Author: Jonas Heide Smith

Bragt i COMUNICARE no. 3, september 2003.

Når to mennesker mødes på gaden, har de et arsenal af muligheder for at vurdere hinandens troværdighed. Kropssprog, stemmeføring og sprogbrug er af største betydning, når tillid skal etableres. Men disse cues er nærmest ikkeeksisterende i computerunderstøttet kommunikation. Så hvad gør den designer, der søger konstruktiv og tillidsbaseret brugeradfærd?

Det var sådan set svært nok allerede. Softwareudvikling handlede i årtier næsten kun om forholdet mellem en bruger og vedkommendes maskine. Brugeren skulle, helst uden en doktorgrad i kryptografi, kunne skrive et dokument, tegne en graf eller vedligeholde et kundekartotek. Dette viste sig at kræve et omfattende samarbejde mellem dataloger og psykologer, som sammen fandt hensigtsmæssige standarder for design af brugerflader. Når man i dag kan betjene en PC helt uden forståelse for dens bagvedliggende logik, er det et resultat af dette vanskelige arbejde. Men det skulle blive sværere endnu.

Langsomt blev flere og flere maskiner forbundet til hinanden i netværk. Og det medførte et helt nyt problem: andre brugere. Ikke blot skulle man nu tage højde for individuelle brugeres adfærd og psykologi, men som systemudvikler blev man nu konfronteret med en hel række udfordringer af sociologisk og kommunikativ karakter. Et par eksempler: Hvordan tillader man folk at kommunikere effektivt uden kropssprog eller øjenkontakt? Hvordan forholder man sig til tavs viden, når man gerne vil samarbejde over nettet?

World Wide Web satte fra midten af 1990’erne problemerne på spidsen, men udfordringerne er de samme i f.eks. intranetsystemer og computerspil med mange samtidige spillere. Også et grundlæggende sociologisk spørgsmål blev pludselig yderst relevant: Hvordan får man folk til at opføre sig pænt over for hinanden? At dette er et vigtigt spørgsmål, var ikke umiddelbart oplagt for ret mange udviklere, der så det som deres opgave at levere teknologi og ikke at føre politik. Men erfaringerne talte for sig selv. Det ene demokratisk opbyggede diskussionssystem efter det andet blev udsat for misbrug, og computerunderstøttet samarbejde blev ramt af et vidensdelingsproblem: Det ville være fint, hvis de andre gjorde det, men hvorfor skulle jeg bruge tid og kræfter på at dele min viden?

Lidt forsimplet kan systemudvikleren arbejde på to overordnede niveauer, når han gerne vil tilskynde brugere af flerbrugersystemer til at opføre sig konstruktivt. Disse niveauer kan vi kalde henholdsvis incitament- og æstetikniveauet.

På incitamentniveauet skeler udvikleren til klassisk økonomi og statsteori og sørger for, at venlighed betaler sig. Dette kan gøres på et utal af måder. Når det er mest primitivt, kan systemets ejer smide den bruger ud, der ikke bidrager konstruktivt. Et dagblad har måske et digitalt diskussionsforum, og dem, der pjatter, larmer eller spreder racistisk propaganda, bliver opdaget af en redaktør og smidt på den virtuelle port. Denne trussel får måske nogle til at holde sig i skindet, men kræver også omfattende ressourcer i form af en fuldtidsansat redaktør og garanterer ikke, at alle syndere opdages. En mere elegant og betydeligt billigere løsning går ud på at lade overvågningen være op til brugerne selv. Denne selvovervågning bygger på, at brugerne har mulighed for at rate hinanden, når de interagerer, og en bruger med mange point får særlige privilegier. Dette har vist sig som en overordentlig effektiv løsning.

På æstetikniveauet beskæftiger man sig med f.eks. visuel stil og sproglig tone – alt det, som brugeren ser og hører. Den amerikanske litterat Janet Murray sætter vigtigheden af disse elementer i perspektiv med et godt eksempel: Gennem mange år har det været populært inden for forskningen i kunstig intelligens at udvikle såkaldte chatterbots, hvis eneste opgave er at indgå i dialog med mennesker og helst virke så overbevisende som muligt. Jo mere den menneskelige samtalepartner er tilbøjelig til at tro, at vedkommende chatter med et andet menneske, jo bedre. En klassisk strategi er gået ud på at hælde en stadigt større mængde ord og grammatiske regler i disse bots (sådanne strategier betegnes ofte Good old fashioned AI). Men Murray erfarede, at en meget simpel bot faktisk havde overordentlig stor succes.

Trylleordet her er selvfølgelig kontekst. Systemudviklere har mulighed for, ved hjælp af metaforer og lignende, at skabe en kontekst, som brugeren træder ind i. Og eftersom de fleste mennesker alt andet lige er tilbøjelige til at opfylde opfattede forventninger, har man her en meget effektiv måde at styre interaktionen på. Søger man at skabe et forum for oplyst, filosofisk debat, er det smartere at skabe en visuel og sproglig ramme, der henleder brugernes tanker på Antikkens Grækenland, end en, der får dem til at tænke på skyttegravskrig.

Konklusion

Internettet har skabt en ny virkelighed, hvor sociale interaktioner er blevet centrale. At understøtte ønsket om behagelig interaktion kræver den rette kontekst – en kontekst, der bedst etableres på baggrund af tværfaglig forståelse og med deltagelse af velfunderede kommunikatører.

Date posted: June 1, 2000
Updated: Apr 11, 2007

Historical foreword
The details, unfortunately, are rather sketchy. This should not come as a surprise since, after all, the people involved had little idea what they were doing.

Surely, when Don Woods, a student at the Stanford University’s Artificial Intelligence Lab, sat down at his keyboard one day in 1976 to write an e-mail he didn’t realise he was also about to write computer history. Woods was determined to get in touch with a certain Will Crowther. Not that Woods had ever met or spoken to Crowther, actually he didn´t even know Crowther´s e-mail address.

The reason for Woods´ search was a computer game. Now this may well convey the wrong impression, because the game in question would surely stand out among its digital descendants of our day. Today´s games are audiovisual endeavours produced by teams of dedicated full-time designers. The game that had so intrigued Don Woods was something else entirely. It had no graphics, it had no sound, but nevertheless it was known as: Adventure.

Deducing a person´s e-mail address from a name would soon become hard bordering on the impossible. In 1976, however, only a handful of servers were connected through the ARPAnet. Woods simply tried them all, using ´crowther@´ as the prefix. In the message, which reached Crowther in Boston, Woods expressed his admiration for what he had found on the university computers. Adventure had been born in an attempt to combine Crowther´s interests in programming, caving, and fantasy role-playing. The small program was a crude cave exploration simulator, which let the player interact with an environment of caves interconnected through what the famous opening line of the game described as ´a twisty maze of passageways, all alike.´.

Now, at the time of Woods´ enthusiastic e-mail ´broadcast´ Crowther was somewhat busy contributing to the foundations of the ARPAnet. Woods, however, had less momentous things on his mind. He wanted most of all to expand and improve on the rather bug-ridden and frame like structure that was Adventure. Perceiving this as a flattering display of interest rather than an infringement of authorship Crowther happily gave Woods his blessings.

For this was indeed no scheme to change the world. By his own admission Crowther had been satisfied that his kids considered the game ´a lot of fun´ and Woods thought it appropriate to expand the structure, add to the game´s fantasy flavour in recognition of his admiration of J.R.R. Tolkien and then graciously leave the game behind on the network while going away for the summer.

There was a certain sense of naivety and unconcerned playfulness in the air when interactive fiction was born. The people involved indeed had little idea what they were doing.

Only in retrospect can we hear the distant sound of trumpets.

Section 1: Introduction
Stories are not what they used to be. There was a time when the bourgeois novel, leaning heavily on the time-honored tradition of classical Greek drama, represented the epitome of narrative perfection. The twentieth century saw an end to this ideal. Artists of most media joined forces in an aggressive attack on what was seen as uninspiring conventionalism, a conservative craftsmanship that hindered any radical potential of art. These artists were modernists, and almost by definition a modernist must attack tradition and the shackles of convention. In painting what had been accepted as progress towards truthful representation (see Fig. 1) now was accused of cowardice and even hypocrisy as its striving for truth in perspective was considered both naïve and ideologically biased.¹

Figure 1 ´ Mabuse: St Luke painting the Virgin [section] (c. 1515). The scientific perspective and academic/religious motive would later be considered tasteless by many modernists.

