The success of Netflix was ensured by high technology, but behind them there is a whole philosophy that made this philosophy effective. A system that makes millions of people obsessively click red and white buttons, easily abandoning the long tradition of watching movies in movies and series on TV.
Hello! Yefim Gugnin is with you! And today we will try to figure out how Netflix works. And for this we have to rewind time a bit.
1997 The Internet is only just gaining popularity and so far connects a measly 10 million computers.
People watch all the series and films in the cinema, on cable or terrestrial TV, or they order video at the box office. In this difficult time, 37-year-old Reed Hastings, a former military and Peace Corps worker, now a computer engineer and entrepreneur, decides on an adventure.
Together with his friend Mark Randolph: he opens a DVD rental business and names the company Netflix.
Why adventure? Firstly, the video rental market is almost solely ruled by Blockbuster, which is clearly in no hurry to leave its leading position; secondly, at the time of Netflix's launch, the DVD format has only been around for a few months, its commercial success is not obvious, and most families still use VHS cassettes. And finally, instead of giving customers the old-fashioned way to rent films in video rental, Hastings decides to work exclusively through the Internet and deliver discs to their homes.
And although now it seems to be an ordinary and in all respects a comfortable solution, many businessmen in 1997 frankly did not understand it: they say, why should a client surf some network and order discs home if he can just get up from the couch to the nearest point "Blockbusters", which at that moment were literally everywhere.
But unlike them, Hastings understood a very simple, but important truth for the coming digital age: people do not like to get off the couch.
You know what people love even less than getting up off the couch? Fill out paperwork, sign contracts, calculate the cost of couriers and mail, nervously mark the days on the calendar so that, God forbid, the disc rental is overdue and receive a disproportionately large fine in addition to the usual fee.
All these minor, seemingly obligatory troubles that other companies simply did not pay attention to, taking them for granted, Netflix wanted to minimize to the simplest set of actions: clicked - received - looked - sent back.
Netflix took over all the paperwork, and in 1999 also changed a one-time rental to monthly unlimited subscriptions. Now the client, with all his desire, could not delay the lease of the disk. Unless to steal it, which, oddly enough, in the nineties-zero criminals actively traded.
Many people met the idea of an unlimited subscription, to put it mildly, with skepticism. Analysts predicted that the system would be unprofitable, and not only because of the treacherous "disk stealers". After all, customers, if desired, can rent so many films that they simply exceed the cost of a subscription.
It was not so difficult to do this, especially since the delivery from Netflix worked quite quickly. Rather than ship everything from a central office like other companies, Hastings astutely located offices throughout America and later Canada so that a person from any state could get discs in the shortest possible time.
And this time, the fears of skeptics seem to have almost come true: some clients really exceeded this unwritten limit. Against such desperate domestic "cinephiles", however, they soon came up with a protection system, not particularly honest, but absolutely legal - it is known as Throttling.
Its essence is as follows: if one copy of, say, “People in Black” remained in the Netflix warehouse, and two clients ordered the picture at once, then the package was sent to the one who rents less often. And the second had to wait until the package arrived from another branch of Netflix.
But even this system did not save the company from losses: by 2000, Netflix began to gradually lose profits. At some point, the owners of Blockbuster even tried to buy the company. They never came to an agreement. Blockbuster disagreed with Hastings' $50 million price tag and, of course, made a fatal mistake.
However, one fast delivery and unlimited subscription was not enough to take a strong position in the crowded video rental market, and since the beginning of the 2000s, Netflix had competitors who mastered the Internet and remote communication with customers. Even the conservative Blockbuster launched its own delivery service at one point.
So what made Netflix stand out from them? First of all, content. Netflix's competitors haven't given much thought to their movie library.
For the same Blockbuster, as much as 70% of the catalog consisted of new releases.
Netflix has only 30.
The rest is cinema of any era and genre, from silent classics to strange "grindhouse". The company sought to cover all tastes, including the most specific ones, and constantly expanded the library. If at the time of the launch there were only 952 pictures in their catalog, and then there were no more on DVD, then by 2005 the number had increased to 35.
But the main thing that distinguished Netflix was the algorithms. A complex chain of microservices and small independent programs to customize the site for each specific user. They select those films and series that are more likely to please the client so that he comes back again and again.
It's essentially a digital replacement for those poor, bored rental workers trying to decipher the confusing desires of customers.
The first such algorithm, Cinematch, was successful in 75% of cases. Lucky in the sense that the recommended film received ± half a point from the rating of the film on which the recommendation was based.
Cinematch took into account several factors at once: first, the films themselves, sorted by genre, release year, directors and actors; the second - the ratings of an individual client, a list of his rented films and those that he has queued; and finally, the third is the overall rating of all Netflix users.
This system helped Cinematch avoid platitudes like if you put Pulp Fiction and gave it a high score then Reservoir Dogs would be recommended to you. Cinematch built much more distant parallels based on the ratings of other users and often produced results that were not at all obvious.
It works like this: let's say you ordered and highly appreciated "Pulp Fiction"; Cinematch finds other people who have also given Pulp Fiction a high rating; further Cinematch finds out what other films these people gave high marks and, for example, suddenly discovers that several people at once gave high marks to "Babe the four-legged baby."
It compares the number of these people and calculates the likelihood that you personally might like "Babe". Having finished, he continues the chain, tying together more and more seemingly not very similar films. Cinematch was equally successful in recommending classics, blockbusters and independent films, not limited to recommending only popular films.
