HOW THE YOUTUBE ALGORITHM WORKS IN 2022

Good morning dear readers of Tecnogalaxy, today we will talk about how the Youtube algorithm works, during this 2022.

You may not think of YouTube as an example of social media at first glance. It is much less likely that you will connect with friends, family, and even customers on LinkedIn, than social networks, such as Facebook or Instagram. But at the center, it’s a social network, where you share videos with other youtubers. And like all of these platforms, YouTube uses a complex algorithm to decide the location of videos in its recommendations and lists.

People watch what Youtube wants

It may surprise you to know that most YouTube views are made following the platform’s recommendations. Like Netflix, YouTube uses artificial intelligence to determine “best” videos for viewers (or at least for the person whose account is currently connected).

YouTube Chief Product Officer Neal Mohan admitted at CES last year that over 70% of the videos you watch on YouTube are due to suggestions provided by one of the service’s artificial intelligence-based recommendations.
This is particularly relevant for mobile viewers who devote more than 60 minutes to each viewing session. With such long viewing sessions, YouTube’s artificial intelligence has the opportunity to place many potential videos to watch in front of the average viewer.

The Youtube algorithm has changed over time

Prior to 2012, YouTube ranked its videos by the number of views: the more people watched a video, the more it would be presented to other viewers. The problem was that people learned to play this algorithm easily. All they had to do was give a video a clickbait title, encouraging people to open it and watch a small part of it. Of course, they would soon realize that the video did not do what it had promised, so they left it immediately and switched to the next video in the queue. In a short time people complained about the numerous clickbait videos.

YouTube changed its algorithm in 2012, this time privileging the duration, viewing time, and session time (total time spent on the platform). In a short time the “players” learned that the way to beat the system was to create long and prolonged videos, which met their basic premise, but so waffling and padded.

The problem with the need to make longer videos, while maintaining the overall time of the session, was that youtubers began to run out, with the demand for huge amounts of content. Creators still had to make as many videos as before, but with less time and resources for video.

It was during this era that some new video genres became popular: they had the potential to be long videos but were easy to make. Perhaps the most obvious of these were the gamers who started filming themselves playing. It is much easier for a player to shoot themselves in action during a game than to make a review of the autonomous game, which requires planning and a script, for example.

In 2016, however, YouTube added artificial intelligence and machine learning to its algorithm. This has radically changed the types of videos it has offered people. He was criticized for exposing conspiracy theories and false reports. The most recent (2019) changes to the algorithm are in response to some of the inappropriate materials that people upload to the platform. YouTube recently announced that it had modified the algorithm to ban “limit content”.

The purpose of the Youtube algorithm

YouTube engineers describe the algorithm as one of the “most sophisticated and large-scale industrial recommendation systems in existence”.

However, not all engineers are that polite about the algorithm. Guillaume Chaslot, a 36-year-old French computer programmer with a PhD. in artificial intelligence, he was one of the engineers who created the algorithm. He was disappointed during his three years of work on the project. He said, “YouTube is something that seems real, but it’s distorted to let you spend more time online. The recommendation algorithm is not optimizing for what is true, balanced or healthy for democracy”.

Chaslot added that the algorithm never remains the same. It is constantly evolving, placing different weights on different signals. Engineers were responsible for experimenting with algorithm changes that would increase advertising revenue and the time people spend watching videos. According to Chaslot, “Display time was the priority. Everything else was considered a distraction”.

Ultimately, however, the system has two objectives:

1. help viewers find the videos they want to watch;
2. maximise long-term audience engagement and satisfaction;

How the Youtube algorithm works

As with most artificial intelligence systems, YouTube’s artificial intelligence is sophisticated and YouTube has only released limited information about it. They published a white paper in 2016 ( Deep Neural Networks for YouTube Recommendations ) that clarified some details (and of course people got some guidance from former employees). However, most of the information about the algorithm is still very secret.

We know from the white paper that it uses artificial intelligence to track viewers’ perceived satisfaction to create a personalized and compelling advice stream, that is, it works to determine how satisfied/ happy a viewer is with each video that plays and then adapt future tips to try to increase this level of satisfaction.

There are actually two neural networks in use. The first filters the videos to decide what would be a good match for the viewer’s “Next Up” tips. The second neural network assigns each video a score based on a number of factors (not yet known to the public), but seems to include an allowance for the novelty of a video and the frequency of uploads on the channel that uploaded this video.

This is all about the Youtube algorithm, to a next article.

Read also:

• OPENAI HAS DEVELOPED WHISPER!
• APPLE MAY INTRODUCE THE USE OF THIRD PARTY APP STORES
• HERE IS THE LATEST VERSION OF STABLE DIFFUSION 2.1
• TOP 5 FEATURES WE MIGHT SEE ON IOS 17
• HERE’S WHAT 2022 WAS LIKE FOR THE WEALTHIEST TECH ENTREPRENEURS