Interfacing with Computer Narratives: Literary Possibilities for Interactive Fiction
People have told stories for millennia, and they have passed these stories on to others through many different media. In addition to literal story telling by means of the spoken word, people have communicated narratives in prose, poetry, photographs, film, painting, and other printed or visual forms, and through theater, pantomime, song, opera, dance and other performances. Although all of these forms engage the listener, reader, or audience in some form of interaction with the narrator and the story, the substance of the narrative does not change as a result of this interaction. In most cases, the end of the story is established before the narrator utters the first word, before the reader opens to the first page, before the curtain is drawn open. The interaction of the listener, reader, or audience with the narrative will affect the perception of the story and influence or even transform the meaning that the story's events hold for those to whom it is communicated. This interaction will not, however, cause the events in the story to be any different than they otherwise would be.
In the last two decades, a protean, interactive medium has arisen from a sea of highly technical electronic complexity. The computer, once gymnasium-sized, accessible only to those with years of technical training, and seen as nothing more than a tool for calculation and data storage, has become accessible to those in the arts and humanities. It is slowly gaining recognition as an artistic and literary medium (Note 1), through the works of writers and artists such as Robert Pinsky, Laurie Anderson, and William Gibson (Note 2), and the efforts of computer game designers whose products are now seen to have some literary and artistic merit.
While the computer differs from non-electronic media in many ways, there are some attributes that particularly distinguish the computer as a medium. Computers store all information in strings of zeros and ones, reducing everything to a digital representation. Additionally, the computer is a metamedium that can mimic other media, like film, photography, and print. Finally, computers can add an important quality to narrative: interactivity.
A computer narrative is any story told through the computer medium. This narrative may be told, for instance, by means of a computer multimedia presentation, or through a fixed text that was originally written on or has been typed into a computer to allow the reader to use database capabilities to search for certain phrases. Some computer narratives may be films, cartoons, photo essays or narratives from other media that have been digitized or otherwise translated onto a computer. These narratives need not be interactive. If a computer narrative is designed to allow the software user to explicitly interact with the story, possibly altering the course of events and the outcome of the story through his or her interaction, this narrative lies in the special category known as interactive fiction. The user of interactive fiction software is sometimes referred to as the reader or player, but these terms generate unnecessary analogies to books and games. A person experiencing an interactive fiction will be referred to hereafter as the interactor. The advent of such products as books-on-disks and multimedia computer movies makes it useful to distinguish between the general class of computer narratives and that specific subset, interactive fiction, which allows explicit interaction with the narrative.
In defining the qualities of interactive fiction, it is fruitful to begin by looking through the brief history of this genre, its "interesting infancy" (Pinsky 3), and describing some of the major works that helped to create this new art form. By doing so, the features of this genre (as well as some of its limitations and possibilities) become evident. Also, it is useful to place some of the important works of interactive fiction in a chronological framework so the influences of different works in this field can be more easily traced.
An early, important step towards interactive fiction came when programmers developed text-based games that tried to describe the setting with words and represent the actions of the player as the interaction of a character with the setting.
A Stanford hacker named Donald Woods discovered a kind of game on a Xerox research computer one day that involved a spelunker-explorer seeking treasure in a dungeon. Woods contacted the programmer, Will Crowther, talked to him about it, and decided to expand Crowther's game into a full-scale 'Adventure.' (Levy 132)
Adventure (also called Colossal Cave) became available to networked American universities on ARPAnet in early 1977 (Anderson Winter 1985, 7). Although later ported to personal computers, Adventure was designed on a mainframe computer, where the only available mode of interaction was text-based: the computer could produce lines of text on a terminal in response to typed input. So Adventure gave textual descriptions of the interactor's setting, and "for the first time, the game let the reader answer with words instead of numbers ... or ... choices among two or three possibilities." (Niesz and Holland 114) While other vestigial interactive fictions developed a narrative in response to simple one-letter commands, or the even simpler input of a yes-no answer, Adventure could parse the most basic of sentences: commands containing two words, such as TAKE SWORD or DRINK WATER. Rather than diverting the story one way or the other with a single key press, the Adventure interactor could actually contribute, albeit simply, to the text of the story.
