Over the years, I have published over 200 videos on YouTube [1]: mostly programming tutorials as well as programming-related entertaining and motivational content. Among these videos, I have six courses (roughly 10 videos each) on various topics such as location-aware application development, visual web development, and algorithms. I now concluded a new course on Coding a self-driving car in JavaScript [2]. In this article I want to share my most recent finding related to publishing courses to YouTube.
Coding a self-driving car in JavaScript is a course that teaches how to make a self-driving car simulation by implementing every component ourselves, without libraries (other people’s code). We learn to implement the driving mechanics, to define the environment, to simulate proximity sensors, to detect collisions and to make the car move by itself using a neural network [3]. We learn how neural networks work by comparing them to the biological neural networks in our brains, then implement something similar on a computer. We optimize the network using basic evolutionary methods. The course consists of 10 videos (see Table 1).
Table 1. Individual video statistics.
| Video # | Topic | Views | Likes vs. Dislikes | Comments |
| 1 | Intro + Car driving mechanics | 6,166 | 99.2 % | 94 |
| 2 | Defining the road | 1,313 | 100 % | 34 |
| 3 | Artificial sensors | 1,115 | 99.8 % | 29 |
| 4 | Segment intersection | 4,430 | 99.3 % | 43 |
| 5 | Collision detection | 1,412 | 97.6 % | 17 |
| 6 | Traffic simulation | 775 | 98.4 % | 23 |
| 7 | Neural networks | 8,009 | 98.2 % | 51 |
| 8 | Visualizing neural networks | 1,298 | 100 % | 28 |
| 9 | Optimizing neural networks | 2,163 | 99.1 % | 29 |
| 10 | Fine-tuning | 1,615 | 97.9 % | 42 |
| TOTAL | 28,296 | 98.95 % | 390 |
At the date of writing this (16.5.2022), the course has already collected more views than any of my other courses, despite them being on the platform for much longer (one year, at least). This is mostly due to the channel growing over the past year, and YouTube forming a better picture of the target audience for my content, thus sharing the video to the right people. The likes/dislikes ratio is high, but similar to my other content, the number of comments is significantly higher. This increased engagement shows in the subscriber count as well, which more than doubled (from 2,407 to 5,425) as the course was being published (over a two-and-a-half-month period – one video per week). I consider these statistics remarkable, and that is why I have written the current article: to share what I did differently when designing this course.
How the YouTube algorithm works
Before I discuss my techniques, it is important to note that the success of a YouTube video depends largely on the algorithm [4]. Even if you have the best video in the world, if YouTube doesn’t decide to share it, it will remain unnoticed. The algorithm works in mysterious ways: it changes constantly as users interact with the platform. If YouTube decides to share a video today, it may not decide to do it tomorrow and so on. However, some patterns are clear. For example, when a new video is uploaded, the first minutes are important. If people engage with it in a positive way (like, comment, subscribe), YouTube will show the video to more people and so on.
Tailoring courses for YouTube – overcoming the algorithm obstacle
The fickle algorithm poses a problem when publishing a course on YouTube. As can be seen in Table 1, the first video performed well in terms of views and especially engagement (people were excited for the rest of the course). However, the second video is significantly less popular because one needs to go through video 1 for video 2 to make any sense. YouTube doesn’t know this when recommending video 2 to users, so people who haven’t started the course will immediately close video 2 upon opening it. This gives YouTube the wrong signal. Even if some of these people go to video 1 to start the course, the signal that video 2 is not interesting has already been sent.
This observation is important. It means that YouTube is not the best platform for publishing courses – video lectures that are related to each other, that is. I counteracted this by making videos 4 and 7 independent from the rest of the course. They teach two fundamental techniques: how to find the intersection of two segments and how neural networks work, respectively. These techniques are general and have countless uses outside self-driving, thus fitting as standalone videos. I begin these videos by mentioning the self-driving car project and that one doesn’t need to have started the course to continue watching. This achieves two things:
- people won’t click away and
- people learn about the existence of the self-driving car course (a kind of self-promotion).
Because neural networks are such a hot topic nowadays, video 7 is now the most viewed in the series (see Table 1). This outcome was expected when I planned the course and is the reason why half the effort was put into designing the neural networks lecture.
Alternative solutions to creating video courses on YouTube
This was not the only way to handle this problem. Alternatively, the entire course could have been presented as a single, long video. I disagree with the popular opinion that short videos are better [5]. I think the reason they perform better is because they are usually better quality: it is much easier to make a high-quality short video than a high-quality long video.
On YouTube, long videos can be split into chapters, so they feel less overwhelming, and there are ways to skip to a given chapter if wanted. Moreover, some viewers complained when I published the first video, saying they won’t follow along because they need to wait between the lectures. This lost me potential viewers. However, with the long video strategy, videos on the channel will appear less frequently (every few months instead of every week). This impedes channel growth especially for small channels like my own.
In conclusion, YouTube is not ideal for publishing courses, because it promotes each video independently, and video lectures usually relate to each other. However, by organizing the course so that the independent videos are evenly spaced between the lectures, they will likely be shared more often, and the course can become an overall success.
Author
Radu Mariescu-Istodor, lecturer, Karelia UAS
References
[1] https://www.youtube.com/RaduMariescuIstodor
[2] https://www.youtube.com/playlist?list=PLB0Tybl0UNfYoJE7ZwsBQoDIG4YN9ptyY
[3] Aggarwal, Charu C. 2018. ”Neural networks and deep learning.” Springer 10 (2018): 978-3.
[4] Arthurs, Jane, Sophia Drakopoulou, Alessandro Gandini. 2018. ”Researching YouTube” Convergence 24, no. 1 (2018): 3-15.
[5] https://greenbuzzagency.com/short-video-vs-long-video-optimizing-video-length
Header image: NordWood Themes on Unsplash