Conversational flow:the deal-breaker for interactive fiction?
The missing link between chatbot and reader
Remember Goosebumps?
Or any of those pick your own adventure style novels that allow you to choose what happens next to your favourite character?
Imagine you could do that while logged on to your computer, in a live chat on your Facebook Messenger app with a computer program implementing your choice and responding to you in real time and you´ve got the gist of what a chatbot is.
Referred to respectively as virtual assistants or conversational interface (according to the platform they sit on), bots have primarily gained recognition in recent times for effectively streamlining customer service, ranging from acting as personal shoppers, assisting individuals with online banking queries to curating news articles for them .
One of the key reasons as to why bots seem to have gained such a high level of popularity is due to people believing machines to be less judgemental, especially when confronted with highly personal information as opposed to an actual person.
In recent times, digital storytellers have tried to flip the switch by turning this concept on its head- rather than providing a service that encourages the user to become intimate, the idea is to build a fully immersive environment that manages to be intimate in its own right.
Cue “storybots”, bots supposed to enhance the reading experience via the principle of one-to-one chat.
“LOST” (2016), produced by Sylvaine Souklaye, founder of Copenhagen based start-up Synthesia EXP , is one example as to how this approach has been implemented in the past couple of years.
Written by French author Samuel Petit, the interactive novel allows its readers to assume the role of the protagonist,(“John Deadle”), who seems to be stuck in some kind of metaphysical dreamscape, a world reminiscent of Nietzsche´s abyss where the main character can only escape said plane of existence after discovering the meaning of his or her “life, the universe and everything”, a quest ironically parallel to that of Arthur Dent in the iconic sci-fi book series, “A Hitchhiker´s Guide to the Galaxy” .
It´s an intriguing concept, trying to make existentialist literature à la Camus and Sartre more interactive while aiming “not just to entertain but to create a memory” (Milliken,2016), as Souklaye claimed in a VICE interview at the time when asked about his intention when developing LOST.
Yet it´s a promise that remains largely unfulfilled when one stoops to consider the way the textual flow of the piece is conceived.
You should make it look like a conversation. Don´t throw out a 50 word paragraph and expect your user to read them all (Monteon, 2016).
In an article titled “6 User Experience Tips for Designing your best Chatbot” for Medium publication Chatbot Magazine, UX designer Joel Monteon maintains that one of the main criteria to consider when building a bot is to enable the user to take in the information displayed at a leisurely pace.
This entails not overloading the screen with text bubbles and images (ibid.) and making use of short pauses between posting information as to not overwhelm the user.
“Allow users to finish one paragraph while the bot “writes” the next one, and display this loading speech bubble on their screens as the bot “types” (ibid.) is how he phrases it exactly.
“LOST” at times fails to take this into consideration and thereby risks losing the interest of its audience from the get-go as they are bombarded with long lines of text, interspersed with only a few calls to action where the reader can make decisions regarding the flow of the narrative.
As such, the experience can sometimes feel like a one-way conversation .
When faced with quotes such as “I wander in an endless perspective where I seem to be the vanishing point” or “[…] this endless darkness in which he can detect the many shades makes him a spectator of the unspeakable architecture of his most deeply buried fears”, the reader requires somewhat more breathing space in order to not only comprehend take in this form of allegorical text and properly savour it.
This is where “Humani: Jessie´s Story” comes in, which describes the day of a 20-something year old girl in LA as she navigates the pitfalls of first job interviews, disastrous dates and moving apartments. The bot, which functions more like a game, contains over 3,000 lines of dialogue and nine different storylines, one of which can take up to 25–30 hours to complete (MacPherson, 2016).
The response time is varied, allowing users to take a break and return to the experience should they choose too- “LOST” does not give you that option unless you voluntarily decide to quit the chat window or are at a crossroads where you need to make a choice on whether to go forward or backward or, decide whether you want to pick up a diary or not.
Said breaks in time make the situations themselves appear more natural as well, allowing you to immerse yourself to a greater extent as it seems plausible one cannot continue chatting while on a date, for instance, as “Jessi” maintains at one point.
In some ways, it´s a missed opportunity for “LOST” as the experience could have been enhanced not only by longer pauses between the textual components but also breaks in the narrative which would have been conducive in putting a reader into the more ruminative mind set needed to “investigate the link between emotions and reflection” (2016), as Souklaye describes a project (“Le déserteur”) similar in intentionality to the one mentioned above.
