In the first part of my look at interactive fiction, I outlined IF’s roots in early computer gaming, and its revival through engaged communities and innovative producers. In this follow-up, I’m going to look at the future of the form - technological advances in voice, language processing and AI that could give interactive fiction greater commercial and creative traction within mainstream publishing.
Intent is king
The key differentiator between fiction and interactive fiction is the concept of intent. All interactive fiction requires the user to develop an intent related to an objective, and interact with the work to achieve this. Intent can range from the simple, like selecting the order in which a narrative is presented, to the profound - intervening in the narrative and changing its outcome.
The user needs to be ‘into’ having that responsibility put on them. Audience participation is required. At the same time, the onus is on the interactive fiction producer to make that participation rewarding. Good content is insufficient - the user’s intent needs to be handled in such a way to create a good user experience. And of late, interpreting and facilitating intent is a hot area of technology.
Interactive fiction has traditionally used a software component called the ‘parser’ to derive meaning from the user’s free text input. Simple statements like ‘Open the door’ don’t present too much challenge to parse, but complex instructions, closer to what we would call natural language - ‘put the green key in the red box and then place both on the table’ -require significant processing power, in addition to a sophisticated model of language to call upon.
Complex parsing has now been enabled by the availability of processing power through cloud computing, research into linguistics models, and machine learning, which enables computers to develop their own models. This in turn has led to the development of the voice assistants found in phones, home and increasingly, cars.
Sophisticated parsing technologies will soon become available as services, which can be plugged into products like interactive fiction. In other words, it will become possible to talk to the book to express intent, rather than select from a list of options. The goal is truly conversational interaction with computers through speech and text.
Because of the potential for variation and subtlety, analysing intent from speech is an order of magnitude more complex, and computationally demanding, than written language. It also gives the interaction designer more data to play with, particularly in nuances of intonation, emphasis and emotion that can be used to adapt a response.
In terms of machines reading emotions, Apple’s recent announcement of the depth-sensing technology in the iPhone X, which can detect emotion from facial expressions, points to a future where devices can analyse the emotions of their readers and applications can respond to them by making choices, including content selection.
IF and voice
Voice assistants, like Amazon’s Alexa and Google Assistant, are natural platforms for delivery of interactive fiction. The underlying design of applications for these services - what Amazon and Google call ‘skills’ and ‘actions’ respectively - require designers to predefine dialogue options for the user to navigate the functionality, or content, available to them. Dialogue ‘trees’ like these resemble the presentation of choices to users in interactive fiction.
It's perhaps no surprise, then, that interactive fiction skills were an early addition to Amazon’s Echo speaker list of the skills. The Wayne Investigation is a highly-polished demonstration of audio content mixed with voice interaction, whilst UK game developer Jagex have developed an interactive murder mystery audio adventure for Alexa, One Piercing Note, based on their Runescape fantasy universe. Earplay is a company specialising in interactive audio adventures. Their Codename Cygnus game is one of the first to work across Alexa and Google Assistant.
It could be that interactive audio fiction establishes itself in line with the growing popularity of audiobooks, as purchasers of voice devices look for quality content, which remains limited. These examples suggest that a distinct genre of interactive crime/mystery fiction might be emerging already.
They also have high production values, but there are few barriers to producing more basic interactive content. Building and deploying simple voice skills is inexpensive and Amazon have already released free software that functions as a template for building voice-based adventure games.
Showing real character
Sophisticated voice parsers capable of learning will enable us to interact in a more human-like way with stories. But beyond that is the possibility that stories, or perhaps more specifically the characters within them, will become adaptive through the application of artificial intelligence. Emily Short is Product Manager for Character Engine at Spirit AI, a firm developing an AI capability that can power interactive narrative. She summarises Character Engine thus: “Character Engine is designed to model the behaviour of a character who can improvise around the player’s inputs while still serving the author’s narrative goals. The idea is to allow the character to continue to express an authored vision of who that character is, but sometimes in circumstances that the author did not specifically envision in advance.”