In literature, following examples set by the provocations of Joyce, Kafka, Borges and others, modernism made a spectacle of challenging narrative cohesion, particularly by employing motives and techniques inspired by the thriving science of psychoanalysis. Filmmakers such as Bergman, Fellini, Antonioni and others later employed a modernist aesthetics as part of a quest to raise cinema to the level of art form (Thompson & Bordwell, 1994:492-557) and relied heavily on a deconstruction of Hollywood conventions and a focus on internal (often, it seems, psychopathological) realism.

In what may be termed linear media one can observe a tireless attempt to confront the limits of narrative form. The streams of consciousness and rigorous jump-cutting have been an attack on the vice like causality of linear projection (and printing). It has been an attack on the seductive qualities of traditional fiction and a call for a more critical audience, an audience in charge and in a state of constant reflection. It has, in many ways, been a call for interactive fiction.

This article attempts to map the promises and the limits of interactive fiction. This task is undertaken in a spirit of enthusiasm but also with a critical sense of purpose. While this article is no manual for the construction of such fiction and certainly has no wish to promote one design style at the expense of others it does take the form of a critical inquiry into what has been said and done in the field. For, as I hope to make clear, some rather unfortunate choices have been made.

The title of this paper refers to the themes that will be examined within. First of all it hints at the very pleasure or pain that may ´ or indeed may not ´ be specific to interactive fiction, that is the issue of deliberate choice. Secondly, and more importantly, it serves as a heading for A) a description of how the future of interactive fiction seems to lie in the very opposite direction from where it has been headed since Adventure, and B) how a disturbingly large number of people in an effort to theorize the phenomenon have walked down a crooked path that unfortunately leads nowhere in particular. The road they travel spins in circles because they have chosen to focus on questions of the ´what´ type. In this paper I argue that to understand the limits and potential of a phenomenon one must start with the ´how´s and ´why´s.

A note on language
Some of the cited literature is only available in Danish. To ensure the usefulness of this text, however, I have focused on English language works and wherever possible I have supplemented all references to Danish literature with a reference to the English version of the text. All Danish quotations are translated by the author.

Focus
Essentially this paper seeks to prove the following hypothesis: There is no logical contradiction between interactivity and engaging narrative experience. It is argued that the widespread conception that such a contradiction exists is a result of a particular (and according to its own ideals somewhat peculiar) design choice and that the most loudspoken branch of theory has mistaken this choice for the nature of interactive fiction.

Structure of the argument
As implied this paper attempts to add bricks to the foundation of a research field. Rather than applying diffuse terms to a highly specific aspect it aims to supply a rough map of this field. To stick with this metaphor I find it arrogant, if not down-right wrong, to believe that one can just pick any plant in a field and use it as a basis for wide reaching generalizations (analyzing one computer game without bothering with the history of the medium, for instance). This is very different from saying that one can learn nothing from a detailed case study ´ quite the opposite is true ´ but this case must be chosen with care on the basis of a broad understanding of its context and history.
The argument presented here falls into four sections.

• Section 1 presents the case.
• Section 2 tells the history of interactive fiction and attempts to explain why it went astray (according to its own ideals). In the service of illustration and ´grounding´, the game Gabriel Knight III (Sierra, 1999), a typical example of the genre, is analyzed.
• Section 3 confronts existing theories of interactive fiction. This section argues that, while impressive work has been done to systematically approach the subject, much effort has been wasted in a battle of words. Theorists from humanistic disciplines have fallen victim to, what may be termed, the ´substantialist fallacy´, thereby actually working within the same faulty (or unconstructive) framework of the game designers. As an alternative I suggest an increased focus on why anyone would want to spend time on fiction in the first place and call for heightened attention to the need for interdisciplinarity.
• Section 4 sums up the arguments and presents ideas for future research.

Terms used
Not surprisingly the explosive spread of digital communication and the rapid change these technologies have undergone has left us somewhat at a loss for words. What was almost securely categorized under a heading of number crunching hard science broke loose rather unexpectedly and became a full-fledged medium. Computing so obviously a method of calculation became a medium of communication. Meanwhile, with the growth of an industry able to promote its services almost entirely on grounds of novelty, considerable sums were spent on the word game. New terms were coined and shaky metaphors constructed in a haphazard manner. What we have from this source is a plethora of ad hoc terms counting such notabilities as ´Cyberspace´, ´Information Society´, ´Global village´ and a true magnet of fragile definitions: ´Interactivity´. Certainly the major players on the market were never at a loss for words. Some would argue they used far too many.

To reach any sort of progress in any scientific endeavor knowledge must be systematized, hypotheses operationalized, and terms clarified. With this in mind I carefully use the following:

Interactive Fiction: A highly contested, some would say ideological, term. It was formally introduced in a Byte article in 1981 (Aarseth, 1997:48) as a label for what had previously been known as ´storygames´, ´compunovels´ etc. It was used as a label for story centered computer games. At other times, however, it has been applied to books* * Most notably the Swords and Sorcery series of the early 1980s. The cover of board and role-playing game designer Steve Jackson´s The Citadel of Chaos promised “A fantastic story with YOU as the hero.”, theater (Laurel, 1993:52) and film². Espen Aarseth (1997) claims that the term unjustly connotes freedom and revolutionary potential. Indeed like ´interactive entertainment´ it appears to be a way for both industry and players to escape possible pubescent connotations of ´computer games´. Finally it should be noted that many fans exclude anything from this category that isn´t the real (that is: the old) form of the adventure game, in other words they exclude anything that is not purely textual interaction. In this paper the term signifies any fiction in which the user is required to participate at the level of the syuzhet³. The term ´adventure game´ is used interchangeably.

Interactivity: While this term has proved extraordinarily slippery and hard to define within media studies it will be used loosely as a measure of a medium´s potential for letting the user shape contents and form (for a discussion of the concept see Jensen, 1998).

Syuzhet: The denotative level of a text. As opposed to fabula, the mental construct of the individual reader (following Bordwell, 1985:49ff). These terms have the disadvantage of being obscure but the clear advantage of being uncontested (unlike various semiotic equivalents).

Computer game: Any piece of software on any platform, which explicitly rates the performance of the user/player or demands a certain performance in order to make the program proceed in a manner described as desirable. Although one may find software that only border on this category (mainly virtual isotopes but also goalless strategy games such as SimCity, Maxis 1997) this definition is adequate for the purposes of this paper.

Interactor: The user of interactive fiction⁴. The interactor exists between old categories such as author, reader and viewer (Laurel, 1997:152; Montfort, 1995) but is best thought of as an agent in a world made by others. A virtual representation of the interactor will be referred to as an avatar.

Interactive fiction is a young research field to say the least. Terms employed within the design community are often unknown even to players and in order to avoid misunderstandings I make comprehensive use of explanatory notes and illustrations.

Section 2: The ´how´s of interactive fiction
Adventure games were always a little different. Set apart from the mainstream of computer games by their focus on contemplative deduction and their lack of immediate commercial appeal to owners of arcades they have long been considered worthy of attention from even the most serious of ´old media´ (e.g. Rothstein, 1983; Lassen, 1997; Frost-Olsen & Schmidt, 1998). Far from the hard rock soundtrack and junk food of contemporary arcades (for a nostalgic history of arcades see Herz, 1997:43-61) adventure game players would sit silently in the night pondering the logical puzzles that soon became an important hallmark of the genre. Thus these games would fall squarely within the category of fine art in that they:

´ Required intense contemplation/reception (as do literature and painting)
´ Required (or invited) solitary reception (as do literature and painting)
´ Were obviously based on story types found in literature (although the crime mystery and fantasy templates often employed would not qualify as high-brow)

Western culture speaks of art in the language of literature. Artists are identifiable individuals with a need to express deep-felt insights into the human condition, preferably with no thoughts of marketability. Franz Kafka, in this perspective, was an artist par excellence, as he preferred to see much of his work burned before he died. The artist-as-person criterion has forced filmmakers and critics to perform a rather peculiar maneuver. In order to define film ´ a highly collaborative form of expression – as art the filmmaker has had to become an auteur, a writer, who paints with the ´camera as his pen´. In the words of film critic Alexandre Astruc:

´After having first been a jester attraction in a marketplace, then an entertainment form much like boulevard theater, or finally a means of storing the images of an epoch, film is little by little becoming a language. A language, that is a form in which an artist may express his thoughts, however abstract they may be, or articulate his problems, as had it been an essay or a novel.´ (Astruc, 1948/1970; see also Bordwell & Thompson: 37-38).

