by Cory Barker
Finding One: Method Identification
Of the 27 millennials who submitted journal entries on Netflix, 20 identified, explained, and analyzed the methods used by company to tailor recommendations and programing. Frequently, these explanations appeared early in the writing samples, and were used as a means of entry to talk about larger issues of privacy, ease-of-access, and data mining.
Most millennials identified algorithms and computer learning strategies as Netflix’s method to making film recommendations. For example, a 20-year-old male said:
Netflix uses algorithms and metadata to provide recommendations for users. These professionals are gaining intimate knowledge of your interests, emotions, and beliefs in order to gain a perspective of how you are so that the artificial intelligence can make a profile and try to curtail advertisements and certain recommendations based solely of this profile. A perfect example of this type of intelligence is Netflix.
These posts strictly identify Netflix as a company seeking to learn more information about the user so that they may more finely make recommendations and maintain user loyalty and attention. Importantly, these reflections are mostly descriptive in nature, rather than critical or persuasive. This description is a technique used to introduce the topic and then later as evidence.
There were also posts that delved deeper into the types of data collected by Netflix and the extent of their data mining. Participants clearly identified the outward limits of the company’s ability to target, monitor, and collect information about users. Again, these are descriptive reflections rather than a persuasive tone. Three other 20- to 25-year-old respondents said:
Netflix digs deep and tracks: When you pause, stop, fast forward or rewind, the date and time you watch content, what zip code you are in, ratings, browsing and scrolling, searches and when you leave content and if come back. Netflix basically tracks every movement. A good thing about this analytic is that Netflix will know what shows are more likely to be canceled, or help deal with ending credits.
Platforms such as Netflix are able to learn about your interests because they monitor what exactly you search, and exactly what movies you watch. After they gather this information, they analyze it in a database, and everything else that you do in your account, and then they are able to guess what movies they think that you would want to watch.
Netflix learns about your interest by analyzing the different genres, titles, and actor based off of you history on the platform. While some of the recommendations are good, there are also some that are completely off.
The two posts above particularly emphasize the participants’ focus on the personalization and extent to which user data is documented and analyzed to provide insight into customer preferences. The terms “monitor” and “tracks” suggest that participants understand the practices that are used regularly by Netflix to tailor content to the user.
There were also posts that provided examples of the Netflix process in action. These examples demonstrate how aware participants are of the methods and uses of the data provided. Here, participants started to connect narrowcasting to their own lives, particularly emphasizing the recommendations made for their own accounts. The introduction of the subjective “I” as well as sharing of personal experiences suggests that millennial participants are not only aware of the general practices of Netflix; they also understand how it influences their own lives. Three other participants added:
For example, I watched a Disney movie and after I was done, Netflix recommended similar Disney movies.
The more someone uses Netflix, the more the Netflix computer learns about the user. For example, if someone seems to be watching a lot of comedies, Netflix retains this information and suggests more comedies that it thinks the user may be interested in, and the process repeats.
Like millions of other Americans, Netflix is an important part of my life. While there are still some shows that I will watch on an actual television, most of my TV watching occurs on Netflix. The more I use Netflix, the more it learns about me as a person. While this sounds like the worst nightmare for Dwight Schrute [Rainn Wilson] from The Office [2005–2013], it can still be beneficial to the user. Netflix learns about my viewing interest through several ways. First, is me watching shows or movies. If Netflix, sees that I am re watching The Office for the fourth time, it may recommend me to watch Parks and Recreation [2009–2015], which I am currently re watching for the second time, as that too is a quirky NBC sitcom. Second, Netflix has a rating system for its programming and I can give shows a rating from 1 to 5 stars. Netflix can use this and recommend me shows that are similar to the ones I give four or five stars. Third, you can also tell Netflix the type of genres you prefer. I like comedies, but I am not generally a fan of anime shows. Thus, Netflix will not recommend me something like that.
These posts also describe how the relevant content worked within the context of their lives. The participants give Netflix agency in their description of the processes used to analyze user behaviors. Through phrases such as “Netflix sees” or “Netflix computer does,” the participants document the work being done by the company to be a part of their user experience. This agency will be discussed in the coming sections.
However, some participants noted their confusion over the Netflix process and method. Seven of the 27 journals noted they were familiar with the general process, but still confused as to the specifics of how the system worked. Again, these entries connected the practices of the organization to the user experience.
