The Hyperfiction Writing Experience

Chapter Three: The Hyperfiction Writing Experience


Following previous chapters, which examined different forms of hyperfiction for insights into the process of producing my own work of hyperfiction, in addition to the unique affordances for writing that the hypertext form enables, this chapter will concentrate on putting these lessons into practice. First will be examined my experience of writing the hyperfiction, and the challenge it represented in transporting the hypertext to the internet. This will also take in the difficulty of choosing a suitable vehicle or setting for the story, and of incorporating the various different narrative techniques that distinguish hyperfiction from standard linear fiction. Next will be examined the tools for creating and hosting hyperfiction works, and exploring their strengths and limitations, including my methodologies in this project. Then a brief outline of the hypertext’s plot will be given, and how it satisfies the conditions of a successful hyperfiction.

Challenges in Writing the Hypertext

Deciding on the type of hyperfiction to write was difficult at first; would it be an attempt to convey a shattered consciousness in the post-modern style of Afternoon, or a diverting and fun exercise in text-adventures, like the fan-fiction games that populate the web? In the end it was decided to write a work of literary fiction – this being the genre I am most familiar with, and in keeping with the style of the different types of hyperfiction analysed in the previous chapters.

This decision to write in a literary-fiction style proved quite hard, as not every story could necessarily contain the disparate elements normally found in hyperfiction. I settled on writing two or more parallel texts, as this would employ the narrative features I found most interesting: non-linearity, multi-perspectives and synchronicity, with interactivity being incorporated by the use of hyperlinks to connect different sections and narrators.

A hypertext that proved very instrumental in helping to decide what form the hyperfiction would take was A Long Wild Smile by Jeff Parker. This was a parallel text from the dual perspectives of a couple in a relationship, and included many of the effects that I wanted to reproduce in my text, such as miscommunication, one character mishearing another and picking something up wrong, as well as the large differences between what one character and another think, which is part of multiple perspectives. “In the hypertext short story A Long Wild Smile by Jeff Parker… the reader can move back and forth between two narrative strands: one narrated by a woman’s fiancé and another by her lover… Some links make very quick incursions into the other perspective, highlighting micro-level conflicts” (Ryan).

This structure of interconnecting and parallel narratives is extremely filmic, an early example being Rashomon (1950) directed by Akira Kurosawa, but also the films of Robert Altman, most noticeably Nashville (1975) and Short Cuts (1993). In these films a large event occurs which throws together a diverse array of characters: in Short Cuts there is a fleet of helicopters spraying for insects at the beginning, then an earthquake at the end, and in Nashville with a series of country music concerts, and also a motorway pile-up. This motorway scene would prove the most fruitful, and also presenting me with a setting for my eventual hyperfiction.

It was decided to use a motorway as a setting for the hyperfiction, as it solved the immediate problem of a convincing setting shared by the different characters who would be travelling in the same direction and so witness similar objects or events, which allowed for multiple perspectives. For example an accident or pile-up if the text were very dramatic, or for something more modest, a roadside ad or a toll plaza, that would be perceived differently by each character at the same time

The motorway setting also helped to solve the difficulty of how to organise the individual lexias. In the case of this hypertext, there would be four characters with individual narratives running parallel to one another. The setting of a motorway would facilitate this, the narrative strands progressing like four separate vehicles moving in tandem. Also by the inclusion of a toll bridge in the text, the different narratives could literally and structurally be made to converge.

The complication of how to achieve simultaneity was also solved by splitting the narratives into quite short sections, that way the user could jump to different parallel narratives without too much difficulty by using the hyperlinks. Each narrative was broken into six short sections of 170-240 words, which corresponded to roughly one or two minutes in the lives of the characters. This made organising the narratives, and having an observed object or event appear uniformly throughout different sections, relatively manageable.

In the ordering of the different lexias, it was initially considered to let the different speakers go unnamed, it falling to the reader to observe the differences in the speech patterns of the characters in order to deduct which narrator was speaking. The narrators were written in such a way that distinguishing between them should be straight-forward, eg one character speaks on his phone for the entirety of the hypertext, another has a limited and repetitive vocabulary because he is eleven years old. In the end however this was thought to be overkill for the reader already coming to terms with a split narrative, in addition to obscuring the more important aspects of the hyperfiction, and so was removed. Instead one of the best examples of a split narrative, Faulkner’s AILD, served as a template for this hyperfiction, with its inclusion of the narrator’s name as the heading to each chapter.

