In a 2009 TED Talk, writer Elizabeth Gilbert pinpoints the moment when people started referring to an artist as “being a genius” rather than “having a genius.” Gilbert waxes poetic on artistry in ancient Greece and ancient Rome, when “people believed that creativity was this divine attendant spirit that came to human beings from some distant and unknowable source, for distant and unknowable reasons.” If an artist fails it is because he himself is a failure. If he succeeds, he is considered to be a vessel for the voice of God.
So artists and storytellers work diligently despite the inherent risk that comes with the job description. But in a modern, data-driven world, it’s possible to reduce the risk of failure by analyzing the preferences of increasingly cynical audiences. Now viewers can instantly access content via computer-generated recommendation engines rather than relying on word-of-mouth, and this user data establishes a collective intelligence that can easily give the people exactly what they want. While this reduces inherent risk for writers and directors, it can compromise the integrity and authenticity of the narrative.
As good storytelling becomes a quantifiable commodity, will consumer data science replace divine inspiration?
The most obvious example of a narrative kindled by user data comes from Netflix, whose successful original series House of Cards was developed based on subscribers’ preference for dark political dramas and movies starring Kevin Spacey. For almost a year prior to its release, Netflix executives explained how their detailed knowledge of user data clinched their decision to license a remake of a popular 1990 BBC miniseries for $100 million.
But before Netflix outsourced this kind of intuition, author Jaron Lanier attributed a swift decline in conscious creativity to an overabundance of media available by way of networked computer systems he refers to as “intelligent agents.” Even before Netflix mailed out its first DVDs to subscribers, Lanier published a rather discursive article titled Agents of Alienation in a 1995 academic journal.
Amongst a few grandiose generalizations, Lanier essentially claims that our identification with intelligent agents lowers our cultural standards by presenting audiences with uniquely tailored content. Since the computer presents an endless variety of ideas, the user “starts to limit herself to the categories and procedures represented in the computer, without realizing what has been lost.”
An agent’s model of what you are interested in will be a cartoon model, and you will see a cartoon version of the world through the agent’s eyes…
This will recreate the lowest-common-denominator approach to content that plagues TV. ‘You’re interested in Balinese ritual, therefore you’re interested in travel, therefore you’re interested in the Infobahn Travel Game Show!
– Jaron Lanier, Agents of Alienation
Even though this article was published in the 90’s, Lanier’s hypothetical categories are strikingly familiar. After perfecting the art of virtual personalization, “Foreign Dramas with a Strong Female Lead” and “Critically-acclaimed Emotional Underdog Movies” became trademark phrases drawn directly from Netflix’s recommendation algorithm.
According to the Netflix Tech Blog, 75% of what people watch is from some sort of recommendation. After opting into a subscription-based service, viewers have an entire roster of entertainment at their disposal and personalized recommendations become a way to, as Lanier suggests, “find the grains of gold in the heaps of dirt.” And by providing explanations as to why the software is recommending a particular movie or show, Netflix subscribers become keenly aware of how software adapts to their tastes.
When building a recommendation engine using a collaborative-based approach, engineers look at user interactions (such as ratings) rather than metadata from the content in order to generate similar items of interest. Netflix continues to improve user preferences based on search terms, scrolls, mouse-overs, clicks, time spent on a given page, and other behaviors. Netflix may also use a content-based approach to provide users with recommendations, but the company prides itself on optimizing the probability that a member chooses to watch (and enjoy) a title based on user behavior alone.
What happens when directors approach the editing room armed with the knowledge that a certain subset of subscribers are opposed to jump cuts or get off on gruesome torture scenes or just want to see blow jobs. Is that all we’ll be offered?
– Andrew Leonard, Salon
The original programming offered by streaming sites still has potential to generate interesting characters and story arcs, but if marketing research plays a major role in the acquisition of these titles there’s no telling how much research will drive the actual creative process.
In an article for Salon, critic Andrew Leonard envisions how “reliance on Big Data might funnel craftsmanship in particular directions” (torture, blow jobs, etc) He pays particular attention to the fact that Netflix and other streaming services track the amount of times we’ve paused a show or replayed a particular scene. Leonard asks, As intelligent as this software is, it can’t take external factors into consideration. Simply put, Netflix has no way of knowing if you’ve paused the show due to lack of interest or because you needed to take a phone call.
The success of Netflix’s original content strategy has spawned other services to take a similar approach, using their own data to choose which shows get distributed, which writers to hire, and ultimately, which stories to tell. And while many of these programs feature established actors such as John Goodman and Gael García Bernal, the key art looks vaguely similar to programs with proven success rates:
This isn’t some sort of brilliant new idea. The film industry has unapologetically placed safe bets on second-hand stories for quite some time. As of May, the only original screenplay to appear amongst the highest grossing films of 2014 was based on a children’s toy (a sequel to The Lego Movie is already in the works and slated for a 2017 release). Over the past decade, the lack of truly unique stories projected on the big screen generated a need for original programming and spawned the over-referenced “Golden Age of Television.”
This isn’t to say that adaptations aren’t successful (Random House has sold more than 24 million copies of George RR Martin’s five-book Game of Thrones series, with sales surging at the beginning of each season). However, rather than constricting a character’s development into a 150-minute timeframe, the season-long story arc keeps audiences engaged while its characters thrive over multiple episodes. It takes immense collaboration between talented writers to develop these complicated narratives, and reliance on data-induced research can severely hinder that creative process.
Roy Price, director of Amazon Studios, believes he can solve this problem by commissioning pilots (and films, for that matter) from writers and content creators who many not have access to industry tools. Users can upload a full-length screenplay or video that then gets considered to be a part of Amazon’s Development Slate. Amazon then tests storyboards, posters, and promos with by soliciting customer feedback helps Amazon decide which pilots become series.
In an interview with Fast Company, Price insists that Amazon Studios won’t rely solely data on to develop this kind of content. Software within the development slate tracks user behavior, but Price claims that the projects receiving more clicks than others only generate information relative to the show’s marketability. He notes that the defining feature of Amazon Studios is the ability for the audience to provide constructive feedback during all parts of the creative process.
While Amazon’s approach downplays the importance of data in the overall development process, this kind of collaborative storytelling also mitigates the risks involved in any creative endeavor. Whether narratives are developed and structured based on data-analysis or by outsourcing user feedback, the process is still strategic rather than inherently creative. There is a fine line between using and exploiting user data.
“I note increasing reluctance on the part of marketing executives to use judgment. They are coming to rely too much on research, and they use it as a drunkard uses a lamppost, for support, rather than illumination.”
– David Olilvy
Data may provide a good reason for a network to give a great pilot the green light, but it seems likely that brilliant stories will go untold because executives can’t find the data to support their acquisition.
Could data have predicted the success of Seinfeld, the self-proclaimed “show about nothing?” And just because Netflix subscribers loved House of Cards doesn’t mean every network needs a slew of similar political dramas. Ultimately, when we rely on collaborative filtering to dictate user preferences, we start to ignore compelling stories told by actual humans.