Designing Structured Stories
This is a written version of a talk I gave at the Interaction 16 conference on the 2nd March 2016. You can also watch it here.
Stories are everywhere, from newspapers to TV drama, fact to fiction. By deconstructing stories into fundamental, structured building blocks we can present them in new and powerful ways.
We can make experiences that vary the length of the story, let you explore it from different directions, summarise and recap it, change the media used to present it, adapt it to what you already know or personalise it to make it more relevant. And we can create truly responsive content that adapts to your device and situation.
Through two case studies at the BBC, mobile news and drama serials, I’m going to look at how we took a data-driven but user-centred approach to structuring and designing story experiences.
The primary building blocks we use in our work are storylines, key moments and people. The storyline might be the epic journey of a character, the events in a refugee crisis or a love triangle. The essential moments and events in these storylines could be the bits that people remember most fondly, the key to understanding or the jumping off points to other stories. Yet the key people or characters are almost always the most important part of a story.
This is the story of my work at BBC R&D on what I (and others) call structured stories; it’s about what you could make with them, why structure is important and what we’ve learnt doing it.
The structure of stories
My story of this work starts here, with a book — Everything Bad is Good for You by Steven Johnson. It’s about the increasing complexity of modern popular culture. And I was particularly interested in this diagram of plots in American TV crime drama. At the top is Dragnet, from the 60s, where one episode was a single narrative thread, similar to Starsky and Hutch from the 70s. Then it starts getting more complex with Hill Street Blues. Finally it shows The Sopranos which has many substantial parallel plot strands in one episode. If you watch any critically acclaimed TV drama from Scandinavian noir to The Wire to Game of Thrones you’ll probably have noticed this. This started me thinking about what we could do with these structures
Luckily lots of clever people have have already thought a lot about stories. Much has been written on the shape of plots, three acts and five acts. The classic five act structure consists of the exposition, the rise, the climax, the fall and the denouement. If I was really good then this article would use that form.
But it’s often not just one thing after another in stories. Memento, the Christopher Nolan film, plays with time. One strand of it is ordered chronologically (the black and white bit), the other goes in reverse (the colour bit), and the two are interleaved in the film (see this diagram). In fact several Nolan films play with time and telling and I’d suggest that lots of media increasingly does this .
A story model
So it seems generally agreed that stories do have structure. One Russian school of literary criticism defined two things that seem particularly relevant to us — the fabula is “the raw material of a story”, what happened, and syuzhet is “the way a story is presented”, the narrative.
In our work at BBC R&D we look for how to do scalable, reusable solutions using a solid backbone of data. We developed a data model for stories with three layers.
We have the world of the story: the protagonists, people, places and relationships.
On top of that, the events: what happens in the story.
And on top of that, the narrative: or how the story is told.
(Paul, who worked on this, has a similar model described here)
Cards
A small diversion from stories. There’s been a recent design trend for cards and designing things smaller than pages. This quote from Paul Adams of Intercom gets to the heart of it
In a world of many different screens and devices, content needs to be broken down into atomic units so that it can work agnostic of the screen size or technology platform
To make proper responsive content we need re-usable atomic units of content. And to make them properly re-usable we have to add structure or we won’t know how and when to re-use them.
So if we atomise stories into re-usable components then structure and label these components, we can re-combine them to make new things in new places.
Stories in fiction and reality
I’m going to look in more detail at two kinds of stories — drama and news, fiction and reality — and how we used some of these story principles to make some things.
The Archers is the world’s longest running soap opera — it’s 65 years old this year. It’s about a farming community and is broadcast every day on the radio in the UK. This is it’s card index.
It tells you things like what happened in 1967, who married who, who’s got a piano in their front room & what the cows are called (yes, really). It’s used by the writers so that they don’t get anything wrong and are consistent. There are a lots of listeners that will notice. But also to give them inspiration for new storylines. There was an old Windows database but that stopped working and so our team at BBC R&D designed a new web-based structured continuity database.
The point is, there is already structured data about the stories that the BBC tells. But this was so huge it was a bit daunting to start working with so we worked with a slightly shorter radio drama.