How to navigate through the text presented a problem, and the question of where exactly to locate the hyperlinks. The idea was briefly considered of attaching a hyperlink to a word within every text, such as an object like a car or sign post, that is observed by multiple narrators. This would succeed in highlighting the importance of the object, as it would then be coloured blue, the universal indicator of a hyperlink, as in the proto-hypertext HOL analysed in chapter two. Also this approach would directly allow the user to cycle between the diverging reactions to the same object throughout the different narratives in an example of multi-perspectives.

However this was thought to be too complicated and contrived – crowbarring mentions to the same object in every lexia would threaten any realism the text may have, as well as impose an overly rigid structure on the hypertext narrative, similar to the Joyce hypertext Afternoon. Instead cues would be taken from the more open structure of Patchwork Girl, and a dramatis personae at the beginning of my hypertext offered readers the choice of all four characters, while a range of hyperlink buttons at the bottom of the page served as a means to navigate through the hypertext. These buttons or hyperlinks functioned in a similar way to a joystick or game’s control pad: forward, for advancing in a character’s narrative; back, to reverse the direction through the character’s lexias; and a range of different names, which allow the user to shift to the parallel lexias of the characters, in a movement similar to moving left or right in a game.

Saying this, I did allow a series of hyperlinks to remain within the text. In this section, various different characters either adjust or turn on their radios, with the same song registering to each one differently; one character for example, interprets the lyrics to a Beyoncé song in different ways to the passenger travelling beside her, or those in another car. This underlines the multiple perspectives in the reactions to the song, as well as the action of changing radio stations being very similar; in essence, by using the hyperlink, the reader is skipping to the next “station”.

On the subject of multi-perspectives in the hypertext, because the same action and events feature in the different texts although interpreted differently, it was very important to avoid duplicating the same dialogue within each narrative, as doing so would make reading very tedious for the user. Instead in my hypertext there would be one character who is speaking, and another who is either ignoring the speaker or silently criticising them. This disconnect between the narrators provided ample opportunity for miscommunication, which the feature of multi-perspectives is very suited to. An example of proto-hyperfiction that informed this choice was Faulkner’s AILD, where dialogue is repeated between two different narrators but with slight variation, suggesting to the reader that one of the narrators is unreliable. This effect was something I wanted to reproduce; however instead of implying that one narrator was somehow more authoritative than the other, I wanted to include the duplication of dialogue by just one narrator, as he repeats the same story to two different people during separate phone conversations. I felt that the small changes between his first and second telling of the story would signify his unreliability but also chart the embellishment of the story, and that the humour or obnoxiousness of the character would offset the reader’s growing boredom.

While the user is encouraged to read the hypertext any way they wish, some confusion would be registered were they to arrive out-of-sequence to a later section in a character’s narrative; this is an area where the project’s hypertext can be said to be unsuccessful. While hypertexts such as Patchwork Girl features self-contained, non-linear lexias that work to disrupt the sequentiality of the narrative overall, the lexias in my hyperfiction cannot be said to be independent quite in this way, as each lexia forms part of a six-section structure, and still requires the user to read them in sequential order. However nonsequentiality occurs in my hypertext through the non-linear, parallel narratives.

The effort required to make each individual lexia self-contained and complete would have been significant, and similar to the heavy-handed insertion of hyperlinks within each section that was dismissed earlier, may have produced contrived and unnatural results. Also this approach would have been incompatible with the structure of parallel texts, which requires the texts to be read in quick succession to make the most of the divergent and contradictory points of view. While the hypertext differs from Patchwork Girl, it does resemble Victory Garden, the other classic hypertext examined in the previous chapter. Victory Garden has an overarching narrative spread across different lexias, added to this a branching structure that results in contradictory events, and my hyperfiction text is modelled on this.

