Now Trending: What and why we keep watching on YouTube

On Valentine’s Day in 2005, Jawed Karim, Steve Chen, and Chad Hurley, three former employees of PayPal, registered and activated the domain www.YouTube.com.

The first ever YouTube video was published on April 23, 2005, and shows Karim in front of elephants at the San Diego Zoo. "Me at the zoo” now has over 1 million likes, 46,000 dislikes, and nearly 475,000 comments on the world’s second most visited website.

Today, the video-creating, discovering and sharing platform is now used by 73 percent of U.S. adults—split almost evenly between men and women—and 94 percent of young adults ages 18 through 24, according to Pew Research Center.

FortuneLords reported that at least 1.3 billion people use YouTube, over 30 million users visit the site and nearly 5 billion videos are watched per day, and at least 300 hours of video content are uploaded to the site every minute.

With so much content being viewed and uploaded at every minute, Google, which owns YouTube, has created a complex way of organizing and continuous playback on the site: the recommendation system.

Paul Covington, Jay Adams, and Emre Sargin, computer engineers at Google, described the network's recommendation system as an “ever-growing corpus” serving its massive user base, by helping viewers discover personalized content.

It basically works as an algorithm of neural networks that retrieve a user’s YouTube activity history (views, likes, search queries and comments) as input and retrieves hundreds of video recommendations, called “candidates.”

The filtering algorithm then ranks among those candidate selections by assigning a numerical score to each video, according to the user’s history and interests. The highest scoring videos, which are meant to be personalized and engaging to the viewer, are presented as recommendations.

Recommendations, which appear on the home page, bottom and side-bars (depending on the device) of every video, show up as related channels and topics, from electronic dance music to puppies. They also appear as suggested videos and auto-play, maintaining that viewers always have a fresh selection of related content, from the trendiest music videos to “mukbang” and makeup challenges.

For YouTubers with active channels, figuring out the complex algorithm—and getting the desired views and likes on their content—can be a challenge.

Steve Lewis, 30, is a church missionary and vlogger from Oakland, CA, whose channel expresses his Catholic faith as a minister to youth and college students.

“The algorithm rewards you for frequent posting, longer videos, and for acting like a company or a TV channel, a brand. So someone who can post a lot, answer comments really quickly and turn it into a community--those people make money,” Lewis, who has been vlogging since 2014, said. “As a viewer, you used to be able to go down a rabbit hole so you would hop onto a new video, hop onto a new vlogger. Nowadays, YouTube will make recommends for you--it doesn’t want you to just click the most, but spend the most time. So now you’ll see videos you’ve both seen and haven’t seen before.”

Lewis says that in order for his channel, with just under 1,300 subscribers, to get attention, he needs to use a catchy, clickbait title and a good thumbnail. The “tags” and keywords section is also useful for engagement, if done right. His most liked video has 987 views.

One of Steve The Missionary’s most liked videos, which has 898 views as of April 2018.

“A couple of days after you post, the trick is to add your video to one of your playlists, so it re-enters your video into the algorithm as if you posted again,” said Lewis. “What you’re trying to do is make your video connected to other videos...then suddenly, one or two watched videos turns into seven. It becomes a huge time-suck.”

With suggested videos on auto-play, the recommendation algorithm, and streaming services like YouTube Red providing exclusive content to those who pay for it, YouTube wants its 1.3 billion users to stay engaged and always watching.