A similar strategy was employed by adventure game designers. The game industry has traditionally been peopled by anonymous craftsmen, but within the adventure genre the auteur principle has produced such titles as Roberta Williams´ King´s Quest (Sierra, 1984) and Al Lowe´s Leisure Suit Larry (Sierra, 1987). So indeed adventure games have tried to remain different. But are they truly more worthy of cultural recognition than other genres? Obviously this is a dangerous question, but I suggest that a case can be made for the opposite position; adventure games ´ in their present incarnation – are the least interesting of all. This requires an explanation. And for that we must inquire into the history of the matter.

A history of interactive fiction⁵
At the time when Will Crowther was writing the first lines of Adventure dungeon exploration was already a popular pastime in the computer community. In 1972 Gregory Yob of the University of Massachusetts had developed the legendary Hunt the Wumpus (Hunter, 2000; Herz, 1997:9-10). Wumpus is usually considered too crude to qualify as an adventure game as the input of the interactor was restricted to a multiple-choice response in the form of a direction (e.g. ´North´) and one principal action (´Shoot´).

Although at this time producers of arcade games ´ particularly early believers at Nolan Bushnell´s Atari – were experiencing a number of remarkable successes with titles such as Pong (Atari, 1972) programmers of the time-sharing networks had little thought for marketing their entertainment products. Mainframe based games were considered collaborative endeavors in a culture that conceived of computers as a means of resource and knowledge sharing⁶. That is until Zork (Infocom, 1979) came along. Sensing unfulfilled potential in Adventure an MIT based group began work on a more complex and artistically rewarding game, which was to dramatically enhance the interaction between machine and interactor. In an effort to decrease the level of frustration stemming from the inhibiting parsers of earlier attempts the programmers of newly founded Infocom went to work on allowing the machine to make qualified guesses as tothe wishes of the interactor (Lebling, 1982; Murray, 1997:74-82). Zork, communicating only in text, greeted the interactor like this:

West of house.
You are standing in an open field west of a white house, with a boarded front door.
There is a small mailbox here.
>

Figure 2 ´ A map supplied with Zork (Infocom, 1979). The inspiration from fantasy novels and role-playing games of the time is obvious. For a long time elves, dwarfs, and dragons were a mainstay of interactive fiction (as they still are to some degree).

Figure 3 ´ Mystery House (Sierra, 1981). Crude illustrations help convey the story.

Figure 4 ´ King´s Quest (Sierra, 1984). This game used a combination of colourful graphics and traditional textual interaction. The graphical avatar (seen between castle and tree) interacts with the environment.

The ´>´ invited the interactor to state his wish in natural language. The command “Open mailbox´, for instance, would motivate the response ´Opening the mailbox reveals a leaflet´. A strength of Zork was its ability to respond creatively to unanticipated (and unknown) commands. ´Hit mailbox with hand´ resulted in the response ´I´ve known strange people, but fighting a small mailbox?´ (where ´small mailbox´ is merely the particular noun employed in this case. In other situations it would be ´tree´, ´rock´ etc.). The program understands the noun ´hand´ but instead of responding with say ´You hit the mailbox to no effect´ the actual response serves to ´script´ the interactor. It communicates: ´Stay within the framework of the fantasy genre or be ridiculed´ thus ensuring that the interactor will not wander aimlessly outside the range of anticipated actions.

The game became hugely popular, especially when ported in 1981 to the Apple II micro-computer and ensured the immediate future of Infocom, which came to stand for thought provoking high-quality adventure games throughout the early and mid-eighties (counting such highlights as Deadline, Planetfall, Witness, and A Mind forever Wandering; see Wilson, 1991).

Others, however, were quick to catch on. With the growing penetration of microcomputers a market appeared for games that were not dependent on the repeated restarts that were a certain feature of arcade games. Programmers Roberta and Ken Williams founded Sierra (first known as On-Line Systems) on the success of Mystery House (Sierra, 1981), one of the very first adventure games to employ graphics as illustrations of the story (see Fig. 3). At this point, however, the graphics were still ´dead´, inactive in the sense of plot as they merely provided redundant information. In this fashion they may be likened to pictures in a novel – means of establishing ´atmosphere´⁷.

While Infocom´s text adventures became increasingly complex Sierra placed their bet on more spectacular (and thus marketable) family friendly games with colorful graphics. Taking their cue from television series that benefited from familiar characters and low production costs Sierra became famous for their ´Quest´ series. King´s Quest (Sierra, 1984), in particular, promised colorful experiences that would take machines to the limit of their capacity. But perhaps the truly important innovation was the introduction of the on-screen character, or avatar (see Fig. 4). The interactor assumed the role of King Gawain using the avatar as a concrete extension of himself⁸. Although surely felt as limiting to the explosive imagination of die-hard Zork fans this construction did serve to ´ground´ the interactor within both story and genre and to minimize the spatial confusion that might arise from purely textual descriptions.

With one very important exception this was the dominant form of the adventure game until 1987. The exception was a very strange concept known as Little Computer People (Activision, 1986). Now, following a number of unfortunate dispositions the once-proud Infocom had been acquired by Activision who were known for their contributions to the action genre. The take-over, however, did not sit well with the Infocom ´artists´ who saw the development as a threat to their artistic integrity and to what was seen as a style and humor specific to Infocom (Wilson, 1991). Little Computer People was a game unlike most. Actually it was not even a game by most standards. First of all it had no goal, no way to unambiguously win or lose. The role of the interactor was to ´look after´ a man living in a three-story house – no more, no less. The on-screen character needed food, sleep and comfort to thrive. If he were mistreated he would grow listless and possibly sick. Although a similar structure was later used for bestsellers such as SimCity (Maxis, 1987) and Sims (Maxis, 2000) (and obviously also resembles such phenomena as the ´Tamagotchi´) Little Computer People received nothing but scorn from the Infocom crew. The celebrated Infocom creators of interactive fiction saw their own way as ultimately superior to this virtual pet. They should, perhaps, have looked more carefully.

Meanwhile a new player arrived on the scene. LucasArts backed by the thriving LucasFilm empire set out to conquer the adventure game industry and did very well indeed. Perhaps their fresh perspective allowed them to boldly discard what had been considered an essential component of the genre: the textual interaction. Just as likely though it was a sure grip on other forms of narration that allowed them to eradicate the alienating effect of parsers that would often force the experience dangerously towards a word guessing game. In Maniac Mansion (LucasArts, 1987) they aimed at continuity and flow by inspired use of the point-and-click interface⁹.

Figure 5: Maniac Mansion (LucasArts, 1987). The point-and-click interface lets the interactor combine a dynamic set of verbs (bottom) with graphical objects of the game world.

Figure 6: Myst (Brøderbund, 1993). Although the game was heavy on puzzles and still made the interactor follow a certain route Myst presented itself as an experience, as a world to explore.

A set of essential verbs would change to accommodate likely actions based on context (see Fig. 5). While this gave strong hints as to the required action it allowed the designers to focus their energy on the possible rather than devote their time to devising clever ways of avoiding and discouraging attempted impossibilities.

Realizing the potential of this approach Sierra quickly adapted, but LucasArts ensured their position with Zak McKraken and the Alien Mindbenders (1988), Indiana Jones and the Last Crusade (1989), Loom (1990), and the Monkey Island series (1990-1997).

With few variations the point-and-click interface dominated the adventure genre in the 1990´s. Indeed the form seems to have remained essentially the same while dramatically increased storage media capacity has motivated an explosion of audiovisual sophistication in what some have seen as an immature focus on dazzling effects. Most notably the aesthetics of Hollywood cinema were embraced halfway through the decade in a costly attempt to substitute computer drawn 2D graphics with digitalized film. Almost invariably the prohibitive expense of needing to shoot every action performed without relying on any sort of computational short cut resulted in games suffering from painfully rigid structures.

