It seems like this statement from YouTube[1] and this Github issue (referenced by granzymes[2]) have key information being missed by a lot of commenters.
From YouTube:
> Viewers Using Ad Blockers & Other Content Blocking Tools: Ad blockers and other extensions can impact the accuracy of reported view counts. Channels whose audiences include a higher proportion of users utilizing such tools may see more fluctuations in traffic related to updates to these tools.
Quoting granzymes:
> According to the GitHub issue, YouTube didn’t change anything. There are two endpoints that can be used to attribute a view. One is called multiple times throughout a video playback and has been in the easylist privacy filter for years. The other is called at the start of a playback, and was just added to the list (the timing lines up with the reports of view drops from tech YouTubers).
From YouTube:
> Viewers Using Ad Blockers & Other Content Blocking Tools: Ad blockers and other extensions can impact the accuracy of reported view counts. Channels whose audiences include a higher proportion of users utilizing such tools may see more fluctuations in traffic related to updates to these tools.
Quoting granzymes:
> According to the GitHub issue, YouTube didn’t change anything. There are two endpoints that can be used to attribute a view. One is called multiple times throughout a video playback and has been in the easylist privacy filter for years. The other is called at the start of a playback, and was just added to the list (the timing lines up with the reports of view drops from tech YouTubers).
Source from the GitHub issue for easylist: https://github.com/easylist/easylist/issues/22375#issuecomme...
[1]: https://support.google.com/youtube/thread/373195597
[2]: https://news.ycombinator.com/item?id=45277768