Netflix and Big Data PDF

Preview

Summary

Term Paper Information and Communication Theory Netflix and Big Data How big data became important to Netflix. By Candidate: 117. 20-05-2016, Oslo. Department of Journalism and Media Studies (Faculty of Social Sciences) Oslo and Akershus University College of Applied Sciences. i Netflix and Big Data...

Description

Accelerat ing t he world's research.

Netflix and Big Data Kylian Weijers

Related papers

Download a PDF Pack of t he best relat ed papers 

No (Big) Dat a, no fict ion? T hinking surveillance wit h/against Net flix Gloria González Fust er, Rocco Bellanova

Ret hinking int ernat ional T V flows research in t he age of Net flix Ramon Lobat o Cult ure, Quant ified Coralie Zaza

Term Paper Information and Communication Theory

Netflix and Big Data How big data became important to Netflix. By Candidate: 117.

20-05-2016, Oslo. Department of Journalism and Media Studies (Faculty of Social Sciences) Oslo and Akershus University College of Applied Sciences.

i

Netflix and Big Data

Candidate: 117

Preface The topic of big data and Netflix has been brought to my attention for quite some time ago already, when I learned of the fact how Netflix makes use of its user data to optimise its services and products. Also, I have read the book ‘Big Data’ by MayerSchönberg and Cukier in one breath and was very intrigued by it. Additionally, my interest lies in the film and television industry and therefore I wished to specialise myself more in both Netflix and big data and make it the subject of this paper.

Whilst going through the literature, I found out there is a whole lot more behind big data and Netflix: a series of consecutive developments in not only technology, but also culture. Therefore, I was very eager to write this paper and hope you, as reader or assessor, would enjoy reading it.

- The author

20-05-2016

ii

Netflix and Big Data

Candidate: 117

Table of Contents 1 Introduction ..................................................................................... 1 1.1 Theories and framework .................................................................................... 2 1.2 Methodology ...................................................................................................... 3

2 Big data ............................................................................................ 4 2.1 Big data in the media industry ........................................................................... 5

3 Netflix ............................................................................................... 7 3.1 How Netflix uses its data .................................................................................... 7 3.1.1 Recent developments ................................................................................... 8 3.2 Intuition vs big data ........................................................................................... 9 3.3 Netflix and the algorithm culture ....................................................................... 9

Conclusion ......................................................................................... 10 Works cited ....................................................................................... 12

20-05-2016

iii

Netflix and Big Data

Candidate: 117

1 Introduction Big data is seen as the next new resource of the current industrial revolution and even stated to be “the new oil” (Hirsch, 2014). The choice of these words need careful consideration. First, big data is not “new”, it has been out there for decades and crosses multiple civilisations – ranging from the library of Alexandria in the third century B.C., to the Spanish flu epidemic in the early 1900s, and from the fourteenth century A.D., when Italian accountants started with the double-entry bookkeeping, to today’s personal recommendation algorithms by (amongst others) Netflix (MayerSchönberger & Cukier, 2013; Barnes, 2013). Second, big data is said to be today’s “oil” (Hirsch, 2014). This can be interpreted in multiple ways. For example, a company can enjoy great wealth when owning oil. Yet the company can suffer from great damage when its oil starts to leak. Nonetheless, big data is today’s discourse and can bring a media content provider potential competitive advantage, as it has to Netflix (Madrigal, 2014).

When putting big data and Netflix next to each other, terms such as digitalisation, datafication (a term coined by Mayer-Schönberger and Cukier), algorithm culture and a new way of curation come to surface. In this paper, I will explain how big data has made its great entrance to the digital age and how it has changed the way in which information is offered, but more specifically illuminate how it has impacted media providers such as Netflix. When doing so, the following research question will be answered:

How did big data develop in the media industry and how did it become important to Netflix?

In the endeavour of answering the above-mentioned research question, a number of things need clarification. This will be discussed in the following paragraph of theories and framework.