That players could still be swayed by less dazzling displays, however, became more than evident with the unprecedented success of Myst (Br´derbund, 1993 ´ see Fig. 6)¹⁰. To some extent Myst may be seen as a counter reaction to the still more linear structures dominant within the genre. The almost meditative and highly process-oriented experience apparently spoke to new audiences that cared little for more action-tilted designs.

At one point this fact may have been belied by singular bestsellers but the genre was fighting a losing battle. The decline had little to do with interfaces and even less to do with visuals but time was running out for a breed of games that strove to lead one interactor through a prewritten story. As network technology blossomed in the mid-nineties computer game playing became once again a more social activity. From the public arcades of the seventies, the eighties had brought computers into the privacy of consumers´ homes (Egenfeldt-Nielsen & Smith, 2000:51-57/86-89). This was the perfect setting for games of deductive logic but now the possibilities for intelligent and less predictable opposition and team play sparked a cry for connectivity ´ for games that allowed several players to interact within the same environment. This was a demand the adventure genre could not honor. Since adventure games following Adventure have been character-oriented the worlds created have been mere backdrops, objects nothing but stage props, in essence just as ´dead´ as the graphics of Mystery House (everything is set in motion by the choices of the interactor that work rather like the movie director´s call to ´action!´). In a time of networks the excruciating linearity behind traditional interactive fiction stands out in flashing neon. The much-touted interactivity of interactive fiction seems little more than a gimmick added to stories that perhaps ´belong´ in other media. For a clear diagnosis we need a closer look. The following section presents a brief analysis of a carefully chosen specimen.

The linear case of the shadow hunter – an analysis of Gabriel Knight 3
I have claimed that the difference between Adventure and its predecessors is mostly cosmetic. Now whether a certain change in expression is ´purely cosmetic´ may of course be a question of heated debate. While many would probably argue that adding sound to movies may be considered substantial progress in narrative terms other ´surface´ changes such as the switch to a modernist style of painting may also be construed as a purely cosmetic change. Thus whether the alterations and subtle techniques employed in contemporary adventure games are in combination sufficient to count as a ´fundamental´ change is not a question easily answered. The aim of this section is more modest: To show that present-day adventure games suffer from almost the exact same range of problems that faced Crowther and Woods. Secondarily I provide the argument that the present form is a consequence of a way of thinking common to artists of linear media.¹¹

To do this I will perform a brief analysis of a game that seeks to embrace tradition and carry on the genre: Gabriel Knight 3 ´ Blood of the Sacred, Blood of the Damned (Sierra, 2000)¹². I assume that some generalization will be possible considering that the case is situated within the history described above and handpicked for its general qualities. For a recent and highly influential discussion of case study methodology see Flyvbjerg, 1991].

´San Greal´ are the first audible words we hear. Amid swirling shapes and patterns human figures are seen standing in a dimly lit compartment. Details are impossible to make out as the blurry vision is suppressed by a feverish dream peopled with creatures of legend.

Figure 7: Gabriel Knight III introduction. Strange figures loom above Gabriel. The sound of a driving train is heard.

Figure 8: Gabriel Knight III introduction. Gabriel slips in and out of a strange dream.

Figure 9: Gabriel Knight III introduction. Slowly awakening we try to get our bearings. This is the interactor´s first in-game view of the avatar, Gabriel Knight. On-screen credits betray obvious inspiration from cinema.

Figure 10: Gabriel Knight III (Sierra, 2000). In 3D action mode the interactior may move the perspective around as in this church. When an object is clicked Gabriel will attempt to perform the chosen action.

Figure 11: Gabriel Knight 3 (Sierra, 2000). Hotel guests drinking coffee in the hotel eating room. They show no particular interest in the avatar.

This is the beginning of an adventure. Willing or not, from this moment we must pursue the answers to a mystery set in motion by others. While we, the interactor, watched the introduction to Gabriel Knight 3 we were, if we listened carefully, perhaps able to pick up the muffled sound of a train in motion (see Fig. 7-9). As Gabriel, our avatar in the game world, painfully comes to his senses we find that we are indeed on board a train ´ the sounds were diegetic ´ but as we diligently pursue the enigma that is the story and struggle with the puzzles that would bar our way another interpretation presents itself. Whether intentionally or not, the designers, by choosing the sound of a train, has employed a fitting metaphor. Indeed there is only one path.

Still independent of our input Gabriel stumblingly leaves the train at the first stop. On the darkened platform he is greeted by a conductor who informs him that a taxi is waiting outside the station to take him to a nearby hotel.
We have just witnessed the full introduction performed as a 2D cut scene, one of the four modes of the game:
2D cut scene: Bits of the story presented as traditional linear cinematic narrative told outside the game’s own engine.¹³
3D cut scene: Usually dialogues or small one-location drama told within the engine. The default setting has prescribed editing (choice of camera positions) but the interactor may maintain control of the camera if desired.

Special puzzle interface: Puzzles of deductive logic have been a steady feature of adventure games since Adventure. Gabriel Knight often transports the interactor to special modes of interaction in order to solve a given puzzle.
3D action: Most of the game is played with the interactor in control of one of the two main characters, Gabriel and Grace. The action is shown in polygon-based 3D graphics, allowing for extensive freedom of movement within the scenery (see Fig. 10). Whereas Myst and its contemporaries used a limited collection of still images (sometimes enough for animation) the engine of Gabriel Knight is able to compute and display any camera angle on the fly (for an example of several graphic modes see supplement A).

Untraditionally the interactor is mainly in control of the perspective mimicking a moving movie camera. Thus the user may move the perspective around and only needs to click on objects that should be handled by the avatar. In the tradition of cinema the perspective may roll, tilt, pan, and track but may not pass through solid objects.

The limits of freedom
Standing in a sunny hotel room at the first moment of interaction the world of Gabriel Knight appears open to the interactor. A flight of stairs lead downwards and into a small rural French town that seems at first charmingly alive. The hotel staff display mild interest but otherwise most characters in the hotel often ignore the snooping Gabriel unless spoken to. Letting the inhabitants seem to mind their own business and often seem annoyed when the ignorant American commits yet another social faux pas is an inspired design decision. It goes a long way towards masking the fact that these creatures do nothing, and indeed do not exist but for their importance to Gabriel and Grace (see Fig. 11). Now this may appear glaringly obvious as in many ways this is analogous to movie characters that strictly speaking do nothing off screen. But when this principle is translated into game terms it poses a serious problem: a problem of predetermined causation.

Imagine, if you will, a sequence of events connected by causality: A murder leads to a hunt for the perpetrator who is ultimately caught. If this is the story one wishes to tell a free protagonist becomes a threat. Surely one cannot allow the protagonist to make choices that derail the narrative; the risk is obvious that the story would never get told and that the interactor will be bored as nothing happens. A designer who wants to tell the story of how A leads to B while maintaining an interactive element will start going to great lengths to ensure the interactor that the choices he makes are important while making absolutely sure that they are not¹⁴. Enter the puzzle.

The interactivity of cross-word-puzzles
As the designer cannot let the interactor upset or change the series of events in any important manner he is faced with the choice of either letting the interactor watch passively as events unfold (not a game by any definition) or maintaining that the interactor is important by other means. The strategy chosen throughout the history of interactive fiction is the interactor as starting gun. Only when cued by the interactor the inertia of the story is dispelled and the narrative progresses¹⁵. Such cues may take any form but since walking through an open door is seldom associated with glorious victory doors in adventure games tend to be locked. The quest now becomes one of finding the key, the crowbar or better yet the rope that when used with the hook will grant the avatar access through the window instead. Puzzles rarely have more than one solution so often the freedom of the interactor is limited to discovering the ´right´ choice ´ it is a question of ´solving´ a story rather than participating actively. In Gabriel Knight much time is spent on perusing obscure evidence for clues to the many deductive/geometric puzzles that sometimes threaten to halt the game entirely. The fact that finding the solution to such a puzzle instills the interactor with an unambiguous sense of victory may explain the genre´s penchant for detective fiction that often celebrate the relationship between logical reasoning and success.¹⁶

The horrors of geometric progression

Figure 12: Standard narrative model for linear fiction. This model may be traced to the poetics of Aristotle and remains the basis of traditional literary fiction as well as a cornerstone in the Hollywood style of script writing (Adapted from Larsen, 1995:102).

