I log onto Netflix on a pretty regular basis. Today, the home screen boasts: 44 TV shows and movies added in the last 4 days. The featured image is A Series of Unfortunate Events, a Netflix original show based on a book series I loved in elementary school. Add to My List? Netflix tempts me. I take the bait. I click. Add to My List!
As I scroll down, there’s a rotating menu of Top Picks for You. Interspersed among Bollywood movie recommendations, I spot Netflix original series The Crown (because you watched Downton Abbey), Gilmore Girls: A Year in the Life (because you watched Gilmore Girls...all 7 seasons...thrice, don’t judge), and Black Mirror (because it’s trending with your demographic).
Netflix knows me. The buzzworthy shows that appeal to my social circle, the Bollywood movies I watch with my mom, the guilty-pleasure reality shows I’d rather not publicize in this blog post. And every time I watch – or don’t watch – a show its algorithm recommends, I help Netflix know me just a liiiiittle bit better.
In this week’s Raw Data episode, we’re taking a deep dive into the ways data is toppling the power balance between traditional media producers like Sony or Paramount and online distributors like Amazon and Netflix.
Netflix isn’t the only online video platform that’s using a treasure trove of consumer data to inform creative decisions about original content, but it was undoubtedly a pioneer in the data revolution that’s shaking up Hollywood. And this disruption is creating a lot of excitement in the entertainment industry as an outlet for new creative opportunities.
“They [Netflix] are presenting a mecca and a beacon of creative freedom and power of distribution for all storytellers right now,” independent film producer Steven Berger tells us. Steven runs his own company, and while he hasn’t collaborated with Netflix directly (at least, not yet), he told us they are definitely the talk of the town.
Once upon a time, there were only a handful of primetime TV slots – so networks had to pick content that appealed to the greatest number of people. In other words, time slots were a scarce and valuable resource. But with video on demand, more and more streaming services, and direct access to viewer data, these tech companies can target audiences with niche content – and create new shows that appeal to specific groups.
“Because these great storytellers can find exactly the right audience, they're going to be able to tell different types of stories,” said Michael Smith, a professor at Carnegie Mellon University. “They're going to be able to tell a type of story that just wouldn't work in traditional broadcast channels.”
On this episode, Smith gives us insight into the recent changes in the entertainment industry, especially how data-driven decisions coexist with creative decisions about casting and storytelling. We spend time exploring a book that Smith co-authored with Rahul Telang titled Streaming Sharing, Stealing: Big Data and the Future of Entertainment. We also talked to a data scientist at Netflix, Todd Holloway, to get the inside perspective.
Streaming platforms know so much about me that they can make my leisure time more enjoyable. I don’t have to check whether the recent Hindi movie Udta Punjab is on Netflix yet, because as soon as it’s available it’s recommended to me. At the same time, these companies are fundamentally changing our entertainment on the creation side. YouTubers filming in their bedrooms and living rooms are becoming full-fledged celebrities creating high-quality video content. Online media distributors are giving millions of dollars to niche-interest TV shows and revivals that otherwise may never have seen the light of laptop screens worldwide.
We’re in the midst of a dramatic shift in power that the Hollywood studios didn’t see coming, and perhaps have been slow to respond to. Companies that once relied on major studios to create content in order to distribute DVDs by mail are now racking up awards at the Emmys and Golden Globes. Streaming platforms seem to be taking the advice of House of Cards protagonist Frank Underwood: “If you don’t like how the table is set, turn over the table.”
So happy streaming, and we’ll be back in a couple weeks with an episode on propaganda armies. Stay tuned!
MICHAEL C. OSBORNE: Hello Raw Data listeners. Before we get started with today’s episode, Leslie and I wanted to say a few words about where our show is at and where it’s going. We’ve always wanted Raw Data to be a space to explore how data and digital technologies change the relationships between people and the institutions of society. It’s becoming harder and harder to know what sources of information we can trust. And one of the reasons we run this show out of Stanford is that, from our perspective, the need to share science, research and expertise has never been greater. It’s our view that universities should play a more active role in making knowledge accessible to the world.