Following close on the heels of Adventure came Zork, called Dungeon for a short time. Zork was developed at MIT. Like Adventure, it ran originally on mainframe computers, was set in a dungeon, and had a treasure-seeking theme.
"When Adventure arrived at MIT," wrote Infocom game designer Tim Anderson in an article on the history of Zork, "the reaction was typical: after everybody spent a lot of time doing nothing but solving the game (it's estimated that Adventure set the entire computer industry back two weeks), the true lunatics began to think about how they could do it better." Among the "true lunatics" at MIT were Anderson, Bruce Daniels, Marc Blank, Dave Lebling, and Joel Berez. ... "Zork was done as a midnight programming project, as a reaction to Adventure," says Berez, the first ... president of Infocom. ... "Adventure was lots of fun, but with our advanced technology and the literary skills of some of the people involved, we thought we could do a better job." (Hochberg)
An early version of Zork was running by June 1977; the program was expanded throughout 1977 and 1978. (Anderson Spring 1985) In 1979 and 1980, the more advanced version of the program was broken into three pieces so that it could be run, and sold, on personal computers. These were marketed by the company Infocom, composed of Zork's MIT developers. (Anderson Winter 1985) Zork I, Zork II and Zork III were the first of the next-generation text adventures. They were publicly available, were profitable (thus encouraging greater efforts in developing such interactive fiction), and incorporated meaningful computational advances. Zork's interface, called the Interlogic parser, allows the interactor to communicate with the program by typing sentences of moderate complexity. This interface is fundamentally the same as that of Infocom's later interactive fiction, including the more advanced works A Mind Forever Voyaging and Trinity. Additionally, although Zork was designed by programmers, not writers, Berez's statement above indicates a concern for literary skills which Infocom continued to voice throughout the years it developed interactive fiction.
After the commercial success of Zork, Infocom went on to produce dozens of other text adventures that used the Interlogic parser, including two sequels to the original Zork trilogy. Zork and most of its progeny are puzzle-based, but there are exceptions. A Mind Forever Voyaging, for instance, emphasizes the exploration of a simulated world rather than rewarding the interactor with a higher score for completing a series of puzzle-solving steps.
In 1984, Synapse released Mindwheel, one of an eventual five electronic novels. Distributed by Brøderbund Software, this interactive fiction reflected a new approach. Synapse "looked for writers who knew little or nothing about computer games, whose ideas would not reflect the biases which might creep into the mind of someone familiar with ZORK..." (Bateman 40). These electronic novels consisted of disks packaged with hardback books that were to be read beforehand by the interactor. The disk portion of the work was a text-based interactive fiction with an interface similar to the Interlogic parser.
In 1986 Activision released Portal: A Computer Novel, written by Rob Swigart. Portal has an exploratory rather than puzzle-solving structure and was one of the earliest attempts at an interactive fiction that combined literary efforts with the use of graphics. Earlier interactive fiction graphics were usually placed on top of an all-text framework, and were limited to illustrating the interactor's location. Some interactive fictions (notably King's Quest in 1983) also added the ability to direct a representation of the interactor's character across this graphic landscape.
A major attempt to create a literary work that would not be marketed as entertainment software came in 1987 when Michael Joyce designed Afternoon, A Story. This hypertext work, created with Eastgate Systems's Storyspace, consists of screens of text networked to create a space of text, or storyspace. (Note 3) The screens are not always linked through explicit, visible screen options; sometimes clicking on one word of the text will take the interactor to a new thread; sometimes it will just advance the interactor to the next screen in the current thread as if RETURN had been pressed. It is not even always evident which of these events has occurred.