While the main barrier to consequential interactivity is the problem of causality the interactor as starting gun approach presents the designer with more pragmatic difficulties. If an object only reacts in strict correspondence with a predetermined reaction (and thus is not an object with special properties but rather known to the engine merely as a graphical element) then every reaction to every possible action needs to be programmed. Similarly if, for instance, the interactor of Gabriel Knight guesses the truth about the free masons operating in the area he may not act upon this knowledge before the appointed time. In essence the interactor will know more than the avatar does. The exposure of the secret society is carefully entwined with other story elements and if the interactor were able to alter the order of events however slightly the whole structure would have to be rearranged. This problem presents itself to the designer, as a fear of geometric progression ´ if choices really mattered a tremendous amount of expensive material, or ´lazy bytes´ (Crawford, 1982/1997:46), would not be displayed to a given interactor and the manuscript of the game would soon become impractically complex.¹⁷

Figure 13: A model of traditional interactive fiction. Within normal sections (or chapters) the interactor may operate with some freedom. But to get to the next section he must bow to the prescriptions of the story and thus temporarily abandon his freedom in order to progress.

Roundabouts and plot points – a model of traditional interactive fiction
Models work best in retrospect. While one may ´ unconsciously perhaps – base a narrative upon time-honed templates a scientific model should only (and indeed can only) deal with realities, whether concrete or theoretical. What models do best then is alert us to general principles in what may seem at first glance random and indeed to remind us that what may seem natural is often, though far from always, merely traditional.

Figure 12 is the basis for much linear fiction. Such a structuring of events may even, as cognitive film theory has suggested, be a standard model of human perception (Bordwell, 1985; Branigan, 1992). Traditional interactive fiction often works towards maintaining this overall form. From a position of ignorance the interactor is taken through a learning process that ends in a climax.¹⁸

The above model (fig. 12) works on the level of syuzhet but as the interactor may well use or abuse his control to ´flatten´ or break this curve a more enlightening approach may focus on the amount of interactivity present throughout the course of events. Such a model has been proposed by game designer Michael Valeur on the basis of a rigorous analysis of the ´how´s of the genre (Valeur, 1998). Figure 13 is a loose adaptation.

While the interactor may enjoy some freedom within chapter 1 and even some influence on temporality he must conform to the logic of the narrative if he wants to proceed. Although this model gives little indication of the circular paths that the interactor must often follow in search of vital clues it is a very accurate description of how the structures work.

Against text
But the structures are arbitrary. They can be changed. Before we attempt this, however, we need to realize one thing: The problem is in the words. Culturally we measure quality by standards of literature. Artists of new media need to present themselves as individual vehicles of creativity, as writers, as auteurs. As in the model above we speak of chapters as if they were a natural way of delaminating experience. Adventure game design is done by writers who produce manuscripts. If these manuscripts map any sort of narrative process they will have endings. Endings are like destiny ´ if you have it, then every choice you make is unimportant. We can´t have endings.

Deistic narration
Deism is the belief that the Christian God set the universe in motion and then left it to its own devices. Presumably God did not plan for all contingencies. Actually, according to this belief, God didn´t plan for anything, but merely constructed a frame. But what is narratologically important is that event will occur, stories will appear, on the basis of ground rules and relationships. Perhaps J. C. Herz´ analysis has been underrated due to its journalistic style:

´There are some major advantages to creating the world first and worrying about the characters and plot later ´ the Old Testament approach to game design. First of all, it´s a way around the ´Pirates of the Caribbean´ syndrome, where you feel like you´re on some kind of monorail through the game. You can veer slightly in one direction or another, but you can never go outside the lines. Either the characters push you back into the main lane by implacably parroting the same three lines, or the virtual camera takes you prisoner on a forced march of zooms and dolly shots.´ (Herz, 1997:154).

Herz mockingly describes Dragon´s Lair (Bluth, 1983) as a worst-case scenario of cross-media hybrids. In her own words the game was “just a television that someone had made really, really difficult to watch” (Herz, 1987: 147). If the choices of the interactor are to have any significance beyond the continue/stop dichotomy situations must be open-ended. Such situations are object-oriented.¹⁹

To achieve this kind of freedom the designer must lay behind the thinking of the writer and become an architect. The architect conceives of a building holistically. This is highly important for consistency and verisimilitude but far more importantly, a building is not dependent upon any specific action being performed inside. Instead the architect draws the building with the highest degree of non-linearity, has it build and then steps back.

An example: A virtual living room is designed, the avatar is placed on a couch, and a dragon is placed under the couch. What we then have is not a story but a story is what we may get. What we have is a starting position with narrative potential but without direction.

To turn the dragon into a true object it must be given properties and preferences. It must have a system of interpretation and a range of reactions. This may for instance take the form of a system that interprets anyone coming within 100 pixels of the dragon carrying a visible weapon as a threat. According to parameters such as cowardice/bravery, sleepiness/restedness (the dragon´s personality) a reaction is chosen. If the dragon´s options are limited to ´fight´ and ´run´ the interactor is unlikely to feel part of an outstanding artistic experience but with only a few variables what we have is a self-supporting system of unpredictable direction and better yet: We need no longer worry about the branching paths of geometric progression.

Experiences from virtual worlds
The computer game industry may not appear to be sparkling with the youthful fervor characteristic of the innovative early 1980´s (Crawford, 2000). Still, newly developed technologies such as improved hardware for on-the-fly 3D graphics rendering, are quickly adopted by a business that often markets its products by reference to minute improvements in frame rate. When the Internet came along it was not ignored. Newly found distribution channels were pushed to the limit (Herz, 1997:83-90) but inexpensive network technology also revolutionized multi-player functions by letting several players interact simultaneously. As mentioned above this was a dead end for character-centered adventure games but the other genres competed aggressively for the bandwidth.

Figure 14 ´ Ultima Online (Origin, 1997). Players log on to a virtual world of medieval adventure. The world is persistent and in principle every choice may change the world forever. In this picture a party of avatars is attacked by a band of lizard men. A traditional role-playing style character sheet is available in the upper right corner.

In 1997, though, a world was created. Origin, makers of the long-standing Ultima series (1980-2000) challenged the need for endings.²⁰ Ultima and its predecessors have been fantasy role-playing games, often considered a sub-genre of the adventure game. Role-playing games, often heavily combat-oriented, typically follow J. C. Herz´ principles of ´Old Testament game design´ by establishing worlds in which stories may take place. The principles of random encounters and the importance of trade make a game such as Baldur´s Gate (Interplay, 1989) highly flexible and, while not open-ended, highly unpredictable and customizable. Appropriately enough Origin named their latest creation Ultima Online (see Fig. 14).

Ultima Online was soon followed by competitors Everquest (Verant Interactive, 1999) and Asheron´s Call (Microsoft, 1999). All follow the same general principles: Players purchase the game and must then pay a monthly fee (along with any dial-up charges) to be allowed access. Once inside the player creates a character from a set of options, deciding on skills, character class, and various aesthetic settings (hair color, gender etc.). What happens then is a combination of the player´s choices in relation to the choices of others and of course the natural ´laws´ of the game world.

The poetics of the starting point
Short of a frame it is difficult to construct a model for truly non-linear fiction. But the frame is all-important. At a glance it may be hard to appreciate the difficulties of deistic narration; it would perhaps seem that the players do all the work. In many ways Ultima disproves this, for the game is crippled with problems of design.²¹ Despite the collaborative ambitions of Origin (and indeed of the role-playing genre) teamwork in Ultima Online is more than difficult. The interface is arbitrary, communication is challenging, and the concessions to a range of connection types make the game appear sluggish and unstable. But although these problems must be solved before much else can be achieved the challenge of imbuing a starting point with narrative potential while weighing all elements against one another is the more daunting task. The unpredictability of this self-supporting system makes it imperative that no major imbalance is introduced, as this imbalance is apt to increase over time. These are the problems most often mentioned by game designers but as Janet Murray hints (Murray, 1997:283-284) a deistic poetics will also need techniques for creating an atmosphere of true role-playing. Ensuring that all ´ or just enough ´ of the players remain ´in character´ and thereby add to the enjoyment of others requires the careful use of genre cues and the introduction of techniques that reward role-play without restricting player freedom.²²

A good example of the importance of genre cues is given by Murray. She describes how the liveliness and therefore successfulness of ´chatterbots´ is not dependent upon the size of their vocabulary or their understanding of grammar but rather on their ability to ´script´ the interactor. Thus a ´psychotic girlfriend´, by its very name cuing the interactor, does a good job of subtly limiting the interactor´s responses (Murray:1997:219-220).