When writing for a storytelling chatbot, the entire story needs to put forward through the actual conversation taking place between the characters (Sachdeva, 2017).
This quote, sourced from a Medium post focusing on the art of telling stories via chatbots, sums up in a nutshell the second point I´d like to raise with regards to textual flow in chatbot stories- the importance of allowing the user to interact with the characters in question by voicing their own opinions or questioning the actions of the protagonist (s).
“Jessie´s story” makes accommodations for both, for example by allowing a potential user to select her dress for a dinner engagement from a carousel displayed inside the chat window to giving one the choice of advising her not to tell another character she rides an elevator with that they have a wad of gum stuck on their backside.
What´s especially striking about this narrative game is that chatting with “Jessie” feels natural at all stages- while playing the narrative games, I tried to trip her up occasionally by giving monosyllabic answers to questions or making use of emojis rather than responding in the form of clear sentences.
This was the result of one of those interactions:
Despite all efforts to the contrary, I managed to become more engaged with the experience as I gradually started to view the responses to my comments as similar to those “a friend would send, further increasing my sense of Jessie as a real person” (MacPherson, 2016).
Chatbot responses to user messages should be smart enough for user to continue the conversation (Surmenok, 2016).
One might argue that “Jessie´s Game” is a narrative game rather than a literary experience and, as such, lends itself to a higher level of interaction yet there are other instances in which a short story converted into bot format is built on a more conversational level, “A Tendency To Spill” being foremost among them.
The sci-fi story featuring Romy, a human brought up by robots, has a similar minimalistic interface to LOST and does not even employ any form of graphic imagery- it´s basically just a textbox via which you can converse.
Disregarding stylistic concerns, the story world in which “A Tendency To Spill” is set appears richer due to the fact that “Romy” asks some very profound questions in the course of the literary experience, allowing the user to respond in a similar fashion.
“I think my parents are supposed to love me-how can I tell?”
“Are all memories bad?”
“What makes someone a friend?”
These are just a couple of examples for question a user can expect when engaging with “Romy”- what is striking, however, is that the overall experience is still satisfying despite being based on a retrieval based system.
Despite at times feeling very scripted, which is the case when the user´s answers to the questions asked above are at times disregarded in favour of asking the next batch of questions, the idea of convincing a 13 year old girl who grew up thinking she is a synthetic form of being that the modern world is more than an outlet for hatred, crime and chaos (Grian, n.d.) is intriguing from the onset.
If “LOST” had made use of the occasional open-ended question, (merely one or two!), within its narrative, it could have served as an effective distraction from the fact that certain plot points appear incomprehensible to the average user, thereby disrupting the textual flow as the user tries to make sense of what he or she is reading .
One specific instance where I felt this to be the case is when the narrator of the story, John Beadle, starts writing in a diary, with those diary entries itself being inserted into the regular speech bubbles without any change in design (ex.colour, font) to mark the transition, something that feels jarring and disorienting at times.
On the other hand, the strictly linear and narrative arc which this story of self-discovery follows does have its perks as it allows the bot to be focused on the subject matter at hand- introducing the reader to a new way of experiencing a short story. By not deviating and offering multiple points of entry as “Jessie´s Story” does, the structure of the experience remains relatively easy to grasp- it´s a matter of falling down the rabbit hole and seeing where it takes you.
A focused bot that does a few things right is more useful than ones that barely breach the surface (Krumhausen, 2016)
Audience is key when it comes to analysing whether the conversational flow in “LOST” is a success in comparison to “Jessie´s Story”- despite the potential of immersion both bots offer via use of sophisticated imagery, use of colloquialisms and emojis, as mentioned previously in this article, there are also risks to this approach- namely, of losing one´s users in a sheer wealth of detail and, more importantly, forcing them into a trajectory that is not of their own choosing.
At the first glance, this interpretation may seem an illogical conclusion, seeing as I argued at the onset of this article that choice is more limited in “LOST” due to a limited number of calls to action, grating the user scant opportunity to drive the narrative forward.
Yet “Jessie´s Story” is aptly named- it´s truly her story, with the user assimilating to the likes, dislikes, wants and desires of this character driven bot. From designing a mantra for her day to guessing the country her date hails from, these are instances which carry the potential of “not being the information the user wants, making the user likely to get frustrated and leave” (Vishwanathan, 2017).
Personally, despite being fascinated with the potential of being able to have my own input to the storyline, I at times grew exasperated with constant calls to my attention, especially when I´d just gotten to enjoy a specific plot point.
