Your "censorship" framing is just a speculative assumption. There is no proof that censorship is going on here.
> What hypothetical "pure" intentions would you suggest are possibilities?
There are billions of comments on YouTube and anyone who actually works in software understands that software very often does things that the programmers didn't intend, and I'm not even talking about ML which is literally statistically applied logic, i.e. the wrong thing happening is inevitably assured and assumed from the start.
This whole argument is moot anyway, there's no commercial incentive to hire a legion of moderators because for all that extra money spent they'd simply be accused of bias even if the moderation was actually fair, a statistic of which there is literally no trusted party to arbiter.
That's hiding information that's normally available due to the nature of the content. That's censorship.
Censorship is bad, even if done for the right reasons.
That's the nature of censorship.
But especially so when the "alpha testing" is specifically done on videos of the new White House administration.
Unless you're questioning whether removing the downvotes qualifies as censorship at all. Which I would find a little weird.
What hypothetical "pure" intentions would you suggest are possibilities?