One foreseen complication was that users will be unfamiliar with the mode and confused about how to navigate the hypertext. To reduce any confusion, tables of instruction were included, explaining the function of the hyperlinks and how they are to be used to advance through the different parallel narratives. It was also stressed that the user can read the text any way they wish, and that if one character is found to be boring then users are permitted to skip their section entirely.

An area where my hypertext diverges from Victory Garden, and also Afternoon, is in the ending. Both these established hyperfictions carry alternative endings within their structure, whichever ending is arrived at depending entirely on the series of hyperlinks that the user chooses beforehand. For my intended hypertext, recreating such an effect would have required a very large commitment of time, and also was not germane to the narrative. It may be argued that alternative endings represent an important use of interactivity, one of the features of hyperfiction. However I feel that interactivity is already well-served with the large number of hyperlinks and the diverging routes the reader may take through the text.

Challenging the linear style of endings, and taking Joyce’s Afternoon as a source of inspiration, was the question how to conclude the hypertext. While my hypertext was not fully cyclical, it was a very important part of the overall structure for the user to pass back and forth between the sections and characters for as long as was necessary, without the constraints that Afternoon placed on its readers. Therefore in this hypertext, the limit of six sections per character meant that the reader could only proceed so far before changing to a different narrative strand.

The arrival at the ending to a section is signified by the removal of the “Forward” button, however the other links to parallel narratives remain, allowing the reader to explore the parallel narratives, in addition to returning to the narrator’s previous sections. This is because it did not seem logical to apply an ending to something that is intended to be open and interactive. The same then as in Afternoon, it is more straight-forward to let the point at which the reader loses interest or drops off represent the point where closure is arrived at. Also by including a definite end-point, if users are given the “suspect quality” according to Michael Joyce that represents the end, it rules out the possibility of the user re-reading the text, or back-tracking, which like in Afternoon is extremely important.

 Methodology: Building the Hypertext

There are two web tools used for building hyperfiction or interactive fiction available online. The first is Inklewriter, created in 2011 by two games developers. I found Inklewriter to be the simpler of the two, but also the most basic. It is primarily used for interactive fiction where two or more options are presented to the user, and they click on whichever option appeals to them. The strength of Inklewriter is its simplicity and ease of use, though a weakness is the tool’s lack of functionality. For instance to build my hypertext using Inklewriter, the structure of the hypertext would have to be drastically altered. However one feature of hyperfiction that the Inklewriter software is suitable for is with interactive, alternate endings, as the narrative paths fork based on the choices that the reader makes.

The second web tool is Twine. This was developed in 2009 by Chris Kilmas, and carries far greater functionality than Inklewriter. For example it is relatively easy to create new text passages and to link them using the special syntax. It is not limited by the same restrictions on the narrative structure as in Inklewriter, instead there are as many hyperlinked passages as are needed. What else was extremely useful was the way in which Twine displays the text passages, resembling a virtual notice board with each text passage presented as a moveable post-it-note. This feature proved very valuable in helping to visualise the order of the story, and the way in which readers would navigate the text. Figure 1 is an image of another Twine story’s plan, which gives an impression of how intricate the links between the sections can become. This is taken from a Tumblr site named Twine Garden, which displays screenshots of Twine plans based on their complexity and strange beauty, rather than any storytelling merit they might have; my hypertext map was not this intricate, seen later. It is also interesting to note how the below plan and other plans of Twine hypertexts resembles the snaking paths of Victory Garden seen in the previous post.

Figure 1

Figure 1: Example of an intricate Twine plan (Source: Twine Garden).[1]

The Twine plan in Figure 2 represents the actual plan of my hypertext, entitled Road Works. The text moves in a left-to-right and top-to-bottom direction; the arrows representing what direction the user takes can be clearly seen. The four large columns represent the parallel narratives. The arrows between the columns show that they are connected or linked; it is clear that the arrows are double-sided, which represents the backwards and forwards movement between the different passages. There are 28 passages in total, four introductory passages, and 24 containing the hypertext. I also include in Figure 3 below the statistics relating to my hyperfiction that are provided by the Twine system.

Figure 2

Figure 2: Road Works Twine map (Source: Twine).

Figure 3

Figure 3: Road Works Twine statistics (Source: Twine).

