I suppose that could be considered a very mild form of trolling.
I guess the theory is that Trump's followers are force-joined into subscribing to the Biden White House's videos. So then they react by downvoting every unsolicited video that pops into their streams. And the response is to delete their downvotes? So I guess the message to them is, "You really must look at what the Biden White House has to say. Oh, but you're not allowed to express your opinion about it."
Actually, you are allowed to click "dislike" and to feel like you expressed your opinion about each video, but later we will delete your opinion when you look away.
It is NOT trolling. I made the site. You can easily verify my data.
- The hard way: Use YouTube's API to track the stats on one of the videos. Compare to my data, and the two will line up.
- The medium-hard way: Over the course of a day or so, simply compare the official likes/dislikes that you see on youtube.com to my stats at 81m.org. You will see that my stats line up with the official numbers.
It might be hard to believe at first, but I guarantee you that if you look into it, you will see that YouTube is deleting dislikes. There are tons of random threads about this on Twitter that are totally independent of my site 81m.org.
You can see many examples. The real approval ratio, which I calculate by periodically checking the YouTube API, is typically around 2% to 5% on all Biden White House videos. However, the official YouTube approval rate, the one YouTube wants you to believe, is usually around 20% to 30%, which they achieve by deleting dislikes periodically. (And I think the latter approval rate is going up over time, though I haven't done a rigorous analysis yet.)
Why is YouTube/Google manipulating the like/dislike ratio in favor of the Biden White House?
[edit: Disclaimer: I made this site. I don't profit from it in any way. I just want people to be able to examine the data and come to their own conclusions. The code is not open source (yet), but the data are available for download, and the data are also trivial to spot-check and verify.]
Disclaimer: I wrote the linked article.