SDF ARPA member & Saint

  • 1 Post
  • 27 Comments
Joined 1 year ago
cake
Cake day: June 14th, 2023

help-circle








  • I think that would only work when the number of instances is small. Two solutions to this might be:

    • have instances act like relays where the home instance of a community notifies 10 instances and then each of those instances notifies 10 instances, etc.
    • batch updates on a timer such that once a minute all posts, comments, boosts, etc within a minutes are buffered and sent together.










  • That was an incredibly comprehensive, well articulated, and dare I say, exhaustive essay on some important issues you raised. On top of that, creating sample documents is next level.

    Privacy

    I don’t think the word “privacy” is a good word for the concept. I believe “user data control” or “right to be forgotten” is more appropriate for the “deletion issue”. However, there are few privacy issues such as instance admins having access to private messages and the potential for a hack to expose users e-mail addresses and usernames.

    I believe you are 100% correct that we need to do a much better at communicating exactly who has access to their data and what (if any) control they have over that data once it is federated. I don’t believe we will ever have an guaranteed federated delete, and we need to make that crystal clear so users can proceed accordingly.

    Legal

    Running a self-hosted service is one thing, but running a public service raises a myriad of legal issues. In the US, children under 13 must not be allowed to have accounts (COPPA). CSAM (child pornography) is another problem that can expose admins to serious repercussions. In the US, it is not enough to delete it, it must be reported to the NCMEC. Federation will make this especially treacherous. Other issues such as criminal investigations, subpoenas, and possibly even national security letters are not a matter of “if” but “when” they will occur.

    If Lemmy continues to grow, instance admins will need to be prepared for these issues. I would suggest that the public instance admins reach out to an organization like the EFF who has experience dealing with these issues. If not, I’m afraid a high profile incident may be all it takes to kill it.