Twine was the preferred option because of its clean presentation and functionality. The difference between the HTML located in the individual passages/post-it-notes, and what appears in the web browser, can be seen in Figure 4. The hyperfiction published to the web using Twine was very attractive, with large-sized type and a clean, minimal lay-out, which made it easy for the readers to become accustomed to and follow the text. Also Twine could be used together with HTML tags and CSS to increase its functionality further, and for a broader range of design options.

Also significant was the large support-base that came with Twine, which included forums and other resources. There was a range of different websites specially designed to host Twine stories, eg the website This demonstrated there was an audience already interested in hyperfiction texts, and with publishing my hyperfiction online through Twine and it would be straightforward attracting a large enough number of users in order to analyse how they navigated the hypertext. For example, there was a Twine twitter page that automatically posted links to new hypertexts, which had close to 1000 followers. By publishing using Twine my hypertext could be accessed by many of these followers.

Figure 4

Figure 4: Twine passage (left) and browser display (right) (Source: Twine).

Twine proved extremely useful in helping to plan and construct my hypertext, also the display when published to the web was exactly how I imagined the hyperfiction to look. However I encountered major obstacles in trying to use Twine together with Google Analytics (GA). Using GA to track how users navigate through the hypertext was an important part of the project, and is outlined in Chapter Four. But due to the way that Twine constructs the different passages making up a hypertext, not as individual HTML pages as GA requires, but as cascading sections within one single HTML file, the tools were not compatible. Searching on a Twine forum, I became aware of the possibility of setting GA event tracking tags on each of the hyperlinks in Twine. However these were instructions issued for Twine version 1.0, which was dramatically overhauled for version 2.0, including the syntax. This made the instructions on how to use GA in connection with Twine out-of-date and unworkable.

In the end it was necessary to relinquish these gains to allow the use of GA. I decided to export the Twine file to HTML, and using HTML files together with CSS I constructed a hypertext modelled on the Twine version. It was possible to segment the individual sections into separate HTML files, which would be displayed as different pages. Following this, I was also able to include GA tags in addition to Event Tracking tags on the hyperlinks, meaning it was possible to monitor what pages were viewed in what order but also which hyperlinks were clicked on. Uploading the 28 HTML and one CSS file to Google Drive, I was able to share the files publicly and publish them to the web. The HTML code and how it is displayed on the Web Browser can be seen in Figure 5.

Figure 5

Figure 5: HTML code (left) with page displayed in Web Browser (right).

Without the established audience of hyperfiction readers that came with using Twine, in order for my hypertext to rack up enough views it was necessary to mount a real effort through social media in promoting the hypertext. I am relieved to report that it was a success, receiving close to 87 users and 130 user sessions at the time of writing.

Overview of Completed Hypertext Work

The user is presented with four characters to choose from at the page dramatis personae, with no restriction placed on them once they decide to read a character’s narrative, instead it being encouraged to shift between narrators while progressing through the hypertext, in order to better appreciate the parallel narratives. These different narrators come in a particular order in the dramatis personae: the character of Lawrence is first, he is an eleven-year-old boy travelling in a car with his father; next is Lawrence’s father, here called Dad, and they share the same car. Then there is Barry, a passenger in another car behind Lawrence and Dad, which is being driven by Barry’s girlfriend, Debra. It should be possible for the user to read each of the four characters’ first sections in order, and for it to make sense.

Placing the hyperfiction in two separate cars manages to both contain the action in a shared setting, and convey the action along so as to allow the characters to observe new objects and places. This was partly inspired by the Ulysses episode “Wandering Rocks”, which features the movement of the Vice-Regal cavalcade in the final section, and facilitates multi-perspectives. Also there is an earlier episode of Ulysses with a similar setting, the episode “Hades” which takes place largely in a carriage, and follows the central character Bloom together with three friends on their way to Glasnevin cemetery, the narration jumping between the internal monologues of each of the four passengers.

The setting was also influenced by other modernist texts such as Mrs Dalloway, as well as Faulkner’s AILD, which although substituting a horse-drawn cart for an automobile, finds the various characters engaging in multi-perspectives. This and other examples highlight the disconnect between what one character and another are thinking, which is something I wanted to exploit in my hypertext.

