The subs going dark should have only been half of the protest. Users should have also stayed away from the site but I don't think that was really coordinated.
The number of new posts didn't drop much, the comments dropped a bit more but only by like 20%, which isn't a lot given the amount of subs that went dark. Reddit doesn't care about subs, they care about users and it seems engagement was still pretty high.
The next protest should be to all users to stop using the site. Drop the users and they'll start to listen.
I don't know a lot about Lemmy's implementation but a difficult thing to deal with is how do you "rank" a post? Like you have a small community of a few active people, but there's federation with a massive community with lots of users - which posts are "better"?
Worse still, there's an inherent lag/delay with the federated posts, a post that was very active in the last hour might have only been federated to the server in the last 5mins - so what do you do, do you bubble up all those posts or ignore it because there's more recent and relevant things?
The kicker is that these decision points aren't instant either, any system that's doing this kind of ranking will have an algorithm as you describe, but that algorithm will take time to process all the data, while the data is coming in batches as each server federates with each other. It's a difficult problem to solve.