LESLIE CHANG: So Mike and I have been thinking hard about how best to deliver stories that offer insight and clarity in these confusing times. And in future episodes, we hope to do just that. In fact, if you have ideas for topics you think we should cover or people you think we should interview, please get in touch – email us at raw data podcast AT gmail DOT com. In the meantime, we have an episode about a decidedly lighter subject that we’ve been wanting to do for a while now: big data and entertainment. We recognize this topic doesn’t address the problems that we’re now facing with information and misinformation, but on our next episode, we promise we’ll get right into the thick of it. For now, we hope you enjoy this story. Thank you as always for listening and joining us on this learning journey.
[Raw Data theme music]
LC: Welcome to Raw Data. I’m Leslie Chang.
MCO: And I’m Mike Osborne. Today’s show: A New Day in Hollywood.
[theme music continues]
MCO: So first, it's late January, but it's not too late to say happy new year. Happy new year, Leslie.
LC: Happy new year, Mike.
MCO: Nice to be back. I can only do this for a couple more days before its February. So on our last episode we talked with Tim Wu about attention and the idea that attention is sort of the currency of the Internet, it's what people are fighting over. I really liked that framing of like attention is money-
MCO: Essentially. Leslie, can you tell us a little bit about what that interview was about and where it went.
LC: Yeah so I basically talked to Tim about how the Internet runs on clicks, and shares, and time spent on webpages. These are really ways to kind of quantify how much attention were giving to something on the web, right. Because so much of the Internet runs on advertisements. They need to know how much you're paying attention, how much you're clicking through, but-
MCO: So when you say so much of the Internet ... I mean ... You mean-
LC: I mean like Google, Facebook, all of these are ad-supported, right. Like some of the biggest sites are all ad-supported.
MCO: Its become the predominant business model of a lot of tech companies.
LC: Yeah absolutely, and towards the end of the conversation we kind of got to this point where we were talking about, is there a better Internet? Is there a better business model that we could look at and Tim really pointed us in the direction of Netflix. Which, first of all is fun to talk about, right, because it's TV, and movies, and films, but he was saying that they have really figured out how to get people to buy into the subscription business model and through that they're able to license some really great shows and now get into making new shows.
MCO: Yeah, can I just say a word about TV.
MCO: I love TV.
LC: I know you do.
MCO: Yeah, no I don't-
LC: No, I know you do.
MCO: I don't know if I'm expressing to you just how much I love it. I was thinking about this the other day like, I don't know if street cred is the wrong word to use, but I definitely evaluate new people in my life a little bit based on their TV tastes you know-
LC: Yeah, It's like a good entry question.
MCO: Totally and when somebody tells me about a show they're watching and I go home and watch it, if I like it, like I'm endeared to that person and if I don't, then it's like, maybe they're not for me, you know. Like it really is sort of social currency in a way, am I taking it too far?
LC: No, I get it, it's like TV is a really easy entry point to kind of get a conversation started.
MCO: Totally and the thing is though, you mentioned a second ago, that you know, Netflix is now doing original productions. So is Amazon, and have even started getting nods from the Academy Awards.
LC: Yeah, Oscars, Golden globes-
MCO: Golden globes, exactly ... Emmy's, and my perception as a fan of TV, as a TV viewer, is that like Amazon and Netflix are kind of the cool places to be doing TV shows these days. I mean you see a lot of big stars and big directors launching television shows on platforms that are funded, these companies are historically tech companies. But we wanted to know, is that really true? I mean what's the word on the street, and of course you and I don't really know too many people in Hollywood.
LC: Not really, but we made some calls.
MCO: Yeah we have friends of friends, and eventually we got put in touch with this guy, Steven Berger. He runs his own production company and he's involved in a number of different indie projects.
STEVEN BERGER: I am an independent producer which means that I don't work for a studio. I don't work for somebody else's company. I started my own company right out of grad school. I find material and filmmakers. I develop scripts, I raise private equity to make those movies. I sell them for distribution and basically just help people get their stories told, get them made, and get them in front of an audience.
MCO: We asked Steven for his take on how studios in Hollywood [the Hollywood establishment] are reacting now that tech companies like Netflix and Amazon are making their own films and tv shows.