While suffering from a broad range of childhood diseases online role-playing games seem apt to change the future of the medium. It is true that a good case can be made that the industry shows signs of conservatism (and certainly the fantasy templates seem hard to put aside) but in this case what is, from the perspective of this paper, a narrative revolution has been started. Hard-core adventure fans mourn the apparent lifelessness of their genre, but if they looked a little more closely they might realize that instead of dying the genre has adapted.

Section 3: The ´why´s of interactive fiction
´You are standing in an open field…´. Not long after Zork had greeted its first interactor in the second person voice literary theory turned its searchlight upon this curious phenomenon. Since then the theories of interactive fiction have tried to accommodate this new form of literature into an existing framework. There is nothing wrong with this approach. In this paper I have made numerous references to cinema and film theory without, I hope, bending interactive fiction to fit old categories. The problem is that much theory has focused on the ´what´s of the phenomenon asking if there is a fundamental contradiction between ´interactivity´ and ´fiction´²³. In the following section I argue that such a question cannot be answered scientifically and that asking it hints that one has committed a ´substantialist fallacy´. While not underrating the important pioneer work done I suggest that energy be focused elsewhere, possibly on the ´why´s of interactive fiction.

Espen Aarseth and the rhetoric of revolution
Literary theorist Espen Aarseth (1997) makes the important point that attempts of nonlinear fiction are not tied closely to computer technology but can be found throughout the entire history of written literature. Secondarily he aims to cut through the ´hype´ of interactivity, seeing the term as highly ideological and as connoting revolutionary/utopian expectations that can never be fulfilled:

The industrial rhetoric produced concepts such as interactive newspapers, interactive video, interactive television, and even interactive houses, all implying that the role of the consumer had (or would very soon) change for the better. [´] To declare a system interactive is to endorse it with a magic power.´ (Aarseth, 1997: 48).

The industrial rhetoricians did not argue alone. From literary theory proponents of a postmodern aesthetic would advocate ´open texts´ that didn´t force the reader into narrow interpretations. Within media studies John Fiske spoke of the ´writerly text´ that

…requires us, the readers, to participate in the production of meaning and thus of our own subjectivities, it requires us to speak rather than be spoken and to subordinate the moment of production to the moment of reception.´ (Fiske, 1987:95).

Similarly Jacques Derrida made use of a society-theater metaphor to attack the constraints placed upon modern man and upon users of traditional fiction:

[The author-creator] lets representation represent him through representatives, directors or actors, enslaved interpreters´who´more or less directly represent the thought of the ´creator´. Interpretive slaves who faithfully execute the providential designs of the ´master´´ Finally, the theological [teleological?] stage comports a passive, seated public, a public of spectators, of consumers, of enjoyers. (quoted in Ritzer, 1996:597).

In this way interactivity has been promoted as a way to escape the shackles of determinist interpretations. Aarseth makes this point well but his more polemic attack on the term (he claims that the real fiction is that there exists such a thing as interactive fiction) is equally based upon on a far weaker argument.

The substantialist fallacy²⁴
That interactivity as a term has been stretched to the breaking point cannot be argued. Aarseth´s account of the infancy of the genre and his commitment to a strict typology is exemplary but the overall interest in mapping the relationship between interactivity and narrative is more problematic. Though later modifying this slightly Aarseth notes: ´To claim that there is no difference between games and narratives is to ignore essential qualities of both categories´ (Aarseth, 1997:5). In other words, the problems of traditional adventure games are fundamental. I hope to have shown that this is not so. The problem with Aarseth´s claim, of course, is the reference to ´essential qualities´. ´Games´ and ´narratives´ are terms, they are words applied to groups of phenomena. In other words: They are definitions, and definitions have no ´essential qualities´. Definitions are means by which we distinguish between objects; they do not exist externally of language. Thus we cannot discover the ´true´ definition of games. One cannot even argue with a definition made by someone else.²⁵ What Aarseth does is argue that A is not B on the grounds that they are considered different letters.

Aarseth´s problem is widespread. The idea that one can somehow prove that narrative and interactivity are separate and opposite categories is popular. Assumably this logical mistake springs from an unconscious conception of narrative as linear storytelling. This comparison has motivated game designer Walter Freitag to claim that:

´There’s a conflict between interactivity and storytelling: Most people imagine there’s a spectrum between conventional written stories on one side and total interactivity on the other. But I believe that what you really have are two safe havens separated by a pit of hell that can absorb endless amounts of time, skill, and resources.´ (quoted in Juul, 1996).

Now this claim is obviously true (if we follow common definitions). The combination of linear storytelling (based on conventional written stories) and interactivity is the problem of traditional interactive fiction. Jesper Juul agrees when he claims that:

´Computer games and narratives are very different phenomena. Two phenomena that fight each other. Two phenomena that you basically cannot have at the same time. Any interactive narrative or attempt at interactive storytelling is a zigzag between these two columns.´ (Juul, 1996).

But this doesn´t say much. In essence much effort has been put into proving the following: If narratives are linear stories written without the presence of the interactor then narrative and interactivity are opposite categories. This is truth by definition. As the fascinating but unpredictable series of events that may occur in online role-playing games imply, too much time has been spent in claiming something, which may be true but is not very important.

Reasons for interactive fiction
In 1996 film theorist Joseph Anderson published a book entitled ´The Reality of Illusion´. This book is an eloquent attack on most major film theories articulated before the mid-1980´s. It contains a controversial approach to film and states its case soberly although accessibly. It was widely ignored.

Perhaps this was due to its somewhat obscure subtitle ´An Ecological Approach to Cognitive Film Theory´. The word ecological is probably poorly chosen, as it tends to have different meanings within different academic fields. Furthermore other more fitting terms might have been used to indicate that this was an interdisciplinary approach to film, trying to span the gap between (socio)biology and film theory.

In the following I shall briefly outline Anderson´s theory and suggest how it may be adapted to include interactive fiction.

All higher animals play. This is the basic tenet of Anderson´s argument. If biological creatures almost as one exhibit a given behavior then there is a reason for this behavior. Standard natural selection (by most interpretetations) prescribes that a general trait, to have surfaced and subsisted, must entail (or have entailed in the past) an advantage to the survival and reproductive chances of the individual. Thus the ability and the inclination to play is helpful to our survival.²⁶

Drawing upon psychology Anderson notes how play entails cues that continuously remind the players that what they are doing is to be considered quite different from the simulated activity. Thus when children play at wrestling they should always consider it different from a fight for survival. Similarly when dogs play at fighting with their owners they should never bite with any strength. This is hardly controversial. Anderson then goes on to describe another activity that displays almost the exact same characteristics as children´s play: Movie going. This activity is so entrenched in ritualistic behavior serving the function of framing that we normally fail to realize it (Anderson, 1996:122). Since playing is the ´framing´ of a certain activity a probable evolutionary advantage would be the ability to rehearse various situations (dangerous or otherwise) in a safe setting. The only difference between the playing done by children and that of adults is that children´s games are about (´about´ in the external sense) learning senso-motorical skills that adults have already developed. Adults play different games but the basic reason remains the same. In Anderson´s words:

´A motion picture makes it possible for viewers, in a purely cognitive space, to test the efficacy of certain strategies and feel the exhilaration of victory, the relief of a ´close shave´, or the devastation of defeat without the risks that would attend that behavior in the real world.´ (Anderson, 1996:114).

One important point needs to be made. Inclinations and preferences developed in our evolutionary past are of course not to be considered ´good´. They are ´natural´, nothing more. Darwinism tells us little of morals (although it may explain why we have them) and does not proclaim, of course, that the preferences should be followed. The survival value of eating fresh fruit carries over to our day into a less desirable (even less reproductively wise) desire for candy.²⁷ The same holds true for playing. The inclination to play doesn’t filter out games that might in other ways be harmful (although other functions might).

It may well be that Anderson overstates his case when noting the similarities between different types of play but the chance for grouping phenomena according to something as concrete as their function presents itself as a welcome alternative to the far more arbitrary genre and media categories of much theory.

