[almog]

This just goes to demonstrate what might be triggering these false positive auto-moderation, but it shouldn't justify it.

What I mean is that if the model they use for flagging would infer that legit spam (comments to a post by Justin Bieber AFAIK isn't related to the Israeli-Palestinian conflict) could mean that similar content when applied to relevant content is still spam (or marked for deletion for other reasons).

It's a bit like if google page rank algorithm identified a website being copied by multiple websites and punished the website that was the first to publish that content instead of other way around. If that was the case, it would have obviously been exploited to bury competitors pages.

[charlesju]

I think what probably happened is these people (real or bots) posted on every major influencer and liked each other's posts to get them to the top of the posts.

Which is very similar to what a spam bot ring would do.