Again, this is technology created to adapt to intent. Gaming is an obvious application for this kind of technology, with the complex and rich fictional landscapes of ‘open world’ games, populated by human players and non-player characters. These AI characters will have some capacity to choose their own responses to humans that communicate with them - not simply to understand what they have been told, but also to have a simulated emotional response to it and develop memories they are able to recall, making interaction more lifelike and unpredictable.
The same principles could apply to a textual fictional world. Instead of characters being chained to the page, they could be given a degree of liberty - provided of course they still hit their story ‘beats’ so the reader gets to experience a coherent narrative.
Interactive fiction which is centred around characters, and where the reader intent is to achieve a goal by building a relationship, may appeal to some who are not interested in traditional gameplay. I can see intriguing possibilities for this in areas like the romance genre.
Ultimately, stories could also be constructed where fictional AI characters have an awareness of each other. If two AI characters met in a fictional scene, the fundamentals of the scene might be determined, but the nuances might be unpredictable, making those characters behaviour akin to actors.Truly autonomous characters would be as troublesome for writers as the hosts of HBO’s Westworld are for their creators, roaming outside of their story arcs - but at least not lethal to their readers.
AI characters interacting with each other in novels may be blue-sky thinking, but the trend in interactive fiction over the last few years has generally been towards character interaction over solving puzzles, coupled with more sophisticated narrative mechanics that make the reader’s journey feel less pre-determined.
The simple 'branching narrative' approach where the reader selects a path through a story 'tree' is being overtaken by ‘choice and consequence’ designs. In these, actions the player takes have rippling effect throughout the rest of the story. Apparently trivial choices can have significant, unintended consequences and expand the the range of possible story outcomes. This is done by maintaining values related to actions open to the player, representing several ‘states’ that are computed to produce a result at key points in the story.
Choosing what characters say to each other is a common way to present these kind of choices, with the effect being an update to the state of the emotions or intentions of non-player characters (NPCs) which affect the story later. Telltale Games’ works based on TV properties like The Walking Dead and Game of Thrones employ this frequently and tantalise the player with the ambiguous response ‘He/She will remember that’.
This style of games arguably has broader appeal because they are more about participating in drama than solving puzzles. The breadth of that appeal is demonstrated in the creative and commercial success of Until Dawn, an interactive homage to teen slasher movies, published by Sony for the PlayStation 4 which surpassed initial sales expectations and went on to win a bafta Game Award in 2016. The complexity of Until Dawn’s underlying choice and consequence story mechanics - branded as its 'butterfly effect' - is demonstrated in that the player can successfully save, or fail to save, any number and combination of the game’s eight player characters from the game’s psychopathic antagonist.
Similar mechanics are used in the forthcoming game from Sony, Hidden Agenda, by the same Guildford-based developer, Supermassive Games,. This taps into the observation that many enjoyed playing Until Dawn in groups, or streaming it online. Hidden Agenda makes use of a new social gaming feature for the PlayStation 4 called Playlink, that enables multiple players to connect their mobile phones to the console over wi-fi and play together, using their devices to make choices and view their own secret game objectives.
Interaction that requires users to partake in moral dilemmas rather than solve puzzles has a broader appeal than hardcore gaming. It's easy to see this working for a party video game like Hidden Agenda, but extending the concept to collaborative (or competitive) reading where readers experience and influence a connected text-based narrative is entirely feasible.
The power of social has yet to be fully tapped in a storytelling context. Projects don’t necessarily need to be on the scale of James Frey’s Endgame, with global participation and huge prizes for cracking puzzles. Less expansive concepts, where groups of users could ‘enter’ stories together, and interact with each other and story elements, is the essence of multiplayer gaming. This is associated with graphically advanced action games but it could work as textual, narrative-focused experiences. Social networks are well positioned as distribution channels for such experiences as they develop.
Interactive fiction is an idea that has found new form in every generation of information technology. It can’t challenge the enduring appeal of the story told as a holistic and immutable experience by the author, but innovations in AI, natural language technology and adaptive narratives point to an exciting future.