Any feature relying on a willing cooperation of every node in the network is impossible to enforce.
Sure, it should be attempted, but don’t expect it to actually matter against actors that just don’t give a fuck.
Any feature relying on a willing cooperation of every node in the network is impossible to enforce.
Sure, it should be attempted, but don’t expect it to actually matter against actors that just don’t give a fuck.
As a general note, the design of Lemmy is around preserving information against censorship over having private conversations (like Matrix for example). You're probably aware that once data is delivered to other servers, the origin server can't guarantee deletions of data from them.
I'm not an admin, but I remember that Lemmy admins are able to purge content from their server, and send out a request for other servers to purge their copy of data which by default would be accepted if received from the canonical server. This was used to remove some harmful content that was posted on Lemmy a few times. This feature isn't user accessible, at least as of now.
Media uploaded to Lemmy by the user can be deleted.
Edits are propagated through federation, so if you are concerned, you can edit posts you want to delete before deleting them which is close enough.
Thanks for the detailed explanation — this helps clarify a lot. I understand that federation makes guaranteed deletion impossible, but knowing that admins can purge content and send out a federated delete request is exactly the kind of mechanism I was hoping existed.
From a user perspective, having some version of that available — even if it can’t guarantee 100% deletion everywhere — would still be meaningful. A user‑initiated delete request that other servers can respect would give people more control over their own content without undermining federation.
Editing before deleting is a good workaround, but it still feels like something that could eventually be built into the platform in a more direct way. I appreciate the insight; it’s good to know this isn’t a dead end technically.
Thanks for the explanation — that actually helps me understand the situation a lot better. I get that federation makes things more complicated, especially when content is already spread across multiple servers.
From a user perspective though, it still feels important to have a way to fully delete posts, or at least send a federated delete request that other instances can respect. Even if it can’t guarantee 100% removal everywhere, having a proper deletion mechanism would give users more control over their own content.
I’m glad to hear it’s not technically impossible. Hopefully it’s something that can be explored in the future, even if it takes coordination across the Fediverse.
I am not a programmer, but as I understand how it works, when you access content (user profile, post, etc.) that is not on your server, a request is sent and two things can happen. The request receives a response and the content is updated or it is not received and you see a cached version. I suppose a third option could be added, you get a response and it says that the content is actively deleted, causing the cache to be deleted. Maybe it took a while to clear everywhere, but I don't see why it wouldn't work.
That makes a lot of sense, and it’s exactly the kind of mechanism I was hoping existed or could be added. Even if deletion can’t be instant across every server, having a federated “this content is deleted” response would give users real control instead of just hiding posts locally.
A system like the one you describe — where a delete request clears caches on other servers — would solve most of the privacy and user‑control concerns people have. It wouldn’t need to be perfect to be meaningful.
I appreciate the explanation. It’s good to know that this isn’t impossible, just something that needs coordination and implementation across the Fediverse.
The solution: A three‑layer deletion model This is the only model that satisfies both Lemmy’s architecture and user expectations.
Layer 1 — Local hard deletion (guaranteed) When a user deletes a post/comment:
the content is wiped from their home server
the object can remain as a placeholder to preserve thread structure
media files are fully removed
This part is already possible.
Layer 2 — Federated delete signal (best‑effort) When deletion happens, the home server sends a message:
“This content is deleted — purge your copy.”
Servers that respect federation will:
delete their cached copy
update the thread
remove the content from search
Servers that don’t care will ignore it — but that’s already true today.
This is the missing piece Lemmy needs to implement.
Layer 3 — User‑initiated purge request (optional escalation) Admins already have a purge tool that:
deletes content locally
sends a federated purge request
is accepted by most servers
Expose this to users in a controlled way:
rate‑limited
confirmation required
optional admin approval
This gives users real deletion power without enabling abuse.
You say you understand the technical limitations, then immediately ask if there are technical limitations in the next paragraph? And then this comment is clearly written by an LLM. Maybe if you tried to use your brain you'd have better luck.
I’m trying to understand the current behavior as clearly as possible, which is why I asked for clarification. The replies from others in the thread have helped fill in the gaps about how deletion works across different servers.
My goal isn’t to argue about federation limits — it’s to understand whether the user‑side deletion experience can be improved, even if perfect deletion across all nodes isn’t possible. That’s the part I’m trying to explore here.
all 15 comments