20-05-2016

1

Netflix and Big Data

Candidate: 117

1.1 Theories and framework First, it is important to establish how the term ‘big data’ will be used throughout this paper. Big data is a term that came up during the ‘explosive increase of global data’ and was used to describe those datasets (Chen, Mao, Zhang & Leung, 2014). MayerSchönberger and Cukier (2013) and Barnes (2013) have pointed out that data has already been gathered and compiled for a long time, as exemplified with the Library of Alexanderia. It is therefore interesting to look into how big data became ‘big’ data. Even though literature has pointed out that big data impacted society greatly (MayerSchönberger & Cukier, 2013; Harper, 2016), it has also become clear through research that it was society allowing the shift towards the proportions of data and thus impacted data itself. This two-sides influence is the algorithmic culture, a term presented in literature often along with the concept of recommendation systems. With a focus on the media landscape and culture, it is therefore interesting to look into how big data has grown in relation to culture.

Datafication is a term coined by Mayer-Schönberger and Cukier and relates to the ‘quantifying of the world’, or rather, the act of quantifying a phenomenon in order to further tabulate and analyse that data (2013). This will lead to the case of Netflix: a media company which has proportionally datafied its content for the benefit of its recommendation algorithms. There has been great discourse surrounding Netflix’s recommendation algorithms and how effective it is (Hallinan & Striphas (2014). However, Netflix has dedicated a tremendous amount of time in developing its algorithms. Therefore, it is not only interesting to look into how Netflix makes use of big data, but also how it has made efforts in developing its systems.

Conclusively, the framework of this paper will be that of two parts, with (1) a study on big data and how big data has impacted the media industry and (2) a study on Netflix on how it uses big data and how it has grown to become its competitive advantage.

20-05-2016

2

Netflix and Big Data

Candidate: 117

1.2 Methodology Results of this study will solely be based on secondary research by means of text analysis. This is a qualitative research for which purely academic literature has been consulted. The reason for doing so, is because the aim of this research is to pick up on studies of big data and Netflix where other academics left off as to continue the discourse on both subjects. The aim is to illuminate the development of big data in combination of Netflix’s algorithmic developments. However, given the fact that there is a vast amount of research conducted on big data, it is important to connect as many dots in this paper between the words big data, media industry and Netflix.

All the works referred to can be found back in the works cited list at the end of this paper. Especially Big Data: A Revolution That Will Transform How We Live, Work and Think by Viktor Mayer-Schönberger and Kenneth Cukier (2013) has been the main source of reference on big data for this paper. However, this text has been thoroughly juxtaposed with studies of other academics with the intention to contrast findings and to draw new conclusions.

Key terms are: big data, datafication, media industry, Netflix, recommendation systems, and algorithm culture.

20-05-2016

3

Netflix and Big Data

Candidate: 117

2 Big data Schmidt and Rosenberg (2014) exemplify in their book How Google Works in a few instances the usage of big data:

‘Companies like [Google] aggregate anonymous signals from mobile phones to provide accurate traffic data in real time. London’s water pipes are monitored by thousands of sensors, reducing leakage by 25 percent. Ranchers embed sensors into their cattle that transmit information about the animals’ health and location (…)’ (Schmidt & Rosenberg, 2014).

The way in which the above-mentioned phenomena are quantified is, in fact, defining the term datafication by Mayer-Schönberger and Cukier. Netflix also makes extensively use of datafication; not only by means of tabulating and analysing its user data for the benefit of its algorithms and by that its recommendation systems, but also by means of datafying all their movies to the extend where Netflix created more than 75,000 micro-genres (Madrigal, 2014). Accordingly, the media content Netflix subsequently creates can be called ‘big data media’: a term coined by Harper, which he describes as media content that is ‘constructed based upon the typically algorithmic analysis of large sets of data’ (2016).

However, big data, according to Boyd and Crawford (2012, p. 663), is less about data being big, than it is about ‘the capacity to search, aggregate, and cross-reference large data sets’. This attributes to the fact that the phenomenon of big data is technologically driven, since the Digital Age gave way those technological advancements.