With the growth of the Internet, multi-player interactive games (including interactive text adventures), run from Internet-connected computers, have become accessible to more and more people. By the late 1980s many college students, including a small but growing number in non-technical fields, were participating in these games. The most common type of fiction with multiple interactors is the MUD, an acronym for Multiple User Domain (or Dungeon, or Dimension). MUDs all use similar text-based interfaces that approximate the abilities of the Interlogic parser, with additional commands to mediate interaction between the characters whom the different interactors direct.
Another type of interactive fiction available on-line as well as for personal computers is conversational software. This type of all-text work usually minimizes plot and setting in an attempt to genuinely portray a character with which the interactor can converse by typing one voice in a dialogue. The first such work was ELIZA, a simple program from the artificial intelligence community, written by Joseph Weizenbaum, which asked questions like a psychoanalyst. A slightly later contribution to computer-simulated conversation was Kenneth Colby's program PARRY, which imitated a paranoid and was, like ELIZA, based on a simple set of rules. In 1986, MicroProse introduced Racter, a commercial entertainment program that converses with an interactor. MicroProse says that Racter wrote The Policeman's Beard is Half Constructed, a book of poetry. Modern conversational programs include the on-line Maur (in which the interactor converses with a dragon who will eat him unless the conversation goes well) and Alice (in which the interactor chats with a fellow student before a class), both available on the Conversational Hypertext server. (Note 4)
By the late 1980s, all-text interactive fictions were on their way out, at least out of the commercial sphere. Infocom, reduced almost to non-existence by financial difficulties after it attempted to market a business product, was bought by Activision.
Infocom was relocated to Mountain View, California, but most of the staff were laid off. Infocom became a label, and Activision went through a radical reorganization. Finally, Mediagenic was formed. Mediagenic was the parent company of Activision and Infocom. ... the original writers have moved on. Mediagenic went nearly bankrupt, and merged with The Disc Company. (Graves sec. 5)
Although the original Infocom was no more, in 1993, the third sequel to the original Zork trilogy was released by "Infocom: An Activision Company." This interactive fiction, Return to Zork, does not use a text parser like the first five Zork products. Instead, the interactor selects his or her action with a mouse, choosing from pop-up icons that symbolize different actions. The Return to Zork interface attempted to combine features of a text and graphical interface.
The most recent style of interactive fiction, to which Return to Zork belongs, immerses the interactor in a world of sophisticated graphics, animation, and sound. The most prominent example of this is Cyan's Myst, by Robin and Rand Miller, released in 1993. Myst, a program on CD-ROM, presents the interactor with an image of the narrative's world, from the perspective of the character he or she is controlling. The interactor directs his or her actions in this simulated world by a simple interface that contains no pop-up menus or icons and requires no typing. The interactor moves the mouse arrow to different points on the screen and clicks to move around the world and manipulate objects. Myst, like Return to Zork, includes digitized video and sound recordings of live actors. Myst displays near-photo-quality rendered images, whereas Return to Zork's graphics have a more painterly, less realistic appearance.
These examples of interactive fiction demonstrate some of the possibilities that lie before the genre today. They reflect the limits of past and present technology and the style of interactive fiction that the software market would or would not support financially at different times. Most importantly for those concerned with interactive fiction as a genre of literature, these examples illustrate some literary possibilities. The authors of interactive fiction can develop fragments of text and then link them, determining the multiple ways an interactor can view these texts. Designers can create surreal yet photorealistic worlds and engage the interactor by connecting him or her to the world with a simple interface. Texts can be written that the interactor must literally complete by typing the appropriate action of the central character, or by typing half of a two-character dialogue.
To understand the literary possibilities offered by this genre, the properties of the computer medium must first be considered. The computer medium has important attributes in addition to its interactivity, and these are discussed in the next section. Sections three and four, in part two, deal with the ramifications of interactivity on the literariness of interactive fiction. Sections five, six, and seven discuss three specific categories of interface and how they influence the interactor's connection to the work. The last part, composed of sections eight, nine, and ten, concludes with a discussion of the future of interactive fiction and of new types of interface that will yield literary possibilities for interactive fiction designers.