Of course, it is based on dry mathematics. But, unlike other similar systems that fixate on popular names and genre affiliations, it rather mimics something like word of mouth, a chain of perfectly live recommendations from people with similar tastes.
And, according to a New York Times article, this system has resulted in people across America paying much more attention to independent releases, including those that didn't make big money in theaters.
For example, it is believed that it was Netflix that saved Francis Ford Coppola's The Conversation from oblivion. It was impossible to find it in video rental stores and it was not played on TV.
But Reed Hastings 75% was not enough.
He's generally not one to settle for "good enough," and in 2006, Netflix launched a competition to create a new referral system with a $1 million prize pool.
In 2009, the BellKor`s Pragmatic Chaos team won the prize - their algorithm worked better than Cinematch by as much as 10%.
Now the Netflix algorithm works not only with user ratings, people with similar tastes, but also with its demographic data - gender, race, place of residence. Moreover, he adjusts everything to the Netflix client, right down to the posters of his films and TV shows.
So, recently, the Skynews portal raised a scandal: they noticed that Netflix shows black users special adapted posters in which black heroes are brought to the fore, even if they play very minor roles in the film itself.
Accordingly, the more you use Netflix, the more data the site collects about you and the more personalized its main page looks. Creepy? Perhaps. But apparently it works.
In 2007, Netflix announced a beta test of The Watch Now, a new service that offered customers to watch movies online instead of renting DVDs. This is the direct ancestor of the Netflix as we know it now. Unlike many competitors who were baffled and bankrupted by the popularity of the internet, as Blockbuster did, Netflix was ready for change.
In order to simplify the viewing process and create a recommendation system, Netflix has embraced the possibilities of the digital age from the very beginning. Now you don't even need to get up from the couch to take delivery.
Distribution centers across America have evolved into networks of servers that quickly stream movies from a huge catalog to anywhere in the world, all in the highest definition.
Advanced algorithms make millions of recommendations every second, so that customers can’t stop and watch, and watch, and watch, and watch, and watch…
Since 2013, Netflix, which already had experience in creating original films, began to produce original content for its service. The first series to proudly bear the Netflix Original title was House of Cards, an adaptation of the 1999 BBC series of the same name.
The choice, by the way, is also partly based on dry mathematics: the thing is that the original British mini-series was very popular among Netflix users. In 2015, Netflix began a collaboration with Marvel, which, however, has already ended. In 2016, after a rebranding, the company finally went beyond the continent and started working in 150 countries around the world.
In 2017, Netflix bought out Mark Millar, the author of the Kick-Ass and Kingsman comics, Millarworld. In 2018, he signed a deal with Paramount to create joint films and in the same year received his first Oscar nomination for Roma for Alfonso Cuarón. In 2019, most recently, Netflix signed an agreement with Dark Horse to create series based on comics.
Obviously, the company is not going to slow down: since 2013 they have released 249 films and more than 400 series, not counting those for which they bought the rights after production. They have dozens of times more in their plans, including new films by Martin Scorsese, Ron Howard, Charlie Kaufman, Zack Snyder, Noah Bombach and, suddenly, Michael Bay.
Of course, Netflix still has a lot of unresolved problems: a strange regional policy, in which customers in countries like Russia pay the same as in the United States, but receive a tenfold smaller library of films and series; a large amount of frankly passing content, especially among feature films; finally, not a very adequate attitude to theatrical distribution - it’s still a shame that the same “Roma” and “The Irish” by Scorsese cannot be seen on the big screen outside the United States.
But to deny that Netflix, like its streaming competitors like Amazon Prime and Hulu, changed the film and TV industry forever is pointless. Content has become dozens of times more, competition has increased, small author's films finally got their way around cinemas, which had long been captured by big blockbusters. Shows no longer have to wait for their broadcast and cable bosses to give them an adequate spot on the screening grid.
And their influence is not limited to the creative part of the film process. Netflix, again, for purely marketing reasons, has been moving the technical part of the film industry forward for several years. So, in 2018, the company banned the authors who shoot original content for it from using Arri Alex cameras (for a second, the most popular in Hollywood).
And all because it supports a maximum resolution of 3,2K, and Netflix, according to the standard, needs all 4K. After much bickering and scandals, ARRI has released a new camera with a 4,5K sensor.
Netflix even requires its creators to use HDR and Dolby Atmos technologies, and the company recently launched the Netflix Post Technology Alliance program to develop filming technologies. Now all tools, from cameras to software, that meet the company's standards can use their logo.
It remains only to observe where progress will lead art this time. Maybe all author's cinema will soon completely leave the cozy halls of cinemas for streaming services. Maybe streaming will end the long-running era of blockbuster movies.
Or maybe something else will happen ... But it will happen for sure. In 2019, the number of Netflix viewers began to fall for the first time since zero, and the company is clearly not going to just sit and watch.
After all, sitting and watching is our job, the audience.
Thank you for staying with us. Do you like our articles? Want to see more interesting content? Support us by placing an order or recommending to friends, 30% discount for Habr users on a unique analogue of entry-level servers, which was invented by us for you: (available with RAID1 and RAID10, up to 24 cores and up to 40GB DDR4).
Dell R730xd 2 times cheaper? Only here in the Netherlands! Dell R420 - 2x E5-2430 2.2Ghz 6C 128GB DDR3 2x960GB SSD 1Gbps 100TB - from $99! Read about
Source: habr.com
