YouTube-The most exciting tech

Aditya Aggarwal
2 min readOct 7, 2020


While writing this blog on the tech product that I am most excited about, I was thinking about numerous things from mouse to self-driving cars. But suddenly, I got a notification of a certain video posted on YouTube, and, after an hour, I still hadn’t finalized my idea. At that instant, I realized how YouTube is so captivating and influences our daily life.

How YouTube changed the world?

Back in 2005, the popular way of consuming videos was limited to cable television, which provided local content and major world updates. But with the advent of YouTube, watching and sharing videos became very convenient. The platform allowed anyone to post anything they wanted at any time. With very little attention on production-grade cameras and more focus on the authenticity of the content, the platform empowered new forms of communication to the masses. Over the last 15 years, it has dictated internet trends (Harlem Shake, Ice Bucket Challenge), discovered talents and spawned careers, started social campaigns, bridged cultural differences, and many more.

How does this all work?

It is fascinating to think about how more than a billion people can find what they want to watch in an uncountable list of videos. Behind this lies a marvelous tech product — the recommendation engine of YouTube. In the past, the recommendations were based on primitive non-personalized features like view count (a.k.a. the number of clicks) and watch time of the video. But in 2020, with the machine learning wave, these recommender systems use real-time user feedback and deep neural networks for ranking the videos.

The idea here is not to generate the best results based on relevance but to maximize viewers watch time. It consists of a candidate generation network that takes a corpus of millions of videos and generates hundreds of candidates based on query parameters and video’s metadata. Next, it uses a ranking network to score each video based on user engagement and degree of satisfaction based on likes, dislikes, and comments.

Wrapping up

YouTube recommendation engine is an impeccable example of modern computing and redefines content recommendation based on personalization. Similar ideas are explored in numerous other products as well including Facebook, Spotify, Amazon. But the user experience paired with the scale at what the Youtube algorithm works really fascinates me.