It is of course possible that we may prefer to actively participate in a certain kind of game whereas others lend themselves better to ´passive´ watching or listening. But these are questions for later studies.

At the very least Anderson´s theory makes a good foundation for the argument that the narratives we ourselves interact with serve a similar function to the ones we engage with through others, such as the movie protagonist. They are means of procuring skills and experience.

If one accepts the general direction of this argument, and it seems difficult not to, it is hard to see how such an insight might have been gained through a continued narrow focus on questions of the ´what´ type.²⁸

The relative strength of Anderson´s interdisciplinary approach provides food for thought. Within specific theories of interactive fiction perspectives combining computer science with humanist methods seem to hold the greatest explanatory value. Janet Murray´s ´Hamlet on the Holodeck´ mixes a solid knowledge of computer architecture with a well-founded literary perspective. The result is a number of valuable insights that surely rival the importance of others of a more structuralist and systematic bend such as Aarseth (1997) and Konzack (1999).²⁹

Section 4: Conclusions
Although we work within the limitations of hardware and human perception, design is primarily a matter of choice. So is the way we choose to approach a phenomenon scientifically. Within game design the storytellers of interactive fiction have chosen to adopt time-honored traditions of linear media. Many, it seems, have been content to be mere translators perhaps seeing contemporary movie making as the culmination of centuries of narrative experience. While we cannot blame them from a moral or even a commercial position ´ it may well be that the market was equally conservative ´ we may help to light up what apparently remains unseen. By pointing out the road not chosen we point to an approach to narrative that is truly novel and although novelty is not a virtue in itself deistic narration is a form of expression that is unique to computer technology. While not essentially superior this new form is surely worth the effort of experiment. Game designers may well benefit from thinking of themselves not as auteurs but as architects.

While one may make suggestions for improved designs, mystifying theories are perhaps a graver matter. There is a certain detectable arrogance to the all-encompassing systems of some humanist genre theory. A battle of words is joined apparently without even acknowledging the possibilities for simpler explanations; explanations that take less for granted and thus provide more solid foundations.

Joseph Anderson despairs of the turn film theory has taken and reminds us that it didn´t have to go this way. He mentions the lucid work of early film theorist Hugo Münsterberg claiming that:

´[Münsterberg] set film theory clearly on a path that would have confronted the basic questions about the nature and function of film in a direct and systematic way. Unfortunately, his was a path no one chose to follow.´ (Anderson, 1996:4).

Instead film theory turned essentialist, arguing for decades about the true nature of film, making claims that editing, darkened theatres or certain narrative styles were the qualities that set film apart from other forms of expression. This is the ´what´ approach. This is the effort to separate one phenomenon from others by the introduction of arbitrary criteria. It is the mistaking of means (establishing analytic definitions) for goals (saying something significant about film). And it is the same problem that plagues theories of interactive fiction. One searches for those special qualities that sets the medium apart instead of acknowledging similarities that may hint that what we are dealing with is a medium that can be explained within the framework of media history. With a focus on the ´how´s, the history and techniques, of interactive fiction one would be able to see that a description of the road chosen speaks little of the ´essential qualities´ of the medium and more about conventionalism, tradition and chance.

Humanist approaches have proven their value to a computer science that must understand the computer as both machine and medium. Communication science and aesthetic disciplines are valuable in the study of human-computer-interaction as they help provide the basics of sensible interface design. But they should not work alone, just as engineers should not. In recent years humanists have been accused of posing as experts in areas where they are amateurs. Most importantly Alan Sokal and Jean Brickmont´s ´Intellectual Impostures´ has documented the abuse of natural science in various obscure (but popular) literary theories (Sokal & Brickmont, 1997). While obviously not devastating to humanist scholarship as such the authors point to deep problems in specific theories that may well indicate problems of a more general scope. This exposure is sometimes interpreted as a warning against meddling in the affairs of others. It is taken as proof that one cannot be an expert in all fields. While this is obviously true the problems should not be contributed to too much interdisciplinarity. Rather it should be seen as a warning against lack of interest in the work of others. The theorists attacked by Sokal and Brickmont do not take the sciences they so readily use seriously ´ they find their own perspective vastly superior. This is arrogance, not interdisciplinarity.

A constructive approach to interactive fiction orients itself broadly and humbly. It constructs a solid foundation by taking seriously the history and technical issues of the genre. It even searches for deeper explanations as to the function of fiction and the biological basis for games. Then, with a secure understanding of the ´how´s and the ´why´s we may truly understand the limits and potentials of this thing called interactive fiction.

Works cited
[Dates are in dd.mm.year format and are included in references to newspaper articles and used to indicate versions of web pages.]

* Aarseth, Espen (1997). Cybertext ´ Perspectives on Ergodic Literature. London: The Johns Hopkins University Press.
* Anderson, Joseph D. (1996). The Reality of Illusion ´ An Ecological Approach to Cognitive Film theory. Carbondale: Southern Illinois University Press.
* Astruc, Alexandre (1948/1970). En ny avant-gardes f´dsel: Kameraet som pen. In: Monty, Ib & Piil, Morten (eds.). Se, det er Film – i klip. Copenhagen: Fremad.
* Bordwell, David & Thompson, Kristin (1993). Film Art ´ An Introduction. New York: McGraw-Hill.
* Bordwell, David (1985). Narration in the Fiction Film. London: Routledge.
* Branigan, Edward (1992). Narrative Comprehension and Film. London: Routledge.
* Crawford, Chris (1982/1997). The Art of Computer Game Design. http://members.xoom.com/kalid/art/art_of_cgd.pdf. [Download: 04.05.2000].
* Crawford, Chris (2000?). Computer Games are Dead. The Journal of Computer Game Design, vol. 9. http://www.erasmatazz.com/library/-
JCGD_Volume_9/Games_are_Dead.html. [Download: 25.06.2000].
* Dalum, Astrid & S´rensen, Finn (1996). Interactive Fiction ´ A Case Study. University of Roskilde. Thesis. http://www.centrum.dk/users/finnv/intfict.htm. [Download: 05.05.1998].
* Egenfeldt-Nielsen, Simon; Smith, Jonas Heide (2000). Den digitale Leg ´ om b´rn og computerspil. Copenhagen: Hans Reitzels Forlag.
* Fauth, Jurgen (1995). Poles in your Face: The Promises and Pitfalls of Hyperfiction. Mississipi Review Web. http://orca.st.usm.edu/mrw/mr/1995/06-jurge.html. [Download: 30.03.2000].
* Flyvbjerg, Bent (1991). Rationalitet og Magt. Copenhagen: Akademisk Forlag. [English version: Rationality and Power ´ Democracy in Practice, 1998].
* Frost-Olsen, Peter & Schmidt, Rigmor K. (1997). Den 5 alder eller Riven. Weekendavisen Berlingske. 21.11.1997.
* Gombrich, E. H. (1997). The Story of Art. London: Phaidon.
* Graetz, J. Martin (1981). The Origin of Spacewar. Creative Computing. No. 8. http://www.wheels.org/spacewar/creative/SpacewarOrigin.html. [Download: 24.04.1998].
* Hafner, Katie & Lyon, Matthew (1998). Where Wizards stay up Late ´ The Origins of the Internet. New York: Touchstone.
* Herz, J.C. (1997). Joystick Nation. London: Abacus.
* Hunter, William (1999-). The Dot Eaters ´ Classic Video Game History. http://www.emuunlim.com/doteaters/index.htm. [Download: 12.04.1999].
* Jackson, Steve (1985). Kaos-borgen. Copenhagen: Borgen. [English version: The Citadel of Chaos, 1983].
* Jensen, Jens. F. (1998). Interaktivitet og interaktive medier. In: Jensen, Jens. F. (ed.). Multimedier, Hypermedier, Interaktive Medier. Aalborg: Aalborg Universitetsforlag.
* Konzack, Lars (1999). Softwaregenrer. ´rhus: Aarhus Universitetsforlag.
* Larsen, Peter Harms (1995). Faktion som udtryksmiddel, Viborg: Amanda.
* Lassen, Nikolaj M. (1997). Knockout!. Weekendavisen Berlingske. 18.04.1997.
* Laurel, Brenda (1993). Computers as Theatre. Berkeley: Addison-Wesley.
* Lebling, P. David (1982?). Zork and the Future of Computerized Fantasy Simulations. Byte Magazine. http://www.lysator.liu.se/Infocom/Articles/byte.html. [Download: 22.02.2000].
* Montfort, Nicholas (1995). Interfacing with Computer Narratives ´ Literary Possibilities for Interactive Fiction. The University of Texas at Austin. Thesis. http://nickm.com/writing/bathesis/. [Download: 05.05.1998].
* Murray, Janet H. (1997). Hamlet on the Holodeck ´ The Future of Narrative in Cyberspace. Cambridge: The MIT Press.
* Ritzer, George (1996). Sociological Theory (fourth edition). London: McGraw-Hill.
* Rothstein, Edward (1983). Reading and Writing: Participatory Novels. The New York Times Book Review. 08.05.1983. http://www.lysator.liu.se/Infocom/Articles/nyt83.html. [Download: 22.02.2000].
* Sokal, Alan & Brickmont, Jean (1997). Intellectual Impostures. London: Profile Books.
* Therkelsen, Inge-Lene & Dalum, Lisa (1998). Blackout ´ Interaktivitet og fort´lling. University of Roskilde. Thesis.
* Thompson, Kristin & Bordwell, David (1994). Film History ´ An Introduction. New York: McGraw-Hill.
* Turkle, Sherry (1995). Life on the Screen ´ Identity in the Age of the Internet. London: Phoenix.
* Unknown (2000). The Collossal Cave Adventure Page. http://people.delphi.com/rickadams/adventure/a_history.html. [Download: 01.06.2000].
* Valeur, Michael (1998). ´Blackout´ ´ erfaringer omkring arbejdet med interaktiv manuskriptskrivning. University of Roskilde. Thesis.
* Wilson, Johnny (1991). The Rise and Fall of Infocom. Computer Gaming World. http://www.lysator.liu.se/adventure/Infocom/Articles/rise.html. [Download: 22.02.2000].