SB: I think that the perception within the business is that they are being radical disruptors. And...I would say that they're bringing the movie business back around to what it originally was, which is that you can walk into a place with an original idea that doesn't have to be a comic book. You can find an executive that's really excited about, that has the power to say, "Go and make this thing that's not reverse engineered based on an international marketing plan." You can go and make the movie that you want and get it in front of an audience. So I think that while the studio system has their crosshairs on Netflix and Amazon, I think they're closer to what the intention of the studio system always was.
LC: You sound excited about that.
SB: I'm extremely excited about it. I don't only want to go and see comic book movies. I don't only want to go and see movies that are the 12th sequel to an original movie that was made 30 or 40 years ago. I think Netflix offers a really great alternative to what traditional studios are doing.
LC: Our conversation with Steven echoed a lot of what we were hearing. Creative people seem to be genuinely excited about the opportunities at Amazon and Netflix. We wanted to get a high level view of the industry, and our awesome intern Isha told us to check out a book that came out last year called “Streaming, Sharing, Stealing: Big Data and the Future of Entertainment.”
[Music: “Go ‘n’ Drop (2003)”]
The book outlines a lot of the economic and technological forces shaking up the entertainment industry. We decided to reach out to one of the co-authors.
MICHAEL SMITH: I'm Michael Smith, I’m a professor of information technology and marketing at Carnegie Mellon University. My colleague, Rahul Telang, and I do a lot of research on how technology is changing the entertainment industry.
MCO: Their book is a fascinating dive into how the industry is adjusting to the digital age. A lot of things are in a state of flux. Audiences are changing their behavior – for example, there are a lot of cord-cutters – people who are cancelling their cable TV subscriptions because they’re able to stream movies and shows from places like Amazon, Netflix and Hulu – or from sites hosting pirated content. These new options have massive repercussions for the traditional power players in Hollywood.
LC: It’s worth taking a moment to set the scene here – it turns out there are actually just a handful of major studios that make most of the movies and TV shows you’re familiar with.
MS: One way to think about the studios is they're almost like venture capitalists. They got a big pot of money and they place that money on different startups, if you will, different movies or different projects. Most of them fail and their hope is that a couple of them are going to go on to be huge successes.
MCO: But in the past few years, the economics of all this has started to change. Cameras and equipment have gotten better, so productions have dropped. On top of that, people are going to the movie theaters less, and there’s a ton of pirated content available. When you start to put some of these factors together, having a state-of the-art movie theater or holding once a week spot on primetime is just not the same scarce resource it once was.
MS: What’s scarce today is customer attention and information about the customer. And we're seeing the rise of powerful players like Amazon, Apple, Google, Netflix who own the customer, own the customer data and who are using those skills to get into the business of making content.
LC: The entertainment industry is used to dealing with disruptive technologies – think about the 1950s when families first started buying televisions, or the 80s when VHS hit the scene, and later DVDs in the 90s. Throughout all of this, the business structure with just a few major studios has remained intact. But today, the forces of change are coming from several directions.
MS: You can look at these technological shifts individually and they don't look like much of a threat. The rise of user generated content, digital piracy, big data, powerful downstream distributors, when you look at them separately, they don't individually look like a problem. But when you look at them together, it's a perfect storm of change that is fundamentally shifting power in the industry.
MCO: Big data is really at the heart of all this. Companies like Netflix, Amazon, Apple, and Google are able to collect information directly about their audiences. And somewhere along the way, Amazon and Netflix looked at the market and realized they could and should make their own shows.
LC: Back in 1997, when Netflix first started as a company, they rented out DVDs by snail mail. But when streaming came along and they wanted to offer content on demand, they had to get more involved negotiating licensing rights with the studios. This experience offered them insight into the power of owning content.
[House of Cards theme music]
MCO: All of this was going on behind the scenes, but the turning point was in 2013 when Netflix released the political drama House of Cards, starring Kevin Spacey and Robin Wright.
MS: House of Cards was a show that Media Rights Capital was shopping around to a bunch of traditional broadcast networks. It's obviously a show about D.C. politics, it's a show with a fairly complicated plotline. What they were hearing back was, A, no one in the industry believed that D.C. politics was going to work. The conventional wisdom was there hasn't been a successful show since West Wing so people aren't interested in D.C. politics anymore.