I am not sure exactly how it works, but I assume there are little people inside of the TV writing down notes (haha). The notes that these elves take aren’t always the best, because sometimes Netflix go off in a whole different direction regarding programs that its viewer would want to watch.
In more appropriate terms though, Netflix offers relevant recommendations because it uses machine learning to create and compare the data of watch and search history for all of its users.
The reflection on the process reveals that although contributors were never asked to address or describe how Netflix’s platform works, users are still aware and informed about narrowcasting’s presence and the methods used to achieve it. Overall, most of the participants reflected descriptively on the processes utilized by Netflix to tailor content selections. Importantly, most participant journals started with this description, which was then used as a point of entry to critique or analyze the success of these recommendations, which is described in the next section.
Finding Two: User Experiences
Similar to other studies on millennials and the media, journal entries in this study revealed that the group viewed narrowcasting in both positive and negative terms. However, these reflections were almost entirely focused on the user experience rather than larger implications of these trends.
FAVORABLE ANALYSIS
When discussing the positive traits of narrowcasting in Netflix, many identified recommendations as helpful features of the site, as they introduced users to new content or previously unknown films and television series. Two 18-year-old men and one 21-year-old woman added:
This is really helpful seeing as though the Netflix library is so large to sift through the whole catalog to find a program to watch would take time that our fast paced short attention spanned generation would not like to waste just looking for something to watch. This technique is very good for business. There is also a list of titles dedicated solely to the programs that our social media friends have been watching.
This is great considering that the best form of advertising for centuries had been word of mouth and personal testimonials. How great would it be to go to Netflix not knowing what to watch and seeing a list of things that your friends are watching? This creates intrigue and not only creates a larger list of recommendations for yourself, but also gives you something to talk about the next day with your friends.
Netflix and computer learning in general can help people to discover things about themselves that they
would probably never learn all on their own. Sometimes you need someone else to pick out your flaws to fix them, and then sometimes you need someone to pick movies and spam them along the “Recommended for [Name]” tab, the latter is Netflix. I can’t remember exactly what Netflix says about their computer learning, other than how simple recommendations make watching movies and shows. There’s not much more other than the facts I was dishing out earlier in this blog post. Netflix allows a lot of possibility for consumer to reach into the metaphorical bag and pull out a handful of excitement.
Again, the emphasis here is on the user, and how the system benefits the individual. Rather than looking at how this may influence broader society, eight of the millennial participants in this study reflected on how narrowcasting influenced them or the people in their immediate surroundings. As two 18- to 25-year-olds noted:
For me, these recommendations do fit my interest because it gives me options that I would like to see that I didn’t think of searching for. However, it also gives me new options that I never seen, but I usually have no interest in. I would say the recommendations have a 70 percent success rate on my use of the platform.
The benefits with this form of machine learning greatly outweigh the negative aspects. With the learning of your viewing preferences, Netflix basically takes the hassle out of search for something to watch. This type of computer learning is sold by providing a free trial for 30 days with upon completion requires a monthly subscription to continue usage.
These eight reflections ultimately focused on how the recommendations affected them. This is demonstrated through terms such as “I have” or “my interests.” These terms of reference are important as they denote who or what the millennial participants view as most important or critical to the Netflix platform. Previous research suggests that millennials often view their media use as individually, rather than socially impactful. However, while there is some evidence of this, there are also an equal number of journal entries that suggest greater society may also be impacted. These entries suggested that narrowcasting allowed users to access content that they may have otherwise ignored, thus exposing them to new points of view. On a large scale, this practice would help everyone become aware or have the opportunity to be aware of diverse points of view. As one 23-year-old participant put it:
When I use Netflix, I do feel that the recommendations are fitting to my interests. I have found some of my favorite shows and series through the recommendations that Netflix suggests. I think the idea of computer learning is a good thing for society. On a positive note, computer learning makes it much easier for the user to find content that they already have a strong interest in.
Importantly, posts like the one above suggest that millennials view media effects as important to both the individual user and larger society. This is an important finding, as it provides insight into how millennials view the role and impact Netflix has both on the media industry and culture.
NEGATIVE ANALYSIS
Alternatively, 18 entries discussed the negative features of Netflix and the narrowcasting process. First, many added that although the system was based on learning from your own, recorded behavior, the recommendations made were far from perfect. Thus, recommendations were a distraction or took up more time from the user as they attempted to wade through the imperfect fits. Sample responses in this category include:
The downside is what we are limited to from recommendations. If we only watch what is recommended to us then we don’t see much of the other content that we may have an interest in, but YouTube doesn’t know about it. An example would be anomalies; things we don’t usually watch but occasionally search for.