1. The deliberate focus on form and the rhetoric of revolution, however, have been dominant for much longer (Gombrich, 1997:557). [↩]
2. At the 1967 World Expo in Montreal, Canada the Czech Pavilion featured a movie that allowed audience members to make choices at crucial moments in the plot (Laurel, 1993:53) [↩]
3. This definition will serve my purpose although there may be media forms that invite but don´t require audience participation (computer game demos, various forms of hypertext etc.) [↩]
4. Conceptualizing the interactor´s activities proved an insurmountable challenge to game designers at the legendary production company Infocom. In a newsletter the company mused: ´Back when they were simply adventure games, you played them. But does one play an interactive fiction? Or do you read it? Some Californians here at Infocom have suggested that you ‘do’ interactive fiction.´ (quoted in Montfort, 1995) [↩]
5. For the history of computer games in general see Herz, 1997; Hunter, 2000; Egenfeldt-Nielsen & Smith, 2000:28-48. For a discussion of computer game genres see Konzack, 2000; Egenfeldt-Nielsen & Smith, 2000:28-32 [↩]
6. The exploits of the early ´hackers´ are well documented. See for example Hafner & Lyon, 1998. For an entertaining description of the work with the 1962 game Spacewar see Graetz, 1981 [↩]
7. They may also be compared to the use of voice-over in narrative film; a technique often seen as an artistic surrender since it signals a lack of comfort and ability with the visualization of narrative [↩]
8. Of course one may argue that text adventures make use of an ´implied avatar´ as they address a ´you´ that is both interactor and avatar. Perhaps any real difference should be sought in terms of spatial orientation rather than identification [↩]
9. A few other games had used this interface before but Maniac Mansion was the first game to combine it with a technical and artistic expertise that could challenge Sierra´s supremacy. Besides benefiting from the highly usable interface the game was a clever genre pastiche that allowed the player to switch between three avatars with different abilities [↩]
10. As the computer game industry is after all based on numbers and digits it becomes close to ironic that reliable statistics on sales and sizes are exceptionally hard to come by. Although such charts should be regarded with the deepest skepticism one computer magazine lists Myst as the best-selling (PC) game of the period 1993-1998 with 3,8 million copies sold and with Microsoft Flight Simulator as a distant second with 2,4 million copies sold (one should particularly note that this does not necessarily translate into exorbitant revenues as games may be sold at discounts or bundled with hardware). See http://www.barracuda-gssm.com/timelapse/specials/bestgames.htm [↩]
11. One cannot scientifically rate one form of expression as essentially better than another and indeed I shall not attempt to do so. But I will show how the tension between narrative and the interactor´s freedom of choice is not fundamental but merely a consequence of a certain style of design [↩]
12. More than one computer game analysis goes too far on the basis of an arbitrarily chosen case (eg. Valeur, 1998; Therkelsen & Dalum, 1998; Dalum & Finn, 1996) [↩]
13. A game engine is the ´narrative programming language´ through which the game action is presented. It schematically prescribes the qualities and relationships of all objects as well as the representation of any game element. Although this metaphor may certainly be stretched too far the engine may be compared to a system of musical notation. When the system is in place the programmer/designer places the notes and the interactor plays the piece (the last phase becomes meaningless if the game is highly nonlinear) [↩]
14. We need not concern ourselves with whether or not this is normal human behavior. The issue here is adventure game designers and the way they have shaped their genre [↩]
15. A curious exception to this is Leisure Suit Larry 5 (Sierra, 1991) where the story would not be hindered by the interactor’s failure to solve major puzzles. Instead the game could be finished quickly and the climax would then be toned down, as the objectives had not been met. Assumably this was a disappointing experience for many players (but no systematic evaluation has been published) [↩]
16. One may also note how crime-solving themes are well suited to combining complex stories with a fair amount of interactivity as the action centers on piecing together a story that has already happened and thus need not be restricted by concessions to freedom. See also Therkelsen & Dalum, 1998:55 [↩]
17. A range of techniques to minimize this problem have been developed (particularly principles of looping and guiding) but they are mostly utilized as damage control and can never eliminate the problem entirely [↩]
18. [Some adventure games include the possibility of critical failure. For instance the avatar may die or be otherwise incapacitated at crucial moments in the plot. Most games, however, take such occurrences lightly as the interactor is typically just set back to before the ´wrong´ decision. In the case of critical failure it is rarely assumed that the interactor will be prepared to play again from the beginning although this is a common conception in other situations (a common misconception most likely as the principle of lazy bytes preclude that there be much new of interest to explore in subsequent attempts) [↩]
19. While programming may be object-oriented I use the term to describe any situation that evolve according to properties of discrete objects of any type. Thus (unless one believes in divine determination) any real-life situation is object-oriented while situations in virtual worlds may be governed by forces other than those apparently vested in the objects involved [↩]
20. It should be noted that MUDs and MOOs (Turkle, 1995) share many of the same characteristics, although on a smaller scale. A slightly different approach was attempted by Lucasarts in their Habitat project (Murray, 1997:266) [↩]
21. Critics have somewhat appropriately renamed the game Ultima Outline [↩]
22. One can probably draw on the experience of traditional role-players who have struggled with similar problems. It seems likely that many of the techniques that human game-masters have developed can be built into computer systems [↩]
23. The (wrong) arguments most commonly leveled against the possibilities of interactive fiction are clearly summarized in Fauth, 1995 [↩]
24. This is not strictly a fallacy in the philosophical sense where the term typically refers to conclusions that are not justified by the premises of an argument. I use the term in the looser sense to relate it to ´the naturalist´ and ´the intentional´ fallacy. [↩]
25. One may of course question the usability of a given definition and advocate the use of another. Should one define ´art´ as ´everything human´ this definition would have little analytical value, but it would not be wrong [↩]
26. I shall only briefly outline the theory here. There may be some objections to the argument but most are dealt with in Anderson´s book. I will use the present tense only although the correct form would at all times be “is or have been in the evolutionary environment” [↩]
27. The causality, of course, must be reversed. Candy would not exist if our ancestors and we hadn´t been fond of fruit [↩]
28. The focus on causes is also present in a farsighted 1982 account by legendary game designer Chris Crawford (Crawford, 1982/1997) [↩]
29. This is not to say that being systematic is the problem. Quite the contrary: Without such rigor it would be impossible to discover the problems and faults in these accounts [↩]