MCO: Another problem was that they wanted to develop a fairly complicated story, but usually shows get picked up by the strength of just one episode – the pilot.
MS: And David Fincher, the director of House of Cards, and Beau Willimon, the head writer for House of Cards, both looked at this and said, "No, the story we want to tell is not going to be easy to encapsulate in 42 minutes, it's going to be tough for us to lay out the story arc."
LC: Mike and I dug into the origin story of House of Cards a bit, and we read an article in the Harvard Business Review. According to this article, Media Rights Capital was pitching House of Cards to cable networks like HBO, AMC, Showtime, and FX. Media Rights Capital had also been in communication with Netflix, but they were thinking of Netflix only as a secondary revenue stream after the show would have appeared on cable.
MCO: But to their surprise, Netflix came back and said they wanted House of Cards to be one of their first original productions. This was a risky move for Media Rights Capital. But apparently Netflix put on the hard sell.
MS: What Netflix did is Netflix said, "Hey, you know what? We think we have an audience who's going to be interested in this show and we think we can help you find exactly the right audience for this show. We're not going to ask you to make a pilot episode, we're going to give you 100 million dollars upfront for a two-season, 26-episode order," which really was unheard of in the traditional broadcast industry.
LC: Now, you might think that Netflix wanted in on the project because they looked at their user data and saw that Kevin Spacey and David Fincher bring in big audiences. That may be part of the story, but, Mike Smith says that doesn’t make sense – because lots of people in Hollywood know how popular Kevin Spacey and David Fincher are. It’s not some big secret. He told us that the real innovation was in how Netflix used their data about their viewers to essentially do targeted advertising.
MS: When it came time to promote House of Cards, Netflix created nine different trailers for this show. It created a trailer that was uniquely targeted to people who would like Kevin Spacey and so they drew out Kevin Spacey. It created a trailer that was uniquely targeted to people who like David Fincher movies and so they drew out the characteristics of his movies. It also created a trailer for people who like movies with strong female leads that focus more on Robin Wright. They were able to promote it in fundamentally different ways than the traditional studios could because they knew who their customer was.
MCO: Now, House of Cards is not for everyone. If you’ve seen it, you know that the very first scene of the show is pretty shocking. In fact, Mike Smith says it’s a bit of an industry inside joke. There’s a famous screenwriting book called Save the Cat, and the thinking goes that when you introduce the main character you should have them do something heroic like save a cat. House of Cards took a different approach.
MS: In the scene, Kevin Spacey's character comes out to a dog that's just been hit by a car, he explains his political philosophy and then he breaks the neck of the dog to put the dog out of its misery. [00:51:00] What the data tell you is that when Kevin Spacey breaks the neck of the dog, a whole bunch of people tune out of the show. So the thought question is why was Netflix able to include this scene when it wouldn't work on broadcast television.
LC: If you’re running a broadcast network and half your audience tunes out, you’ve basically wasted your scarce time slot. But, according to Mike Smith, people tuning out when Frank Underwood killed the dog, actually gave Netflix a useful piece of information.
MS: Netflix, however, the people who tuned out of House of Cards probably watch something else on Netflix. That even gave Netflix useful information about them, these people like dog strangling and these people don't, I can use that to promote content to them in the future.
MCO: Yeah. I love that point especially, right? I mean, it gets back to the granularity. Not only do we know or what you're watching because they control the distribution platform, they know what time of day, they know at which point in the story you tuned out, they know what you went to next. If you saw something horrific, did you go and watch cartoons afterwards or did you walk away from Netflix for a month and a half. I mean, there's so much inferences to be had.
MS: Right. Or here's another fun theory, you put the kids to bed and then you came back and you watch the show later. You discovered, "My kids shouldn't be ..."
MCO: Hope they didn't have nightmares, yeah.
MS: Exactly, yeah. There are a whole bunch of fun inferences you can take with that information.
LC: We knew that Netflix places a lot of importance on data science, so we wanted to talk to someone there who works with user data.
[Music: “23 Digibass”]
We met Todd Holloway, he’s the director of content science and algorithms at Netflix.
