Have you ever snapped out of a YouTube coma to realize you just lost four hours of your life clicking through hundreds of video links?
You just got ‘Tubed.
The company precisely led you down their mental bunny trail by placing just the right videos in your line of sight, luring you to click just once more.
And maybe one more. (Oh OK, one more can’t hurt.)
So just how do they do it? Are they mind readers?
Well not unlike a magician’s skill set, it all really comes down to slight of hand. In this case, it’s a matter of those hands writing millions of lines of code.
By now most Internet users may have heard that mathematical algorithms are used by big companies like Google, Facebook and others to predict what people want to see.
“There’s a lot of code, scattered all over our code base … I don’t think of it as one thing, I think of it as lots of little pieces that are trying to accomplish the overall goal of connecting you with what you’re looking for or what we think you might want to watch,” Cristos Goodrow, YouTube spokesman, said in an interview with Computerphile.
You start the process with the first click and then the algorithm begins. Users continue to submit information through their online interactions — the choices you make via your clicks — and YouTube’s code determines what you may logically want to see next.
This analytical and predictive relationship is especially important with respect to the way advertisers lean on big Internet companies to select ideal targets for their merchandise.
Essentially, they pay for access to your eyeballs. And this is how YouTube serves up your attention span to the highest bidder.
“It resembles much more counting, than any dark sophisticated thing … I think we’re quite open about what we are trying to achieve,” Goodrow said.
They’re so good at it, YouTube even won an Emmy for its personalized video recommendations.
We can certainly take a little ownership here; you were a willing participate who sat there for four hours while clicking on the perfectly populated videos catered to your taste. But it’s easier to walk away when you aren’t interested in what you see — it’s much more difficult when the company knows how to entice your next thoughts.
“I think when people refer to the algorithm they mean all the code we’ve written to do those things … to decide which search results we’re going to show you and in what order, or which recommendations we’re going to choose for you,” Goodrow said.
[sharequote align=”left”]”It resembles much more counting, than any dark sophisticated thing … I think we’re quite open about what we are trying to achieve,” Goodrow said.[/sharequote]
Does this mean there’s a secret code hidden in some mysterious computer file or case that allows YouTube to predict what each person wants to see?
But they do keep the formula fresh, fluid and rather secret, because they don’t want users to manipulate results for the masses.
“We don’t want people to try and game what we are doing. We’re using certain kinds of information to help people find the videos they are looking for, (but) that depends on that being, that information coming from sincere viewers who are actually looking for things,” said Goodrow.
He continued, “Now if instead it came from a robot that wasn’t really looking for anything — it was just trying to make sure a particular video showed up at the top of some search results because it had a tremendous number of views — then we can’t use that (analytical tool) anymore.”
YouTube calls those tools “signals.”
“We have to try and protect these signals as we call them so that we don’t lose them and then we’re left with actually very little to go on … I don’t want to lock us into a particular way of doing things,” Goodrow said.
So why not just use a simple formula that ranks a video higher based on more clicks, for example?
“Five years ago ranking the search results by the view count may have been a great thing to do and that was because at that time view count was one of the best signals that you could find for what are the quality of the video, but since they we’ve found other things that are even better.”
Goodrow used a recent plane crash as an example. If the company didn’t change up their algorithm, when someone searched for “plane crash” on YouTube, they’d get the oldest, most-clicked file. Instead, the company presumes you may mean a recent crash, and their algorithm will reflect that.
In addition, Goodrow explains how the length of a video viewing session plays into their algorithm as well. He said they are always working on at least 10 different updates to the way their code will read, understand and predict videos for you.
It’s doubtful anyone outside of the YouTube engineering team will ever know exactly how they pick and choose videos for your viewing pleasure day by day, but now we know at least a few more things that aren’t factored in as much.
So the next time you find yourself staring at that next clip thinking “just one more,” remember, it will still be there tomorrow. And YouTube will probably find a way to put it in front of your eyeballs again.
The YouTube executive shared these thoughts with Computerphile in a video interview; you can watch the entire clip here:
Follow Elizabeth Kreft (@elizabethakreft) on Twitter.