Ok, Ok, i forgot about this kind of channel and yes they do collect subscribers, otoh who remembers or actively searches for the episode in which he squeezed toy $xyz and who could have guessed that it broke? And yes, i get these thrown into my suggestions and watch them now and then. He is good in regard to attract so many people and editing his work in a way that it stays entertaining. Surely an intelligent person, avoiding the obstacles. Doing 670 to 10k€ a month with it.
I would guess they completely live off the viral effect, first impression and their reach following that. Yet their video length is completely dominated by the attention span for short excitements, not as for educational content. That has consequences for the value of an ad there vs. here and it is questonable if quantity vs. quality is accounted for at youtube. Back when all was text, more content meant more search hits. Marketing companies are not only selling data and time or area, at its core it is also attention (and money).
I think that channels like yours are just a different type, i see this in youtube machining videos of educational content as well, which practically have a story in a 30 minute video. Viewers are more recurring there, generating more time watched. What they are doing different is occasional collaboration, contests between each other or meetups. It might be harder, as good EE is kind of a niche already, contrary to numbers of units sold that require an understanding of EE to work.
Any group of people has a fluctuation to it, it's normal as highly visible, active members will change their priorities now and then, not even related to the content but their perception of it relative to their life. For a full time single presenter that means viewers can sooner or later end his sentences. Thats nothing bad, it's just 1000+ videos and that might feel boring to some. Imho its got more to do with the watch habits.
As long as it's newby friendly, new people will come and stay for some time as well.
Consider the video topics would specialize extremely in one direction only (without getting viral), that would surely decimate subscriber growth and the mentioned decline part kills the rest.
To compensate a bit maybe a school class like approach (without calling it so) makes sense so there is some learning curve for viewers to get, but still enables new viewers to join? It can be just like it is but with more follow ups or more in depth. Schools do solve this problem for quite a while now, but with a differrent approach and often without any excitement. :-)