principle.
Emphasizing space rather than time, the engine lacks a description of causality beyond simply checking the value of Boolean variables. Gilbert brings this aspect of the game’s spatial form to player’s attention as part of a parody of the graphical adventure genre. Later in the game, for example, the player must direct Guybrush to knock over a cannon in order to steal a cannonball and gunpowder from Herman Toothrot’s jungle hideaway (Figure 3). Herman suddenly appears and comments jokingly that it must just be a “coincidence” that Guybrush happens to be present when he discovers that his home has been disturbed. Experienced players understand the joke to be about the adventure game trope of players accumulating odd collections of objects by adding everything they come across to their inventory. Characters typically do not react to players’ rampant thieving. Yet by joking that he could not possibly suspect that Guybrush is about to steal his property, Herman also reveals his spatial construction to players. Here, in Room 80, Herman’s dialogue is stored as Local Script 20. Local Script 20 is executed by Room 80’s entrance script; after a fixed delay, it draws Herman and initiates a dialogue tree based on the current state of objects within Room 80. Local Script 20’s dialog tree is comprised of if/else statements. Depending on whether certain variables representing object disturbance—but not themselves part of the data structures defining the objects—have been set to 0 or 1, Local Script 20 will make Herman either comment about the canon, his spyglass, or simply Guybrush’s presence. At the level of logical structures, Local Script 20 cannot interpret temporal concepts like cause and effect. From the perspective of the engine, time does not exist because it can only check the state of variables at the precise moment of its execution. The engine, acting through Local Script 20, has no way to determine how or when the cannon was knocked over despite the fact that these concepts are clearly evident to players based on what they see on-screen. By feigning ignorance of causality, Herman both plays on an adventure game trope while hinting at the game’s context within a spatial form designed to render atemporal sets of discrete objects.
Herman Toothrot’s comments regarding the SCUMM engine’s lack of causality.
Moreover, SCUMM’s use of an encapsulated hierarchy to organize its spaces extends this same lack of causality across Rooms. Within the context of object-oriented programming, hierarchies are used to decompose an abstract representation of a process into simpler operations (Booch, 1993). There are no explicit rules regarding how best to decompose an abstract system and so the decision to establish particular components as primitives of others is relatively arbitrary (Booch, 1993). Because the choice of primitives and abstractions is arbitrary, the structural logic used to organize digital media components into hierarchies can be interpreted as a theoretical assertion regarding the location and construction of meaning within a data structure. Furthermore, hierarchies established through abstraction often incorporate the practice of encapsulation: a technique of information hiding. In an encapsulated structure, the primitives used to implement an abstraction are effectively blackboxed within the code representing that system (Booch, 1993). Direct access to primitives is not permitted, and therefore encapsulated elements can be considered less meaningful within the context of a program. Lower level primitives are not meaningful themselves; rather, they contribute to the production of meaning through the relationships formed between them by higher level abstract components. The higher one component or set of components is within a digital media text’s hierarchy, the more directly it contributes to that text’s representational meaning. Within SCUMM, then, any appearance of time flow, causality, or other narrative descriptions of temporality within a Room contributes to the construction of that Room’s space. At the level of code, animations within a Room, as well as the actions players or characters may take, are meaningful only within the context of that Room.
Because SCUMM’s Rooms are encapsulated structures, moments within The Secret of Monkey Island when players experience a sense of causality or time flow across Rooms are the product of representational gaps. Within MONKEY.001, Rooms are defined as blocks of bytes that begin with the four-byte identification header, “ROOM.” The length of the block is included in the Room’s definition, so that the end of a given Room’s block is marked within only by the header of the next. As the indexing system of MONKEY.000 suggests, none of the game’s resources in MONKEY.001 are stored outside of Room blocks. Even non-diegetic resources that do not appear within the explorable space of the game, such as credit sequences or chapter title sequences, are treated at the level of code as spatial constructions and stored in Rooms. More importantly, events that appear from the perspective of players to occur across Rooms are actually separate events executed through a series of scripts isolated within discrete Room blocks. As part of the sword fighting task, for example, Guybrush must trick Mêlée Island’s shopkeeper into leading him to the Sword Master’s house. In order to do so, the player must direct Guybrush to follow the shopkeeper across several Rooms. From the perspective of the player, the two characters move together from Room to Room. However, the engine references separate scripts in each Room to render this scene. After Guybrush asks the shopkeeper to seek out the Sword Master, Local Script 11 in Room 30 sets the variable Bit[326] to 1. As players direct Guybrush to follow him, each Room entered contains a Local Script executed upon entrance that checks Bit[326] and renders the shopkeeper walking across the screen. The shopkeeper at the level of code is actually a series of shopkeepers, each drawn from separate scripts in their respective Rooms. Because the game’s platform understands all Rooms as encapsulated, the only way to produce a sense of time flow across discrete spaces is to reproduce narrative elements within multiple Rooms. Any sense of temporal continuity from the perspective of players is therefore the result of a representational gap between code and screen.
