As a stereotypical Millennial, I’m a huge fan of Netflix. I can finish spending hours binge-watching a whole series, only to stumble upon another binge-worthy series that I just have to watch immediately. Before I know it, it’s 5am and I can hear the sound of birds chirping outside my window. It’s an endless cycle that I find hard to resist…
So how does Netflix keep me so tuned?
Netflix (n.d.) uses what is called a recommendation algorithm. This algorithm tracks our activity (including what we watch and for how long) and uses that data, along with data from those with similar interests to you, to recommend films more relevant to you.
NBC News (2017) discusses this in depth:
To explore this algorithm more, I compared my friend Luke’s Netflix recommendation with mine.
As you can see, the titles vary greatly. Based on these recommendations, Netflix suggests Luke enjoys comedy skits and horror films, whereas I prefer comedy and action films, with a taste for romance. Although these recommendations accurately represent our tastes in film, upon observing my friends recommendations, I found film titles I didn’t actually know existed. For instance, I had no idea Mr. Bean films were on Netflix. I love Mr. Bean! Why wasn’t I shown his films?
Eli Pariser (2011) refers to this phenomenon as being in a “filter bubble”. These algorithms over-personalise our experience so much so that we’re limited to only seeing the content it believes is most relevant to us. In my case, because I haven’t used Netflix to watch Mr. Bean films, Netflix’s algorithm assumes it’s irrelevant and filters it out.
I find this pretty drastic, especially when our world and lives have the potential to be “shaped” and “controlled” by these algorithms (Jennifer Dutcher 2014). Some experts even believe we’ll soon become accustom to these filter bubbles and turn into “zombies who exclusively consume easy-to-consume items” (Rainie, L & Anderson 2017).
I found this was already the case for me. I bounce from recommendation to recommendation without any thought to peek outside my filter bubble. Had I of done so, I would of stumbled upon Mr. Bean. So how can algorithms be improved to avoid all of this?
Well, this “stumble” I just mentioned is referred to as serendipity (an unplanned discovery we find ourselves), and Valentina Maccatrozzo (2012) suggests these algorithms be re-coded to incorporate this effect. Not only will this give us joy in finding a new film ourselves, it’ll create us a sense of self-actualisation in which we discover and explore our own interests (Knijnenburg, BP, Sivakumar, S, Wilkinson, D 2016).
In addition to this, Pariser (2011) calls for algorithms to include a balance of both “information desserts” and “information vegetables” (information that’s relevant to our interests, along with information that challenges our point of view). Maccatrozzo (2012) supports this argument, suggesting they include a “balanced mix of diversity, novelty and relevance”.
Incorporating these theories into the Netflix recommendation algorithm will burst the filter bubble, thus expanding our world, challenging our point of view and avoiding the possibility of our society becoming accustomed to the convenience of instant relevance. It will allow us to discover films ourselves, even if such film isn’t based on our interests. In the end, we might actually enjoy it!
Jennifer Dutcher 2014, Kevin slavin: how algorithms shape our world, UC Berkeley School of Information, viewed 15 July 2018 <https://datascience.berkeley.edu/kevin-slavin-algorithms-shape-world/>
Knijnenburg, BP, Sivakumar, S, Wilkinson, D 2016, Recommender systems for self-actualization, Clemson University School of Computing,viewed 15 July 2018, <https://par.nsf.gov/servlets/purl/10024867>
Netflix, How netflix’s recommendations system works n.d., viewed 15 July 2018 <https://help.netflix.com/en/node/100639>
NBC News 2017, Why netflix’s algorithm is so binge-worthy | mach | nbc news, 25 April, online video, viewed 15 July 2018, <https://www.youtube.com/watch?v=nq2QtatuF7U>
Filter bubble, 2015, image, Perspective IX, 12 January, viewed 15 July 2018, <https://perspectiveix.com/blog/filter-bubble-over-personalised-internet>
Rainie, L & Anderson, J 2017, Code-dependent: pros and cons of the algorithm age, Pew Research Center, viewed 15 July 2018, <http://www.pewinternet.org/2017/02/08/code-dependent-pros-and-cons-of-the-algorithm-age/>
TED 2011, Beware online “filter bubbles” | eli pariser , May 2, online video , viewed 15 July 2018 <https://www.youtube.com/watch?v=B8ofWFx525s>
Valentina Maccatrozzo 2012, Burst the filter bubble: using semantic web to enable serendipity, The Network Institute, viewed 15 July 2018, <https://pdfs.semanticscholar.org/a06d/6219d6724f7a5ae0e4d8b4b03f9d2efba620.pdf>