What I consider:
-
Is an offending post a property of the instance or the user? It may reflect the instance if it’s written or reblogged by a staff member or part of a large pattern.
-
If it’s a property of the user but reporting doesn’t work, do I have the spoons to moderate their instance on a user-by-case basis? I usually do this only if there are existing connections or if the instance is run by and for marginalized groups.
-
If I see slur usage: is there plausible deniability for the slurs to be used in a reclaimed context, or for the users in question to not be native speakers? Glossing over this once earned a subsection in my blocklists’ “Mistakes Made” section.
-
If there’s harassment, I might want to check in with the victim to see if it’s okay to save a receipt featuring them and if I should make a callout.
-
Do staff members run other instances? I could treat all their instances collectively.
Minor factors that do not directly inform the decision to defederate, but do inform the decision to look deeper before dismissing the instance:
- Is the instance staff close to sufficiently bad actors? If so, the staff members almost certainly post similar content. It’s worth the extra time to find the receipts.
- If the instance is brand-new, I might re-visit it in a week or two to see if it cleaned up.
Steps I perform:
- Save receipts with multiple archiving services. Sometimes one service doesn’t render an archive properly, so it’s good to have alternatives.
- Describe reasons with those links, and make it detailed enough for readers to understand the reason without visiting any links. Archived links are for proof more than understanding.
I don’t consider whether the domain is still alive, assuming I have receipts saved from when it was online. Dead instances do sometimes come back, and blocking dead instances is harmless.