Despite the perception of a temporal narrative on the screen, there is no clear description of The Secret of Monkey Island’s narrative structure in either the index files in MONKEY.000 or the Room blocks of MONKEY.001. This absence can be made strikingly visible through a comparison of the order Rooms are inscribed in the data files and the order they are encountered during gameplay. The indexing system SCUMM uses to navigate its data files makes temporal sequence within its data files arbitrary. Table 1 is a list of Rooms in the order their data are stored in MONKEY.001. When mapped, most Rooms that are adjacent to one another from the perspective of players are shown to be distant from one another in the data file—and in some cases, even located on opposite ends (Figure 4). A new playthrough of The Secret of Monkey Island begins, for example, with the introductory sequence in Room 10, advances by script to the chapter title stored in Room 95, and advances again to Room 38 for a cut scene introducing Guybrush, finally ceding control of the interface to players after advancing once more to Room 33. Players are not introduced to the plot’s first conflicts until they proceed through Room 33, enter Room 28, and initiate a dialogue with the Pirate Leaders. Reflecting the influence of the IF design paradigm, any notion of sequence is scattered throughout MONKEY.001’s various Room blocks. Navigational cues that players use to gain a mentally map an island do not reflect the spatial relationship of Rooms within MONKEY.001. The game’s islands are formed by the engine only when deciding which Room to load after one is exited. Even if the sequence in which players encountered rooms were similar to their inscribed sequence in MONKEY.001, their inscribed sequence would eventually be violated when players explore branching paths or in the event that they return to Rooms already visited. 14 Because the SCUMM engine organizes narrative elements according to an encapsulated hierarchy that privileges space, its structural logic does not explicitly support the inscription of singular temporal sequences. Narrative exists from the perspective of the player, but not from that of the engine.
Flowchart of Rooms in chapter 1 of The Secret of Monkey Island. These Rooms represent Melee Island. Note that Rooms that cannot be entered by the player (e.g., cut scene-only rooms like the introductory sequence in Room 10 or the Insult Sword Fighting scripts in Room 88) are not marked. See Table 1 for descriptions of each Room. Note that most blocks adjacent to one another from the perspective of players are not adjacent in the data files.
A complete listing of the Rooms stored in MONKEY.001.
Hierarchies of abstraction and encapsulation may seem unique to digital media, but narrative theories used to describe older media can also be read through these same principles. Peter Brooks argues that meaning within narrative forms is shaped through a structural logic of causality. Narratives are “temporal syllogisms, concerning the connective processes of time;” meaning within narrative texts may be “delayed, partially filled in, stretched out” but its syntax is always “temporally unfolded” (1984, p. 21). Narratives are, in other words, encapsulated structures that employ time as their primary organizational principle. Events, or other discrete units of time, serve as a high-level abstraction within narrative texts, and all other elements serve as primitives in the sense that events establish meaningful relationships among them. Other theorists of narrative form like Roland Barthes (1977) and Seymour Chatman (1975) have even gone so far as to invoke explicitly hierarchical structures. Barthes, for example, uses a hierarchal diagram to illustrate his observation that readers group words and then sentences together to form an awareness of abstract relationships between characters in order to derive concepts like conflict, motive, and intention (1977). Considered as a data structure, this feature of narrative is no less true of texts wherein the events are not presented in chronological order such as those constructed in a spatial medium. We read in “anticipation of retrospection” so that meaning of a narrative text cannot be conceived until the entire sequence has been perceived: “what remains to be read” will inevitably “restructure the provisional meanings of the already read” (Brooks, 1984, p. 23). Understanding and misunderstanding narrative meaning therefore depends on how readers organize all of the events of a story into higher concepts. Because the spatial form of IF that SCUMM is modeled after broadly encourages freedom of movement and action, it only very loosely suggests a sequence of events to players, posing a problem for designers like Gilbert who wanted to present players with a strong, prewritten narrative in addition to their narrativized gameplay experiences.
