Hacker News Reader

Comment by Kye | Hacker News Reader

They'll probably go away once IA's Mastodon mods wake up. I only see good comments from my instance, but it's been around long enough to have encountered all those commenters and/or their instances to know to block them.

View on HN

As someone who has been using Mastodon for over a year, this is the first time I’ve seen comments like these. And I am sure that it gets worse. All the instances we federate with must either have stellar moderation or not federate with these instances.

In fact from my server (and not this link) I don’t see these comments at all.

You can pick instances you federate with, but why do you need to defederate with instances that federate with other instances you dislike?

If I want to follow Madonna, and Madonna follows Trump, but the admin of my instance doesn't like Trump's instance, I cannot follow Madonna any more? Or does it only affect pulling all the content and I can interact normally even if the instances don't federate?

I’m not an admin so I’m not sure. My admin has de-federated with nicecrew.digital as it appears to be full of those kind of comments.

If Madonna is on the same instance as Trump that would make it so when your admin de-federated with it you would lose access to Madonna’s posts.

In my experience it looks like my instance really is only de-federated from garbage like the commenters from nicecrew.

I’m not de-federated from Internet Archive’s instance. I see the majority of the comments when I look up this post from my Mastodon account.

I’m surprised mods are even needed for some of it, a star hash on the n word requiring a double click atLeast would have been a little less aggressive. I’m sure beyond that some basic llm bot could run and flag the rest and humans could just review challenged ones but it seems standard to flag some basics words to demotivate some of the crowd.

There's probably a lively discussion somewhere on Mastodon's issue tracker on this exact subject. It's not as simple as it seems, though. How do you ship a word filter that catches it on a generic instance, but doesn't interfere with an academic discussion of reclaimed slurs?

It's actually a prime example of where Mastodon shines and why shared blocklists have been so controversial, and why any attempt to make a common word filter would fall into chaos. There's no central authority dictating what is and isn't appropriate. I can talk about queer stuff and unabashedly gay shit on tech.lgbt without some Mastercard brainwormed tool telling me what to do. And that's great.

I’d imagine someone not signed in would load up the safe mode generic list from js and the federated server preferences and if you want to reveal a word you can as a anonymous reader or set a cookie with preference sort of like turning off safe mode on duck go without having to sign in. Of course signing in would have your long term preferences. This allows racist to still send people to links with comments revealed I believe based on server or link url param … as aggressive as it can be it’s not really nuking comments and a freedom of speech issue

Didn't say anything about freedom of speech. And again: I'm not the one to talk to. I don't have any strong feelings on the topic, but if you do, you should take it somewhere that people who can do something about it will see.

I tried to find an existing discussion to help get you started, but couldn't. You can start one here: https://github.com/mastodon/mastodon/issues

It's easy to sit here on Hacker News and say "they should just..." Well, there's they. Go suggest it.

Coming up with a standard filter list for an international project will be a long, noisy discussion. You'll tread on internecine conflicts you had no idea about. Old wounds from past related discussions will come out. People will soapbox. The topic of the mere inclusion of the text of the slurs in the filter list will be 500 pages of yelling.

This is why I have no interest in discussing it. It probably won't go anywhere in a place where it actually could. It definitely won't here.

> How do you ship a word filter that catches it on a generic instance, but doesn't interfere with an academic discussion of reclaimed slurs?

Right. Not to mention the Scunthorpe Problem.

https://en.wikipedia.org/wiki/Scunthorpe_problem

Reply on news.ycombinator.com