There is, in my opinion, only one valid use case for this, that being the business use case. But for those cases i'd actually leave the ACL as is and instead allow message logging in the background. A business chat platform will probably want to keep message deletion for usablity reasons and have their compliance handled in the background. As for "granularizing ACLs on existing actions": I think this has the potential to do more harm than good, at least on regular instances. I think users should fundamentally have the right to view/edit/delete their messages for any reason at any time. Fluxer is still not completely where i want it to be for this (i would like a "inaccessible messages viewer" in the settings to allow viewing and deleting of messages that you can no longer access). If you want to preserve messages as evidence in a community you can use a bot to archive messages (allows deniability since bots can manipulate messages).
As for the implications on federation: Federated instances can act in malicious ways no matter the outcome of this feature request. That said i think instances blocking message deletion should be labeled malicious and defederated from the fluxer.app instance since it fundamentally opposes the "right to be forgotten" (even if only blocked for individual messages).
To address issues with my message logging idea: I am aware that this can be misused as well, however this is an unavoidable side effect of having a non encrypted chat platform. If i were a malicious instance i could enable some message logging right now, by simply having up-to-date backups of the database (this wouldn't be enough for the business use case, since there could be messages that are sent and deleted between backups).
I currently do not see how this would be useful in any way (with the exception of the business use case, which i addressed already).
RexSystem2 votesoriginally by @fluxerhost on GitHub OP
i'd actually leave the ACL as is and instead allow message logging in the background.
I agree with this, however, that's more work than expanding out the message action ACLs. If this is more preferred by the development team, however, sure - why not add that too. Expanded ACLs are still important for administration (both at the instance and community level), I think.
As for "granularizing ACLs on existing actions": I think this has the potential to do more harm than good, at least on regular instances.
I don't think so, and anything, any feature at all, can have "potential" harm, since "harm" is within the control of the harmful actor, not the developer, of a technology. (Food for thought: do we stop using computers due to their "potential" to do harm?)
I think users should fundamentally have the right to view/edit/delete their messages for any reason at any time.
Then an administrator who agrees with this can disable the feature on their instance, or if they enable it, a community administrator can set permissions how they feel is appropriate on their community, via the standard and accepted methods already in the Fluxer codebase: ACLs.
If you want to preserve messages as evidence in a community you can use a bot to archive messages (allows deniability since bots can manipulate messages).
As for the implications on federation: Federated instances can act in malicious ways no matter the outcome of this feature request.
Agreed.
That said i think instances blocking message deletion should be labeled malicious and defederated from the fluxer.app instance since it fundamentally opposes the "right to be forgotten" (even if only blocked for individual messages).
Disagreed. Expanded ACLs is not malicious. Expanded ACLs is a technical control. How people use it, even existing features, such as mentioned in the following comment, is up to them: https://github.com/orgs/fluxerapp/discussions/3200#discussioncomment-18753294
Furthermore, the "right to be forgotten" is already covered by two features on Fluxer: leaving a community and deleting all messages, and deleting an account and deleting all messages.
To stay focused on the business use case, though, there is no "right to be forgotten". And that is a valid use case.
I currently do not see how this would be useful in any way (with the exception of the business use case, which i addressed already).
I'm sorry that is the case. This would be useful in: empowering users, as administrators are users of the software after all, and empowering end users, as they can choose where to communicate of their own accord. No one would be forced to enable or disable such expanded ACLs.
Thread
Comment by Noah
Comment by @fluxerhost