The Story Explorer
In 2015 we started working with Home Front, a BBC radio drama serial charting life on the home front during the First World War, who were nice enough to let us play around with their programme. We took the first two series and using lots of spreadsheets we split the episodes up into the storylines, events or scenes, characters and places involved. For each scene we also chopped out the audio from the programme and wrote a quick summary so we had a self-contained “atom” of media for every event.
Remember that bit about the increasing complexity of modern popular culture? Boxsets and the impossibility of missing an episode have meant there is increasingly complex drama on TV and radio. Jason Mittell has written a whole book on this. We wanted to use this data for those times when you’re watching (or listening to) a drama series and think “Who was that?” or “What just happened?” or maybe you can’t remember what happened in the last series or the last episode. There was a space to make something to help explore TV and radio drama and solve a real problem for people.
A colleague even joked that his mum would love a “Remove post-modernism” button on their TV that just re-ordered everything back into a chronological timeline!
This is the Home Front Story Explorer — you can try it here or watch an intro video here. It’s not intended to be a replacement for listening but it’s useful if you want to quickly catch up or remind yourself what happened, or get an introduction to drama that’s new to you.
You can choose the storyline you want to catch up on and then listen to each scene. Or you can just skim through it and read the bits that interest you. Plus you can see which characters are in each scene and find out more about them. You can read much more about it here.
We piloted it last summer and it went down really well with the audience. Some were new to the drama and used it to get into the story, some just wanted a recap on bits they’d missed or forgotten before the next series started and some even used it as a single-purpose radio — listening to single storylines for hours.
We’re now trying to develop this concept across TV drama — designing it to be re-usable & scalable for all dramas and beyond, maybe even for reality TV.
Structuring the news
We then decided to apply these principles to news and look beyond the conventional news article, which hasn’t really changed since modern newspapers were invented.
Our research showed that some young people found it hard to get into BBC News. They found it to be be quite inaccessible and missing essential contextual information. And if you think about it there’s often lots of assumed knowledge in the way news is written. What was the Berlin Wall? What is an interest rate? Why is there a migrant crisis?
This is the model we’re been using for news, which has been in development for a number of years. You can find the full storyline ontology developed by the BBC and others here. Storylines are made up of events and each event can be linked to people, organisations, places and even other storylines. As you can see, it looks awfully similar to the drama model. And the BBC doesn’t just do news in one way; we do it online, on TV, on radio and these can all be linked in to this structure.
This example is the story of Alexis Tsipras’s resignation and the Greek debt crisis. Obviously, as with most news, it’s complicated and subjective. Where do stories start or end? Where do they intersect? What is the truth? It’s really interesting and a bit harder than drama and fiction.
It turns out that the processes of journalism have quite a lot of innate structure that we could use. Journalists are taught to use the 5 Ws — who, what, where, when & why. The inverted pyramid of journalism suggests you should lead with the most essential info at the top, getting less and less important to the end. It was originally so editors could cut a column from the bottom to fit the space available in the newspaper.
The BBC used to run a text information for TVs, it was called Ceefax. Amongst many other things it had news stories, all made up of four paragraphs. This structure made it onto BBC News online because journalists would write for both and, despite the demise of Ceefax, you can still see it now in the first four paragraphs of most articles on BBC News.
Atomising news
We’ve had a go at re-inventing the news article for mobiles using structured stories. Through several iterations of prototyping and testing we discovered that people liked having quick, skimmable summaries of stories, particularly on mobile. But they also wanted the option to be able to go deeper and get more information on the aspects that interest them.
Every story is structured as a storyline — with key events, people and places. The initial view of the story gives you a summary on one page, so you can just skim the key events. But you can expand any of them and dig deeper — into longer-form writing, correspondent reports, social media or video. Plus we have pop-up definitions available of the key people, organisations and places for that extra context if you need it.
Elsewhere
I’ve found that structured stories are everywhere when you look. Here’s some obvious examples I found during this work.
Writers often use structure — this is a diagram showing how long-running stories for a comic series were planned out.
This is a screenshot from Arcadia — a novel available as a book and an app. In the app you can switch between the interlocking storylines at many points.
Games are often structured stories — it’s most obvious in interactive fiction, like 80 Days. Social media could even be described as the structured stories of people’s lives.
