Wait — Tiny Netflix Elves Aren’t Responsible For Tailoring Watching Recommendations?

Peek inside my brain, Netflix.

Peek inside my brain, Netflix.

Whenever there’s something technologically fascinating that is beyond my ken, I just imagine that tiny elves are responsible somehow. Sitting in a workshop, tinkering and tweaking away, perhaps whilst humming a merry tune. But alas, Netflix recommends new things for you to watch not due to the machinations of adorable wee workshop sprites, but through the science of algorithms. Go figure.

So why does Netflix think you’ll be into streaming Cerebral Dark Romantic Comedies With A Quirky Twist And Maybe Some Cute Animals? How do the algorithm elves work their magic?

Wired.com spoke with two of Netflix’s recommendations gurus — Carlos Gomez-Uribe, VP of product innovation and personalization algorithms and Xavier Amatriain, engineering director — who are part of a larger team of 800 engineers working away behind the scenes at Netflix. Which kinda means there are elves at work, only they’re people, and less likely to be wearing pointy-toed boots and darling caps.

For example, if you were to watch the 1960s Star Trek, Netflix will buzz and whirr and spit out a recommendation of the original Mission: Impossible series. Why’s that, asked Wired.com?

“By looking at the metadata, you can find all kinds of similarities between shows,” explains Gomez-Uribe in the extensive Q&A. “Were they created at roughly the same time? Do they tend to get the same ratings? You can also look at user behavior—browsing, playing, searching.”

Similarities can differ depending on the director, for example. Some movies all by one person might be similar just because he or she was the director, like Pedro Almodovar, but others like Steven Spielberg might have a wider variety of movies that wouldn’t necessarily prompt a recommendation.

As for the elves at work deciding what makes a movie or TV show “dark” or “cerebral” or whathave you, Amatriain explains that there are over 40 people going through content by hand and tagging them. Often these people are freelance TV and film buffs, or people who have worked in the industry for years. They know what they’re talking about, in other words.

And also? While you’re watching, you’re being watched. Otherwise, Netflix wouldn’t know what to tell you after you finished up Ladybugs starring Jonthan Brandis (RIP, too soon, too soon).

“We know what you played, searched for, or rated, as well as the time, date, and device. We even track user interactions such as browsing or scrolling behavior,” says Amatriain. “All that data is fed into several algorithms, each optimized for a different purpose. In a broad sense, most of our algorithms are based on the assumption that similar viewing patterns represent similar user tastes. We can use the behavior of similar users to infer your preferences.”

Eventually, Netflix might even start utilizing the time of day you’re watching things to recommend content. So it’ll have one suggestion if you’re bingeing on 1980s romantic comedies at 8 p.m. on a Friday night versus staying up until 4 a.m. devouring Doctor Who on a week night.

For more on how the elves and the algorithms work together, take a gander at the full Q&A in the source link below.

The Science Behind the Netflix Algorithms That Decide What You’ll Watch Next [Wired.com]