Schmidt and Rosenberg (2014) substantiate this and explain that the age of the Internet has allowed transformative rise to ‘the ability to quantify almost any aspect of business’. In other words, the shift towards the digital age and Internet has given data opportunities to quantify and become bigger. Mayer-Schönberger and Cukier disagree on the fact that big data would find its origins in the age of Internet. In fact,

20-05-2016

4

Netflix and Big Data

Candidate: 117

they explain that, indeed, modern-day IT technologies have allowed data to grow, yet essentially the move to big data ‘is a continuation of humankind’s ancient quest to measure, record, and analyse the world’ (2013). Nonetheless, it is technology that allowed data to grow big – with ‘big’ meaning great in volume (enormous quantities), velocity (created in real-time), variety (structured, semi-structured, and unstructured) (Laney, 2001, as cited in Kitchin & McArdle, 2016) and value (in terms of high quality journalism and business insights) (Stone, 2014).

2.1 Big data in the media industry Today’s economy can be described as the experience economy, where products are not only customised but rather personalised and where it is not only about the service but rather the experience (Pine & Gillmore, 1998). This change towards experience economy is amongst others attributed to the increase of possibilities in product development presented by technological advancements. Moreover, when a service is transformed into an experience, it grows substantially in economical value (Pine & Gillmore, 1998). This can explain the competitive advantage of services like Netflix, which offers services and content tailored to each of its users.

Big data has been employed by media companies’ strategies and actions for a number of beneficial reasons: •

To better understand cross-platform audiences,

•

Create powerful data journalism stories,

•

Streamline business processes and

•

Identify new products and services to offer customers (Stone, 2014).

What, however, allowed big data to grow so proportionally in the media industry? According to Stone, this is attributed to the prices for digital media storage and bandwidth, exponential growth of smartphone and tablet usage, and the increase of audience access to digital media (2014). This is another example of how the technological and socio-economical aspects of culture intertwine: technology made

20-05-2016

5

Netflix and Big Data

Candidate: 117

possible the accessible use of smartphone by which subsequently the audience started using more and more smartphones which led to an increase of data growth.

In addition, the media industry is using big data to such an extend that, for example, ‘automated recommendation systems now occupy a central position in the circulation of media and cultural products’ (Morris, 2015). These automated, or rather algorithmic, recommendations systems are different from other forms of prediction by using social scientific methods with behavioural data and individualised predictions, Cappella, Yang, and Lee (2015) explain. In fact, it is not about male/female or what they simply like, it is based on ‘what a specific person has selected previously, what will that same person likely select in the future from a range of possibilities not yet examined?’ (Cappella et. al, 2015). Just and Lazer substantiate this and explain that traditional mass-media selection mechanisms differ from algorithmic mechanism in substantial ways. For example, the automated algorithms operate without time delay and are not targeted at pre-defined audience groups (2016). Furthermore, very interestingly, whereas traditional mass media prominently contributed to gatekeeping, agenda-setting, and framing, it is now the user that creates his or her own reality and is in fact its own gatekeeper and setting its own agenda (Wallace, 2015), albeit indirectly, since users of Netflix, for example, can only see the content that is within Netflix’s databases.

Conclusively, big data is a technologically-driven product of an increase of data aggregation with especially Internet that has contributed to the transformative rise of datafication. Big data is not necessarily about being big in velocity, volume, variety and value, but also big due to the ability to tabulate, analyse, search, aggregate, and cross-reference datasets. Big data is extensively used in the media landscape and allowed to grow due to the exponential growth of other media platforms such as smartphones, the low cost of broadband storage, and increasing accessibility of the Internet. In fact, the media landscape makes use of big data to such an extend that algorithmic systems now inhibit a central position in the circulation of media and cultural products, which differentiates itself from traditional selection mechanisms.

