Embed Notice

@DryBird227 @freakazoid that's actually not how it works at all. It's not an auto block system that detects offensive language used and blocks you for that. That wouldn't be possible anyway since it's an instance level block list so it would have to block an entire instance for one user saying something offensive even if that user was later moderated or banned from that instance.

According to its own website (https://tweaking.thebad.space/about) how it actually works is not detecting offensive language in people's posts and automatically blocking them whatsoever. Instead it's block list is a composite of nine "trusted authority" instances. What it does is use public apis to read the block lists of those nine trusted instances and if three or more of them block any given instance then that instant gets onto the composite block list on TBS. Then the TBS block list is designed to be downloaded as a CSV and imported into an existing instance or a new one that you're setting up.

I have several problems with this:

1. Centralization: we are relying on a core set of trusted authorities instead of communities collecting and sharing information about instances in a distributed and horizontal community-originated manner. I think this is a bad idea even in the abstract.

2. Wrong Authorities: at least two of the instances of the nine trusted authorities — so almost enough to get an instant blocked by themselves — are rage.love and mastodon.art, both of which are known for making poor moderation decisions, going on witch hunts that sometimes even they later have to retract, and one of the admins of mastodon.art is Welsh Pixie who is a well-known bad actor with a transmisogynistic bias. Also at least two and probably more of the instances listed have a private discord server which means that multiple of the supposedly independent votes have a back channel where they can pressure each other to vote in unison.

3. And of course, there are the criticisms about releasing an unfinished product where you imported a gigantic block list and by-default labeled everything in that block list as bigoted and a source of hate speech and poor moderation (instead of something nonspecific) without knowing whether that was true or whether they were part of the block list of the source instances for purely petty or temporary reasons.