Fragmentation
But why use structured story data for any of this? It goes back to the earlier quote. By atomising our stories and fitting those atoms to a data model we create re-usable bits of stories that we know when and how to use and then can put together again in different ways. These structures give us the ability to adapt our stories to different screens, platforms, contexts and experiences. Whether that’s watches, conversational user interfaces, AI or things without screens. We could even adapt stories to you — maybe drop the bits of the story you already know or change the language and tone.
I think this is particularly important for media organisations. There are always more and more new things, devices, platforms and places to get stories, media and news. From Google AMP to Facebook Instant Stories, from Line to Vine, from chatbots to Snapchat. Rather than create more and more stuff for each of these new platforms piecemeal as they appear, if we structure our stories into small, reusable pieces then we can efficiently use them again and again. That makes structured stories very powerful.
Though, I might add, not necessarily a wholesale replacement for long form content or content carefully crafted for the channel.
Design patterns
These are some patterns I’ve noticed throughout this work.
Building blocks
Where should you start if you want to build structured stories? I’d suggest you identify the main protagonists, what happens to them in their story and how you want to tell the story. Then try using the model of the story world, the events and the narrative.
Beginnings and endings
Knowing when to start and stop stories is important because you want things to be the right length for people, but also not hide the complexity or full story.
We tended to try to present our stories at an appropriate length for the format & then used two techniques. We added navigation in the Story Explorer to take people from one story to another. In Atomised News we have the concept of entire storylines embedded within events — they describe themselves well enough for their context but they also link out to the full storyline elsewhere.
Order of telling
As we’ve already seen, many storytellers deliberately change the order in which information is revealed to make the story more engaging. So we don’t always want to present events in chronological order. You might expect that at least news articles would be chronological, but they frequently jump between what just happened and what happened previously.
So we separate the events from the narrative and enable them to be presented in many ways according to the storyteller’s desire, or even the user’s preferences. Examples of ordering include chronological timelines, reverse chronological blogs, an author’s narrative order or even algorithmic timelines (as seen in Facebook).
Skimming and digging
Different people will want to consume the stories in different ways. Sometimes we might be telling the whole story, sometimes just a summary.
In the Story Explorer we flagged certain scenes as essential for a storyline, enabling us to provide a shorter “highlights” experience for each story. In Atomised News the expanding spine provides both an overview and the full story for those that want it.
Mixed media
The most appropriate representation of a story may change depending on the person or the situation or the device. But we can hang multiple media representations of the story off the structure and then choose which to use. These representations might be different lengths of content, different media types or even different tones or styles.
Challenges
In this work we tried to balance using a re-usable data model that could underpin scalable experiences while also designing for user needs. This meant that we were iteratively developing the UX, the stories and the data simultaneously and this required a lot of talking and sketching and compromising to make sure the whole team were in agreement and at the same level of understanding.
We needed tools to make structured stories, but we didn’t want to be putting extra things in the way of writers and journalists. The Archers continuity database is a good example of a tool which is part of the existing workflow and actually useful to writers, while creating structured story data as a by-product. And we are looking to make this process even easier by automatically processing existing media — we’re currently investigating the feasibility of machine-parsing programme scripts to (semi-)automatically create story data.
Finally, we tried to make sure that we kept the stories intact, didn’t mess with them too much and let the storytellers do their job — I think we succeeded.
Reading list
Everything Bad is Good For You by Steven Johnson
Into The Woods: How Stories Work and Why We Tell Them by John Yorke
Complex TV: The Poetics of Contemporary Television Storytelling by Jason Mittell
The end of apps as we know them by Intercom
A three-part model for understanding stories by Paul Rissen
The future of news is not an article by NYT Labs
Events, causation, articles, reports, stories, repetition, insinuation, supposition and journalism as data by Michael Smethurst
The Home Front Story Explorer prototype by BBC R&D
Acknowledgements
Paul Rissen, Michael Smethurst, Zillah Watson, Chris Sizemore, Jacqui Maher, James Jefferies, Andrew Wood and the rest of the team at BBC R&D.
Doctor Who illustration by Yoon Bahk.
