I suppose if you have an instance wide policy against AI moderation, then any mod using AI for moderation is going against the rules. But what anyone “should” do on their own instance is really up to them.

They don’t need to federate even, they’re probably already scraping everything.

Why is that one about the WarEra user green?

Is this some american thing? I’ve never heard of anyone having a different name for a language class, that just seems odd to me. Your name is your name, doesn’t matter what language you’re speaking.

It’s based on lemmy-ui which is AGPL-licensed, so I’d think it would have to be released somewhere?

“Abuse” was maybe the wrong word, but my point is just that maybe liking and disliking shouldn’t be the same as voting up and down.

There are some platforms that offer emoji reactions. That at least gives a bit more nuanced of a reaction than just “up” or “down” - but it is not clear how you would use emoji reactions to sort posts. The good thing about upvotes and downvotes is that it’s a very easy signal for which posts should be at the top, and which should be sorted below.

That’s fine but does that action always correspond to wanting to upvote or downvote the post? I guess maybe? Lemmy at least assumes so.

When are you meant to use them then?

Boost (if its the same as on Mastodon) is indeed entirely different. It’s an “Announce” activity, which is akin to sharing or “reblogging” the post. This is not treated as an upvote in lemmy as far as I know.

What is a KVM? I’m not familiar with the term. I just meant that you aren’t able to SSH into the machine or something? But I guess if it’s turned off and physically unavailable, that’s hard.

Do you not have a way of remotely bringing the server back up?

Is there an issue about this in the GitHub tracker? It could be an unintentional consequence.

Lemmy, which is pretty known for its opinionated devs

The opinions of the devs have so far (at least as far as I know) not spread into the actual code though.

I feel like you’re moving from moderation to sort of oppressive or authoritarian territory once you’re literally building a social credit system into your software. If you want that, sure use PieFed. I don’t want that, so I won’t.

No just the whole thing. It seems sort of extreme to do all this stuff in the code. This is not something the software should have inbuilt if you ask me.

I mean okay sure, you can disable these things. But the fact that they are enabled out of the box in the software as written is a huge red flag.

… Wow. I mean I already knew there was some questionable stuff with PieFed but this is honestly next level.

Downvote and move on, if the name becomes familiar block them.

Sorry, but you forgot a step:

Downvote and move on, if the name becomes familiar report them and then block them.

Moderators should help so not all users will need to block these bad actors.

Perhaps that was exactly the intent, to ensure that only actual humans read the content?

If that’s the case, it has the opposite effect. Humans generally benefit from being able to select text and not just read from an image. It’s of course especially important for blind people.

Meanwhile computers can easily read images via OCR algorithms.