20-05-2016

6

Netflix and Big Data

Candidate: 117

3 Netflix To learn more about big data and the algorithmic culture, we will look at the case of Netflix. First, it is important to clarify what Netflix does and what kind of company it is. In 1997, Netflix was founded by Reed Hastings and Marc Randolph as an online movie rental company. Since then, Netflix has grown to become a worldwide multiplatform on-demand television and movie broadcaster (Netflix, n.d.). Additionally, it has started to offer original content such as drama series and comedy series (2016, Jenner). When it comes to the series, David Carr speculated in his article ‘Giving Viewers What They Want’ that Netflix was already certain that the show House of Cards would become a success, even prior to shooting any scenes (2013). The secret lies, he explains, in Netflix’s big data.

3.1 How Netflix uses its data Firstly, Netflix makes use of its data for its recommendation systems. Amatriain (2013) stresses that the recommendation systems are a ‘prime example of the mainstream applicability of large scale data mining’. In 2000 already, Netflix started with data mining by asking users to rate the movies. According to those ratings, Netflix would personally offer new recommendations and it continued doing so on its streaming service from 2001. (Netflix, n.d.). In 2006, Netflix had released over 100 million rental records for the purpose of improving its rating system (MayerSchönberger & Cukier, 2013). Netflix would challenge others to improve its system by 10% of its current capacity. This challenge is known as the Netflix Prize Challenge

However, the solution to the Netflix Prize Challenge was eventually not used, due to the amount of work but also because Netflix decided to move beyond the 5-star rating system (Mardigal, 2014). In fact, the Netflix Prize Challenge indicated that the 5-star system was less accurate as Netflix wanted it to be and that there was a magic barrier (Amatriain, 2013). This means that a recommender system cannot reach any more accuracy, regardless of the description of each one of the 5 stars or how successful a recommendation is (Herlocker, Konstan, Terveen, & Riedl, 2004). In other words, the 5-star system was not effective enough for Netflix; more specificity was needed.

20-05-2016

7

Netflix and Big Data

Candidate: 117

In the endeavour of looking beyond the 5-star recommender system, Netflix started to rate – or rather tag – each movie in regards to very specific measurements. This has resulted to a multitude of personalised genres, with 76,897 in 2014. Madrigal (2014), who has interviewed Netflix’s VP of Product Todd Yellin, described this phenomenon called the ‘Netflix Quantum Theory’ as follows:

‘The Netflix Quantum Theory doc spelled out ways of tagging movie endings, the “social acceptability” of lead character, and dozens of other facets of a movie. Many values are “scalar” […]. So, every movie gets a romance rating, not just the ones labelled “romantic” in the personalised genres. Every movie’s ending is rated from happy to sad, passing through ambiguous. Every plot is tagged. Lead characters’ jobs are tagged. Movie locations are tagged […] (Madrigal, 2014).

Here we see a repetitive datafication or quantification of certain things – such as a plot, the characters’ job, or movie locations – to meet today’s personalisation standards.

3.1.1 Recent developments From the point of the Netflix Quantum Theory, Netflix has shifted its data mining efforts elsewhere than the 5-star rating system. In fact, Gomez-Uribe (VP of Product Innovation and Personalisation Algorithms) claims that the rating system is not sufficiently useful, but the viewing / playing habit is (as cited in Vanderbilt, 2013). Netflix’s CPO Neil Hunt explains that the problem is that people ‘subconsciously try to be critics’ and when rating the movie, they ‘fall into trying to objectively assess the “quality,” instead of basing the stars on how much “enjoyment” they got out of it’ (as cited in McAlone, 2016). Subsequently, this enjoyment is subtracted from the user’s viewing habits: e.g. when a user pauses, rewinds or fast forwards; when a user leaves the movie before it finishes; when a user watches an episode again; when and how a user watches the content (Amatriain, 2013). Hence Netflix’s decision to move away from the 5-star rating system.

20-05-2016

8

Netflix and Big Data

Candidate: 117

3.2 Intuition vs big data At Netflix, CEO Reed Hastings explain that they start with the data, ‘but the final call is always gut – it is informed intuition’ (Ferenstein, 2016). Mayer-Schönberger and Cu...