Moderation notifications to PDS #3190
benvanstaveren
started this conversation in
Protocol (atproto)
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi folks, @blockstackers.net here - after a discussion with @bnewbold and @dholms I got pointed over here. This topic does probably need a way better title but this is what I had so... change at will!
I'll throw out a little backstory, and then the points (or some of them) I had in my questions to both folks tagged above; I've been thinking about doing PDS hosting, in 2 flavors. The first flavor is where people create accounts on a PDS I run, and either migrate (via a tool I also want to write, still), or start fresh. A low-level entrance into being a bit more decentralized. I also wanted to offer one-click (or maybe two-click :P) hosting of PDS's to people where they can basically be in control of the whole thing top to bottom.
So. That lead to some general discussions about GDPR and how this applies (as I am an EU resident, and everything would be hosted in Europe), which then branched off into moderation questions and discussion. The most pertinent question ended up being (paraphrased):
If Bluesky (PBC) moderates content with a takedown (account or content removal), does this remove the account or content from a 3rd party PDS.
The answer to that seems to be (at the moment): no, it isn't. And there lies a few dragon-like problems:
My "solution" to the problem would be as follows (and I believe @dholms was thinking along the same lines): when a takedown moderation action is taken, whether this is on Bluesky, or another 3rd party appview, the PDS hosting the content or account needs to be able to be notified of this event.
Per @dholms this is possible for Bluesky via a websocket connection to the mod service, but this of course does not apply to any other appviews that may or may not pop up.
I was thinking about a configurable webhook, that can be exposed via the describeServer method; if no webhook is configured, use a "default" in the PDS itself that does nothing but take the notification data, and sends it to the PDS admin email, or perhaps if no webhook is configured (or the webhook isn't responding), the moderation service can use a contact email exposed via describeServer (it doesn't seem to expose the admin email at this point in time).
The webhook itself would be sent the relevant content details, such as did, cid, and perhaps a "reason" flag, and ideally a way to validate that, for instance, a notification of moderation came from Bluesky's moderation service (which personally I would trust 100%) so that the consuming end of the webhook can either just go ahead and blindly follow the takedown, or it can be queued for review - perhaps by way of signing/verification keys.
I believe this offers the largest amount of flexibility for everyone involved (since the PDS really doesn't have to do anything, or require more complicated configuration) and would at that point lower the bar for operators and providers.
Discuss! 😃
Beta Was this translation helpful? Give feedback.
All reactions