We had a random post in an anarchist community on our Polish speaking instance. Some 45 English speaking accounts came out of nowhere to downvote it, with a single one engaging in discussion. None of them were ever active on the instance, nor particularly in this community. Seems they just followed every crosspost.
Mods could not really do anything about it, so the accounts were banned from the entire instance by admins, as this was considered hostile behaviour against our community.
Which rises the question; should people be able to vote, end specially downvote, in communities they are not a part of? Maybe this could be at least a setting?
Another interesting concept that came from the discussion over that was “constructive downvote” - requirement of commenting why one downvotes a post.
I think this is missing the point.
The common thread I see in discussions like these (should X be able to vote, should downvotes exist, etc) is an attempt to separate “useful” votes from noise. None of the suggestions do that, since users will just find a way to get their desired effect another way.
I propose an alternative: a web of trust. Basically, votes from people you trust have higher weight than votes from those you don’t. If someone you trust trusts someone else, their votes would also be partially trusted. That way you only need to really trust a handful of people (pick them wisely) to get a good experience and can ignore the rest of the noise. Treating all votes the same is the problem, so we need some mechanism to weed out the noise.