For instance, in the first part of Lawrence’s section, he observes an image of a scantily-clad woman as part of a road-side advertisement, and remains occupied by this and other related thoughts for the duration of the journey. Meanwhile Dad struggles to effectively communicate with Lawrence, first to chastise him for not being attentive in school, although Lawrence’s thoughts are elsewhere during this as well, and then to impart other fatherly knowledge on Lawrence.

Multi-perspectives come into play at moments when a lorry is observed differently by all four characters, or when the car comes to a toll booth, the interaction between the father and the toll attendant being viewed differently for Lawrence and Dad. As the action of the sections unfolds parallel to each other, the effect of synchronicity is produced. In the second section, Dad says to Lawrence “I hope you’re listening Lawrence?” In Lawrence’s section the question registers in his thoughts at the same time but in a vastly different way: “Dad asks if I’m listening. I tell him “Yeah”. I wish I’d brought my earphones.” This produces a synchronous call-and-response style of narrative that further underlines the breakdown in communication between the two, as Lawrence is patently lying to his dad about having listened to him speak.

Miscommunication forms an important theme of the hypertext, seen in these two sections with the Lawrence character day-dreaming while his father is talking to him, and the dad finding himself unable to reveal an important piece of advice to his son. Like in the hypertext Afternoon, how form mirrors content with the restrictive structure that withholds information both from the narrator and the reader, here form mirroring content can be seen in how the action of hyperlinking to a different section resembles the avoidance of meaningful communication. None of the characters in this hypertext can be said to have strong, healthy relationships.

For example there is virtually no communication between the final two characters, Barry and Debra. Barry is preoccupied with his own act of narration in this section, recounting to two friends over the phone the antics he engaged in the previous night. He takes no time out during the hypertext to speak to his partner Debra busy driving the car, instead generally ignoring her except to issue instructions or to comment on her actions over the phone in a way that Debra does not appreciate. The parallel texts allow the user to see how radically different the characters’ views are.

The conscious decision was made to have Barry repeat large amounts of his story to the second caller. Originally this was intended to bore or turn the reader off after they suffered through the phone conversation already before. However taking inspiration from a scene in Faulkner’s AILD mentioned earlier, where competing viewpoints of the same event suggest the unreliability of one of the narrators, I felt it would be interesting to watch the evolution of a story told by the same narrator. Like a one-man game of Chinese Whispers, Barry’s retelling of his story contains slight deviations and embellishment from the earlier version.

The final character of Debra in her narration carries a disdain for her boyfriend Barry in response to the insensitive way that he treats her. It is significant that like Lawrence’s narration earlier, Debra remains largely silent, only getting the opportunity to speak in the final section where she communicates with the toll attendant. Similar to the Lawrence character, and in total contrast to her partner, Debra’s narration takes place internally. And like Lawrence, this allows Debra to entertain certain thoughts and fantasies, in this case involving Barry in an accident.

It is significant that at no point do the characters in separate cars meet, although they share observations of the same objects, and Debra at one point speculates as to the identity of Lawrence and Dad’s silhouettes. This is in contrast to Barry, who is quite a poor narrator, not being very observant. This decision was made as it was seen to be very unrealistic having their narratives collide in an overt way, unless they were to crash into each other, which did not lend itself to the story overall.


In summary, the hypertext written as part of this thesis, entitled Road Works is successful in its use of the different hyperfiction techniques of interactivity, non-linearity, synchronicity and multiple perspectives. In its setting and content the text also gestures towards other examples of hyperfiction and early proto-hyperfiction. The web tools used to construct hyperfiction are critiqued, with Twine being used for a time for its superior visualisation of the individual lexias or text that makes up the hyperfiction. But owing to the software’s incompatibility with GA tags, necessary for analysing the way that users read the hyperfiction explored in the next chapter, the Twine tool was discarded and the hypertext was published directly to the web instead. However it is important to credit the Twine tool with supplying a very attractive hyperfiction design that my eventual hypertext would adopt. Lastly I outlined the four plots that make up the hyperfiction, how they converge and how they differ. Miscommunication is shown to be a central concern in the hypertext, with none of the characters enjoying healthy dialogue with one another.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s