TODD HOLLOWAY: And on my team, on content science, as the name suggests, we look for ways to use data science to help Netflix have the best catalog, have the best content. On a day to day basis that mostly means we're building machine learning models.
MCO: Todd emphasized to us that Netflix sees data science as integral to the entire company. But he also said that the central role of data does not threaten the creative side of filmmaking.
TH: The data science always plays a supporting role to the creative freedom and making the best possible content for the site.
As far as making actual creative decisions using data, I do think that people sometimes have an exaggerated sense of what we're doing at Netflix on that front. I'm not aware of any casting decisions, for example, that were ever based on data science. Changing the creative decision because the data doesn't back it up. I'm actually, again, not aware of a single incident of that.
MCO: The idea that data science supports rather than inhibits creativity may sound counter-intuitive. But this was actually a point we heard over and over again, including from both Mike Smith and independent producer Steven Berger who we heard from at the top of the show.
[music fade out]
SB: There's a certain level of creative freedom that they offer that is becoming a little bit more notorious around town. There's the cliché of getting studio notes of people's shoes being the wrong color. Traditionally, Netflix and Amazon have been pretty hands off in terms of trusting the creatives that they bring on board to execute their projects.
MS: I don't want a great storyteller like David Fincher, I don't want people whispering in his ears saying, "Hey, you know what? The data tells us that people like bunnies so you need more bunnies in the third scene." I think that will be bad. I also don't think that's what we're seeing.
LC: We wanted to know more about how else Netflix is using data science. If the power of their platform really comes from being able to personalize the user experience, then what other kinds of datasets are they interested in using? Here’s Todd again.
TH: In terms of external datasets, right now I'm excited about something else even more and that's datasets that help us develop a better thesis on the globalization of content on global taste. So Netflix, right now, is available in almost every country on the planet. Places that I don't or the individuals here, we don't have good intuitions about their taste, and so we're really looking to the data and external datasets to help us develop that intuition. There are so many different cinema traditions around the world. There's Nollywood, which is the Nigerian Film Industry, there's Anime, Telenovelas and so on. As a company what we want to do is have a really good understanding of what content travels well as well as what local content might only have local appeal but bring extreme joy to those localities.
LC: It makes sense that Netflix is focusing its efforts internationally. After all, the company uses a subscription model – so growth is based on acquiring new viewers. And right now, Netflix seems to be doing pretty well in that area – about a week after we talked to Todd, news came out that they had added 7 million new subscribers in the last few months of 2016… 5 million of those new subscribers were outside of the US.
MCO: Just to pause on this, because it’s worth calling out – this is a totally different business model for the entertainment industry. Netflix doesn’t rely on traditional metrics like box office sales and Nielsen ratings in order to sell ad space. Obviously the company cares which of its shows are doing well, but it doesn’t matter if all their new subscribers are coming to see one show, like the latest season of House of Cards – or if different people are coming to see The Crown, Stranger Things, or Bojack Horseman, or Fuller House,, Narcos, whatever show you may have heard about lately. It’s all about new subscribers.
LC: Which all comes back to the advantage of user data. Traditional studios are now in this position where their business model increasingly relies on blockbusters, but Netflix and Amazon have figured out how both blockbusters and niche content can bring in new subscribers. That’s one of the reasons we’re seeing so many new shows.
MCO: So far we’ve been mainly focusing the conversation on Netflix but, of course, they’re not the only company diving head first into original productions. After we talked to Todd, we reached out to Mike Smith again for a follow-up interview – we wanted to hear his thoughts about what all this means and where the entertainment industry may be heading.
MS: "We're not arguing that Netflix wins here." Netflix is facing some really significant competition from Amazon and from Google and increasingly from Apple. Each of those companies are companies that also know their customer at a very detailed level, and they have these other very interesting attributes. Amazon not only knows my viewing behavior. It also knows all of my purchase behavior. Google not only knows my viewing behavior on traditional stuff. It also knows everything that I watch on YouTube, and it has all of my search behavior. iTunes, again, knows an incredible amount of my purchase behavior through the iTunes store. So we're seeing this incredible competition between these four very different companies, who are all going after the same space.