In The Secret of Monkey Island, the representational gap that produces the experience of a temporal narrative distinct from player’s narrativized gameplay manipulates SCUMM’s spatial hierarchy so that time is perceived by players as the game’s primary organizational principle. Gilbert uses restrictions on player movement to create chapters with distinct beginnings and endings that are not explicitly marked in the game’s indices. Within the narrative space of The Secret of Monkey Island, there are three sets of Rooms spread throughout MONKEY.001. Considered as a temporally unfolding narrative, the game is divided into four chapters: three corresponding to each set of Rooms and a fourth that unfolds in a subset of the first. Access to each set of Rooms occurs according to a fixed temporal sequence as Rooms in each set appear to players to be isolated from Rooms in another even if they are located adjacent to one another in MONKEY.001. Each chapter begins in a Room that displays a title sequence and that contains a script which places Guybrush within an explorable Room in the corresponding set (Figure 5). Thus, each Room can be understood from the perspective of players as belonging to a chapter in the game even though there is no value for chapter stored anywhere in a Room’s definition. Advancing between chapters is controlled by Scripts stored in the Rooms that mark their beginnings and endings. The SCUMM engine is set to execute the first Script listed in a game’s index upon loading. In The Secret of Monkey Island, Script 1, in addition to defining a variety of default gameplay variables, prompts the engine to enter Room 10. Room 10’s entrance script displays the credits and then prompts the engine to enter Room 96. Room 96 displays the first chapter’s title and then prompts the engine to enter Room 38, part of the Mêlée Island set mapped in Figure 3. Players cannot venture to Rooms outside of the set representing Mêlée Island until returning to Room 33 under specific conditions, triggering a series of scripts that eventually display the next chapter title and place Guybrush in the second set of Rooms. By constraining a player’s ability to move within a platform designed to provide an experience of free exploration, Gilbert creates distinct temporal divisions that establish time as the game’s abstract organizational principle—at least from the perspective of players. The engine continues to organize the game according to the spatial conventions established by IF, yet a gap between code and screen leads players to experience a narrative organized according to time.
Example of a Room used to identify a new temporal sequence.
Gilbert’s manipulation of SCUMM’s spatial form does not merely reproduce a traditional narrative distinct from players’ narrativized gameplay. The engine’s spatialized aesthetic inevitably leaves its mark on the Gilbert’s narrative by creating moments in which time becomes spatialized. Gilbert often uses this for comical effect, allowing players to feel trapped in ridiculous scenarios while permitting them to advance out of these algorithmic loops at any time. One such example occurs when Guybrush must deal with Stan, a used ship salesman, before he can sail after LeChuck and rescue Elaine. The game’s forced sequence of rooms only permits Guybrush to purchase one of the many ships shown at Stan’s shipyard. This scene’s dialogue, however, reflects the engine’s mechanics of spatial exploration by permitting Guybrush and Stan to discuss every single ship available, as well as a looping list of accessory options that have no impact on the events that follow (Figure 6). Temporal progress in the narrative is frozen, yet players still experience a sense of time flow as the dialog advances through its coded loops. Players are free to explore the dialogue scripts stored within that Room for as long as they remain amused by Gilbert’s parody of shopping for used cars. The game’s most substantial puzzle, Insult Sword Fighting, mirrors the conversation with Stan (Figure 7). Here, the game’s time flow seems to halt as players repeat the same scenario multiple times, yet the puzzle advances despite players’ success or failure in each encounter. Presented to players as a series of dialogue trees and seemingly encountered in a variety of locations, the scripts for Insult Sword Fighting are all stored in Room 88, one that is never actually entered by players. A parody of combat in pirate films, the conflict takes the form of players selecting insults or witty comebacks. Players begin the puzzle with limited access to Room 88’s dialog tree. Each time players are exposed to a new insult, a dialogue option is opened; and similarly, responses are opened the first time their insults are parried. Regardless of whether or not players win a particular duel, the puzzle progresses each time the player attempts to solve it. Once players have accumulated what they feel to be a sufficient number of dialogue options they can travel to Mêlée Island’s Sword Master and so resume the game’s narrative sequence.
Guybrush tries to buy a boat from Stan’s previously used vessels. Example of the synchronic exploration permitted through a pause in the fixed, diachronic sequence of Rooms.
Insult Swordfighting Puzzle. Guybrush picks an insult (above) or one of the corresponding responses (below). This puzzle functions through same synchronic exploration illustrated by conversations like those in Figure 6.
