As others have noted, there are some problems with the methodology of how this list has been created.
First, the list of "almost every" Software Engineering conference around the World[1] used, is just 72 items long. This feels light, and some conferences I would expect to see are missing: Brighton Ruby is very popular in the UK, but possibly a bit niche, so fine; but, FOSDEM however, seems more of a glaring oversight, perhaps explained by the fact that most of those talks are published not through a single channel.
The author has a mechanism to add confs to this. We could try and help them with that.
Then, the list seems to have ordered talks published by each conference's YouTube channel, by total views.
When one of those channels posts a talk twice (for example, the Linux Foundation's keynote of Linus talking to Dirk Hohndel), you end up with split user counts which means its lower down the list than you might expect.
I also think this means we're now measuring "most watched" in a slightly weird way: is this now just a reflection of marketing and subscriber reach? GOTO has 1.04M subscribers, ReactConf just 28.5K - is it any surprise that the list has more talks with more views from one of those over the other? Who is to say which conf had the better or more interesting talks?
Other engagement metrics (thumbs up, percentage of subs, number of comments, how many inbound links, and so on), might provide a better number to proxy for "quality" or "best" than just views, although the author isn't making a claim for quality, just views, so perhaps I'm asking for something beyond what is reasonable or what was promised.
This is a good start of an idea though, it wouldn't take much to make it better, so best of luck to the author or those who want to steal the idea: I'd love to see a better list in the future, and perhaps over a longer period of time than one year, too.
[1] https://www.techtalksweekly.io/p/tech-conferences