In theory this system should allow a person to seamlessly have enjoyable content delivered right to them. However, there is a problem with the relevance of these recommendations. First, Netflix asks you to rate content that you have viewed, assuming someone does rate every piece of content than I suppose the machine would be able to learn more about what an individual likes and be more accurate with these content recommendations. However, many people do not rate this content begging the question, how does Netflix know if you enjoyed what you have seen. I could watch ten awful pieces on Netflix and the site would just recommend other content I would dislike. Second, no matter how much the system tries to predict a person mood it cannot. Machine intelligence does not have emotions and simply cannot understand them. Maybe, I watched a romantic comedy because I was with a girlfriend and felt happy, and maybe I watched a break up movie because I was depressed because I had just had a relationship end. Those situations that evoke two emotions that are on the opposite ends of the spectrum are based of human situations that are semi uncommon in nature. Thus, the computers generations are probably going to have a somewhat skewed interpretation of an individual’s interests or desires.
For me these recommendations do not fit my interests too much because I am not consistent with my selections. I do not us Netflix too often, and if I do my selections do not make too much sense. I will jump from watching movies like lord of the rings, to movies like happy gilmore [sic]. Also since i do not watch tv that often by myself, when i do watch it i usually have other people with me, and their opinion gets taken into consideration on my account, which would be inaccurate because its not me who is choosing the movie.
Importantly, the reflections on the negatives of narrowcasting and Netflix were notably longer than the entries that framed the practice as a positive. This will be further analyzed in the discussion section.
Next, there were journals that viewed the narrowcasting as a part of larger trends, such as privacy, artificial intelligence, and computer learning technologies. Netflix was then critiqued as having potentially negative ramifications on overall society, not just the user. One 19-year-old participant noted that
while this computer learning may provide some convenient and useful elements to any application that utilizes such programming, there still remains the matter of privacy and identity. While computers analyzing human thought and behavior could lead to positive ramifications for education and ease of access, there still exists the common conception of computers learning too much and invading matters of privacy and security. While this may not be a matter of significance in regards to Netflix, other applications concerning more personal files and information bring these issues to light. Sensitive information becoming automated—while many companies reinforce their security—is still at risk of being compromised in some way.
Here, it is clear that the millennial participants view Netflix’s practices as tied to larger social and media industry trends. As Netflix continues to recommend personalized content suggestions, the users describe their growing awareness of their data being used for overall negative purposes (despite the increasingly specialized content).
There are also larger connections to the future of society and how narrowcasting could end up hurting the population over an extended period of time. Again, these reflections offer insights into both millennial participants and the perception of the future of technology.
This artificial intelligence provided, I believe, will end up hurting our society but not before it helps our society first. When this AI comes to fruition where it is everywhere and anywhere, this will create more ease in our everyday lives basically because it is like having your brain in a computer, allowing for your thoughts to be captured perfectly. The problem that could happen with this AI technology is the lack of any need for human interaction or need for self-thinking. With the lack of self-thinking, the AI might then be able to overtake the whole of thinking for the user instead of the user generating their own content.
I am usually an optimist, but I’ll say that computer learning is bad for society. From just observing how YouTube operates, advertisers seem to benefit the most from computer learning. Advertising bothers me especially when they can target my interests. It’s bad enough that old media platforms like the newspaper, television, and radio are flooded with advertisements,
but now advertisers can possibly use computer learning as a tool to use through new media platforms.
These quotes demonstrate that many of the millennial participants viewed the practices of Netflix as related to larger social issues such as artificial intelligence and machine learning. Both Netflix and these practices are criticized for their impact on both the individual user and society, despite their success at providing personalized recommendations. This juxtaposition will be explored in the coming sections.
Finding Three: Future Predictions
The majority of participants thought that Netflix’s practices were just the tip of the iceberg of the future potential of narrowcasting, algorithms, and user-based data mining. These practices are linked together through the lens of the media industry and the users’ Netflix accounts.
I think computer learning will continue in the future with the growth of artificial intelligence in more places.
I don’t see how this learning is either good or bad. I think this is an on going case study for what people like since moods vary and peoples [sic] decision making for the next video varies. In terms of time, yes this can be time consuming since watching one video could turn into watching five videos or even twenty videos. Some benefits are that your searching is already done for you and you can find something within that subject that you didn’t know existed under a different title.