When read with knowledge of the game’s code structures, several of the parodies that Gilbert creates through the engine’s spatialized dialogue sequences can be understood as a commentary on the game’s negotiation of both spatial and narrative forms. Tension between the gameplay and narrative is most visible during moments when a player’s choices are apparently ignored. At the game’s climax, for example, when Guybrush bursts into LeChuck and Elaine’s wedding to rescue the governor, players are offered a choice of dialogue options, some heroic and others comical (Figure 8). Regardless of the players’ choice, Elaine suddenly jumps out of hiding and explains that the journey he has just undertaken to save her “really wasn’t necessary.” Insofar as there is only one ending to the game’s narrative, players’ choices at any given moment are, as Elaine’s comments suggest, irrelevant. Regardless of the way players narrativize their gameplay, their choices had little if any impact on the sequence of events Gilbert wanted to present. Despite not granting players control over the game’s fixed sequences of events, Gilbert’s spatialized dialogue gives them a chance to comment on his negotiation of gameplay and narrative. Depending on the player’s choice, the rest of the scene can be read as having a tone of frustration, confusion, or comedy. Shortly thereafter, Guybrush defeats LeChuck and is reunited with Elaine. The two are shown looking out at the night sky together, discussing the events of the game. The player is given a final set of dialog options, one of which is immediately identifiable as a criticism of the game’s form, disguised as a joke (Figure 9). If the player selects the option that challenges The Secret of Monkey Island’s success as a game, Guybrush immediately attempts to cover the tension and return coherence to the narrative by commenting that he’s “not sure why [he] said that.” The events resolved in a climax that was unavoidable, Guybrush’s out of character comment allows players to close the story by reflecting on whether what they have just experienced was a rather limited adventure game or a new form of narrative. As funny as it may be on the surface, this final option hints at the game’s position as a transitional text—a game that rests comfortably between the players’ expectations of the adventure game genre established by IF and the familiar aesthetics of older narrative media. Gilbert thus highlights his experiment in design, offering his representational gap up to players for judgment.
Guybrush faces the inevitably of the game’s linear narrative structure. Players are able only to choose how Guybrush responds to the situation.
The player is offered a chance to reflect on the game just before the credits. Guybrush is confused if the player chooses the option that questions the narrative’s context within the medium of video games.
Conclusion
The consensus among critics and other digital media scholars that games emphasize spatial experiences and older media like literature and film emphasize temporal experiences suggests that technological limitations also constrain our perception of space and time within media. Yet, Gilbert’s ability to manipulate SCUMM’s spatial form to produce a much stronger temporal narrative in The Secret of Monkey Island than present in previous texts suggests that not all technological limitations on aesthetics are fixed. Tools designed to produce one form of experience can be repurposed for other forms. Techniques like Gilbert’s have been used to produce strong narratives in other genres. The Half-Life or Deus Ex series of games, for example, invoke a strong sense of narrative in the first person shooter genre by tightly controlling players’ movements across discrete spaces while at the same time permitting players the freedom to address challenges as they see fit within particular spaces, allowing for variation in narrativized gameplay while simultaneously presenting players with a singular temporal narrative. Just as immersive techniques and gameplay sequences have become more sophisticated within games since their emergence in the 1970s, so too have the strategies used by designers to produce narrative experiences using spatial tools. Video games, in other words, are continuing to mature as a narrative medium even if their stories largely remain within the genres of adventure, fantasy, and science fiction. One reason these genres have dominated video games narratives could be that they involve more movement across spaces than non-genre fiction. Given that narrative in video games is either embedded to push players across spaces or unfolds as a designer’s guided tour of spaces, these genres seem ideal in the sense that they require authors to build worlds at the same time that they create characters and construct plots.
Moreover, The Secret of Monkey Island’s relationship to its platform suggests that the importance of reading source code as part of a digital media text lies not merely in appreciating its construction. Rather, the representational gap that Gilbert employs to transcode a readily recognizable narrative using the engine’s spatial form suggests that study of the transformation that occurs between code and screen is of critical importance. What is lost or gained in that transformation may reveal important assumptions about the roles, identities, and desires of players—or the users of mundane software. A platform-based methodology is useful in this regard—even if it risks distancing itself from the ideas displayed on-screen in favor of technical descriptions—because it can open source code up to aesthetic standards other than elegance or functionality by placing the authoring of code and the experience of its execution within a cultural context. Digital media artists often assume two roles, reflecting a relationship both the with the code they write and the experience produced on screen, and critics should strive to do the same. In the case of SCUMM, the work of hackers and other enthusiasts to reverse engineer the engine and produce ScummVM laid bare much of Gilbert’s work as programmer, making it available for critical study. However, many development studios make modification tools, or even the source code to the engine itself, readily available to those who wish to look at the mechanics beneath what occurs on-screen. Finally, because the development of ScummVM is now directed toward incorporating support for engines other than SCUMM, its source code can continue to serve critics interested in examining how other authors of graphical adventure texts may have negotiated between spatial and narrative forms. Digital media texts like these, which appear to be replicating older narrative forms on the surface, can present a complicated, contested aesthetics that is not visible on the screen without a reading informed by the text’s inscription in various forms of source code.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research authorship and/or publication of this article.