No, this is not needed. Online video streaming nowadays uses a "playlist" of video chunks. These chunks are seperately encoded videos, so can easily be swapped out, or concatted to. This also means that an ad-blocker could ignore specific chunks if they know which ones to ignore.
No, if YouTube displays any UI such as a clickable link, that means it has to know how long the ad is. SponsorBlock could find this data as well. There is also the feature for clicking on a timestamp in a comment that would need to know the duration of the ad, so it should be findable somewhere, it just might be kind of hard.
In the short term, SponsorBlock will not work for people with this experiment.
Probably not. But it makes things harder. As always, uBlock Origin work best on Firefox-based browsers, especially now that we reach the end of manifest v2.
Maybe in the future, but at the moment it seems to only switch to this server-side ad injection via client-side flags.
The more pressing issue for these clients is the new sign in requirement to watch videos for some people.
The offsetting issue will affect DeArrow thumbnail submissions, but should be fixable in the same way as SponsorBlock.
In the worst case scenario, would it be possible to switch to a style of database based on hashes of "bad" chunks? If users have a client-side system that allows them to tag (if they wish to do so) intervals of a video they watched as "ads" (distinct from sponsors), each of those intervals might be expected to consist in bad chunks (the ad) surrounded by a few good chunks included by accident. But an ad is expected to recur in different videos and places, so the boundary good chunks will change. Once enough submissions share a common chunk sequence, all the common chunks can be deemed as the ad and their hashes persisted to be distributed as part of a blacklist database. This doesn't fully solve the problem of being able to watch the video continuously, but it does solve the problem of finding out what is part of the real video and how long it is; Sponsor entries can even be corrected after the fact to account for ad gaps identified later.