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.
Mods could not really do anything about it, so the accounts were banned from the entire instance as this was considered hostile behaviour by 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 see two “deep” issues here.
One of them is that it’s damn hard to decide, in online communities, who should [not] be allowed to perform some action in a fair, transparent, and simple way. There’s always some way to circumvent it, and always someone who should perform it but gets locked out.
For example: what would prevent me from subscribing to a comm, downvoting everything there, and then unsubscribing from it? Or just subscribing to comms to vote-brigade them, while newbies legitimately interested on the comm are unable to vote in it?
The second one is that this sort of Reddit-like voting system doesn’t really work well. It’s at most bidimensional (score vs. controversy, or up vs. downvotes); and yet there are a thousand reasons why people vote, and a thousand pieces of info that they can retrieve (or falsely believe to retrieve) from them. And depending on those reasons, the vote might be completely fine or not.
There are also more practical concerns; I believe that @davel@lemmy.ml’s Hexbear example illustrates this well. If you anyhow hamper the ability to voice negative feedback through downvotes, people do it by noisier ways.