LC: Right now, with so many tech companies getting into the game, it feels like we’re never going to run out of great stuff to watch. Mike and his colleague Rahul make a strong case. But do the studios see things the same way? How is the establishment interpreting the changing landscape?
MS: A lot of people in the traditional industry look at the huge increase in creation, particularly television content, and make the argument that we're in a content bubble, and, in fact, that a lot of people are going to get hurt when the bubble pops. There are going to be a lot of people who were working on projects who aren't going to get paid, and it's going to be bad for everyone. I understand that argument. I wonder, however, whether the idea of, "We're seeing way too much production," is based on the old rules of the business, where you can only show a certain amount of shows on NBC in the primetime hours, and whether, again, Netflix and Amazon and Google are playing by a different set of rules. So this might actually not be a content bubble. This might be the new normal of what is possible in the digital data-driven world.
MCO: Regardless of whether we’re in a bubble or whether this is the new normal, we should all enjoy this golden age of TV. But as Leslie and I were working on this episode, we kept asking ourselves, “Is there any big issue here, anything problematic that we should be worrying about?” And, yeah, certainly there are some privacy concerns with the collection of user data, but honestly the answer is “no, big data meets entertainment is mostly a good thing.”
LC: One thing that did come up, though, is that maybe there’s a risk of being in a content filter bubble. This is a theme we’ve talked about before on our show – with algorithms choosing what information we see, there’s a risk of NOT knowing what exists outside of our personalized universe. When it comes to entertainment and movies and TV, maybe it’s not that big of a deal. But this issue takes on a whole new layer of importance when it comes to the news, and real world events. And if you want an interesting case study on what this looks like on a massive scale, China is a good place to start. We recently talked to Jen Pan. She’s been researching how exactly the Chinese government censors people online.
JEN PAN: I would say that censorship in China is part of a broader effort to control the spread of information, ultimately to shape what people believe and what their preferences are. And...
when you're on social media in China, there's a huge amount of volume of content, so you wake up every morning. There's more than you can ever consume, so why would you then seek out additional information? You might very well perceive there to be plenty of information and no need to try to obtain additional information.
The hard thing is how do you try to get people to know what they don't know because they think they already know. When that's the case, they're not as interested in uncovering censorship or figuring out what propaganda is.
MCO: Propaganda armies around the world, and here at home. Next time on Raw Data.
[Raw Data theme]
MCO: Thanks for listening to our show. If you like what we do, please share the podcast with your family and friends, email it to people, post it on social media. Every mention and share helps, and we really appreciate it.
LC: Our podcast is produced by Mike Osborne and me, Leslie Chang. Our awesome intern who knows good books is Isha Salian. Special thanks this week also to Jackson Roach, Miles Yim, Erika Beras, and Jimmy.
MCO: Professor Michael Smith’s book is co-authored with Rahul Telang, and it’s titled “Streaming, Sharing, Stealing: Big Data and the Future of Entertainment.” Again, the book is awesome, we highly recommend it, and they cover a lot more ground in the book than we were able to in this episode. We also recommend checking out Mike Smith’s TEDx talk, titled “Is big data killing creativity?”
LC: Thank you to Todd Holloway and the Netflix team, as well as Media Rights Capital.
MCO: Our show is a production of Worldview Stanford, and we’re receive additional support from the Stanford Cyber Initiative, whose mission is to produce research and frame debates on the future of cyber-social systems.
LC: Again, please get in touch if you have any feedback or ideas for us. We’re on Twitter @rawdatapodcast, or you can email us – raw data podcast AT gmail DOT com. Thank you so much for listening, and we’ll be back with a new episode very soon.
MCO: We'll let you know what happens from here but I'm very optimistic. I learn a ton in this conversation. I really had a good time talking to you.
MS: Cool, yeah. If you could send me a link once it goes live, that'll be awesome. I'm just still trying to impress my mom and dad.
MCO: Yeah. That never goes away from what I can tell, I'm with you. Awesome. Thanks again, Mike.
LC: Thanks, Mike.
MS: All right, guys. Thank you.
Steven J. Berger, Michael D. Smith, Todd Holloway
A Stanford podcast about how big data and cyber systems are transforming our world. Produced by Worldview Stanford and supported by the Stanford Cyber Initiative.