(This, however, is a matter of personal preference and as such subject to scrutiny.)
People love to play with chatbots and small conversations are great to hide some Easter eggs (Martín, 2017).
What all three storybots have in common, to a different extent, is the effective use of what can be considered as “patterns of randomization”- or, simply put, “Easter Eggs”.
This is one of them:
It´s a piece of text John Beadle (and, in essence, the user) encounters when he tries to escape his dreamscape and one that is referred to as a possible answer to his conundrum- yet the text written in Yiddish and Hindi, amongst other languages, is never referred to explicitly again by the bot while the intentionality behind sharing these specific words with the user remains unresolved.
The mystery and suspense behind this action in some ways encourages the user to continue with the experience, in the hope of discovering similarily unique elements.
In “Jessie´s story”, said Easter Eggs come in the shape of offhand remarks that include pop-culture references such as this one which refers to the classic Madonna song, “Don´t cry for me Argentina” in conjunction with “Jessie´s” failed date with an “Argentinian hottie”.
Despite an attempt to establish which of the bots analysed in this piece make better use of content types (photos, emojis etc) or allow users to get actively involved in the conversation, thereby allowing for better overall textual flow, I have discovered that there is no definite, foolproof answer- not only, because, in the end, it comes down to a matter of personal choice but also due to the fact that it was “not only baby steps for LOST” (Milliken, 2016), as maintained by its producer, Sylvaine Souklaye, it´s baby steps for all of them.
In many ways, storybots do provide us with a wealth of information on what not to do but also allow for “the developing of codes and a grammar for those who will follow in their footsteps” (Hoguet, 2016).
In storytelling, that may very well be a bot that spins its yarn by telling the story you specifically want to hear- or it may not.
We don´t know (yet!)- and that´s the beauty of it.
References
Hoguet, B. (2016) How To Tell Stories With Chatbots? Benjamin Hoguet [online] 18 November 2016. Available from http://www.benhoguet.com/how-to-tell-stories-with-chatbots/ [accessed 6 February 2018]
Grian, D. (n.d.) A TendencyTo Spill: using a chatbot as a storytelling platform. Watershed. Available from https://www.watershed.co.uk/studio/projects/tendencyto-spill-using-chatbot-storytelling-platform [accessed 14 February 2018]
Krumhausen, S. (2016) The Bot Playbook. Chatbots Magazine. Available from https://chatbotsmagazine.com/the-bot-playbook-7bb6d181a6a9 [accessed 13 February 2018]
MacPherson, S. (2016) How I came to love an annoying millenial: Bots are people too. Medium. Available from https://medium.com/ddouble/how-i-came-to-love-an-annoying-millennial-bots-are-people-too-3d25fc2a14d8 [accessed 13 February 2018]
Martín, J.(2017) Design Framework for Chatbots. Chatbots Magazine. Available from https://chatbotsmagazine.com/design-framework-for-chatbots-aa27060c4ea3 [accessed 6 February 2018]
Milliken, G. (2016) This Facebook Messenger Chatbot Tells You a Short Story. VICE. Available from https://motherboard.vice.com/en_us/article/vv7n4x/this-facebook-messenger-chatbot-tells-you-a-short-story [accessed 6 February 2018]
Monteon, J. (2016) 6 User Experience Tips for Designing your best Chatbot. Wizeline. Available from https://blog.wizeline.com/6-ux-tips-design-best-chatbot/ [accessed 13 February 2018]
Sachdeva, M. (2017) Writing Stories For A Storytelling Chatbot. Medium. Available from https://medium.com/@manchitkaur/writing-stories-for-a-storytelling-chatbot-a18948e79632 [accessed 6 February 2018]
Souklaye, S. (2016) le déserteur: Can a machine be a work of art with feelings? Sylvaine Souklaye: Misantrophie d’utilité publique [online] 29 October 2015. Available from https://souklaye.wordpress.com/category/short-novel/ [accessed 6 February 2018]
Surmenok, P. (2016) Chatbot Architecture. Medium. Available from https://medium.com/@surmenok/chatbot-architecture-496f5bf820ed [accessed 11 February 2018]
Vishwanathan, V. (2017) Storytelling and Bot Making. Data Science Central. [online] 20 February 2017.Available from https://www.datasciencecentral.com/profiles/blogs/storytelling-and-bot-making [accessed 14 February 2018]
