YouTube’s Algorithm

youtube algorithm

YouTube is one of the largest platforms to stream and upload content in video form. It has over 1 billion hours of viewership, and over 500 hours of video uploaded to it per day. It does make quite a bit of sense to handle filtering all that data and information to the most reliable data center of all time that is not human. Algorithms, the term is very overused nowadays to describe many things that are not necessarily the original meaning of the word. But for the sake of simplicity, YouTube’s Algorithm will be defined as this system that stores, analyses, and controls data. As an example, if a machine is given a selection of photos of fruits, and then a collection of pictures that are not fruits. If the same device is provided a photo, afterward, it will be able to determine whether it is a photo of a fruit or not. That is the central concept of YouTube’s Algorithm in a nutshell.

1. YouTube’s algorithm failing and repercussions

The majority of gripes that the audience on the platform are having with the system that the algorithm follows is the “unfair” treatment of content creators. What happened was a minor hindrance for some but others were devastated, from people losing years of carefully accumulated date overnight because of accidental channel erasing, to the existence of corporate entities that take the revenue from smaller channels by abusing YouTube’s own fair use and copyright law, making the platform virtually unusable for aspiring content creators. YouTube gave the algorithm the ability to filter the search results, advertisement placement in videos, and of course, the type of adverts to be put according to what it thought to be the person’s liking. And that created a problem in the heart of the system itself, where it was reported by credible sources that the algorithm was making decisions based on false data that it had gathered before. For instance, the algorithm noticed that some, and not all, of the videos speaking about LGBTQ topics used mature language. So the machine acted as it was programmed and made a correlation between LGBTQ subjects, tags, videos, and channels, which it then deemed not advertiser-friendly. And there are quite a few more examples like this one, but the outcome of such error was more significant than anyone could have anticipated. The headline that followed after the news broke out were very much attacking the company, calling them homophobes and unfit for the responsibility that is running such an institution. YouTube went from the most trusted site on the net to share and upload videos to an advertising nightmare. In the state that the site is in today, Google has improved its advertising techniques and upgraded the algorithm so it can figure out better the videos that are advertiser-friendly from the ones that are not. Because YouTube designed the algorithm to keep the viewers in the site browsing in what the internet calls “the rabbit hole of YouTube,” where they go from one video to another thanks to the algorithm recommending the right video. This plan backfired massively on the platform, making it one of YouTube’s most significant failures that they had to contain. On one hand, YouTube cannot assign humans to handle that much data. On the other hand, the only way for an algorithm with the size and speed that YouTube’s Algorithm filters information is if it took more and more time to learn the patterns that make it accurate enough for public use. And there’s a reason that YouTube did not give up on the algorithm, it is because it is already integrated too deep into the system. Everything from the recommendations lists to the trending tab, to the search results and even more, is controlled by a set of algorithmic decisions taken by a machine-learning artificial intelligence that did not quite get enough time to learn. Partly, because it is incredibly hard to tell when it is time to launch an AI. But mostly, due to some of the problems that the advertisers on YouTube had with the videos they were advertising, which lead to financial issues within YouTube.

Related: the new age of connectivity

2. Possible solutions

2.1 Programmers around the planet share an extensive and intricate debate between them on how the collective force of multiple algorithms may solve the problem. The essential idea is simple but challenging to put in action for a company like YouTube. The company would have to spend more resources on developing other algorithms that are specifically designed and created for the purpose of aiding in the making of YouTube’s algorithmic decisions. This would help those main algorithms to reduce the time for them to learn all of the data needed before it can be implemented in the field. But again, it would cost a fortune to proceed with such a project, and there’s the fact that, since this incident occurred, content creators are not really fond of the idea that algorithms are controlling their livings.

2.2 The content creators, however, had other solutions for the issue. Some abandoned the platform altogether in favor of other companies like twitch and mixer. Others opened PayPal and Patreon accounts to seek help from their respective viewers, without having to sacrifice their creative freedom to satisfy the algorithm. And while there was absolutely a group of them who took advantage of the situation and abuse the system to grow their channel and following on social media, most of the community on YouTube are very confused by what the algorithm wants to recommend and advertise.
YouTube’s algorithm is far from a perfect solution to the problem of the problem that YouTube had. But in the digital world that we live in today, algorithms may be the best solution in the long term. Where they can learn Exabyte amount of data in a very short period of time and can be programmed correctly and objectively to use that data for maximum convenience to the user. The only proof of this statement’s validity is time.

Originally published on Live Positively.