Why Does Netflix Keep Recommending Adam Sandler Movies? I Have A Theory

Post by: 23 November, 2020

If you spend any time watching Netflix's Top Ten lists, you'll notice that it's common to see weird and sometimes obscure movies suddenly appear on the lists. Some random Adam Sandler or Mark Wahlberg film will unexpectedly be the 3rd most-watched film and there is often no discernable reason for it to happen. It's clear Netflix's suggestion engines are surfacing the title for users. But why?

Stereogum has a really fascinating look at Spotify's recommendation engine and the story starts with a simple question: The Brighten The Corners album outtake "Harness Your Hopes" is the #1 song on Pavement's Spotify page, with over 29 million plays to date, eight million more than "Cut Your Hair," a legitimate and enduring '90s hit. Quickly, and without any obvious reason, it stopped being a rarity and started to become a standard. So how did this happen, exactly?

Online, people have been casually wondering this on places like Reddit and Twitter, with a prevailing theory being that the song must have been featured on a prominent Spotify playlist, and then simply snowballed from there. Malkmus himself was under this impression, too: “I heard it was on a playlist or something,” he says, nonchalant. “I’m not an expert on Spotify but, you know, one of those ‘Monday Moods’ or whatever the fuck they do.”

It’s a reasonable enough explanation. But looking at a similar situation of his own, Damon Krukowski wasn’t so sure. The musician and writer was fascinated with the question of how “Strange” became his former band Galaxie 500’s top Spotify track — by a significant margin — even though it was not a single, was never particularly popular in the past, and wasn’t being picked up on any prominent playlists. In June of 2018, Krukowski laid out the conundrum on his blog, and soon he received a possible explanation from a Spotify employee.

Glenn McDonald, who holds the title of “data alchemist” at Spotify, had taken an interest in the case, and decided to look into it. What he found is that the sudden jump in plays for “Strange” began in January of 2017, which was “the same time Spotify switched the ‘Autoplay’ preset in every listener’s preference panel from off, to on,” as Krukowski recounted on a follow-up blog post. McDonald explained to Krukowski that the Autoplay feature actually cues up music that “resembles” what you’ve just been listening to, based on a series of sonic signifiers too complex to describe. In this case, “Strange” had been algorithmically determined to sound similar to a lot of other music, and was frequently being Autoplayed to the point that it took on a life of its own, and eventually eclipsed the band’s other tracks. It continues to do so to this day.

It's well-worth reading the entire piece and while Spotify refuses to discuss the issue, it's probably not a coincidence the service is beginning to allow bands to select which songs get chosen for this unintentional attention. 

This is the type of insight you can only get from inside the company and Netflix is notoriously shy about discussing its internal recommendation engines. But this Spotify story makes a lot of sense when you put it in the context of Netflix. These lesser, often obscure movies are being recommended not because they are great, but because they are bland enough to match up closely with other popular movies subscribers have previously viewed. 

Which gets us back to those Adam Sandler movies. Critics might not love them and some of them might feel more like a cast cashing a check rather than working towards the punchlines. But the movies are mainstream and if Netflix's recommendation engine works anything like Spotify's, Sandler's movies are going to get a lot of attention. Even if you're not looking for them. 

We tend to think of recommendation engines as being complex, manipulative creations that are driven entirely by design. But as this Spotify story illustrates, even the most carefully designed piece of software can produce unintended consequences.

Last modified on Monday, 23 November 2020 17:09