I noticed that there were some accounts that were hijacked by the instance owners. All the posts from that user were then edited to say what happened.

This kind of surprised me, I figured instances could delete posts, but not edit them. So how much control do they have?

I assume they can’t see my password (hopefully). Can they post in my name? Do they have all the access to my posts to foreign instances that they do over local posts?

Edit: thanks for all the responses everyone! I’ve wanted my own instance for a while, but maybe I’ll get on it now

  • static09@lemmy.world
    link
    fedilink
    arrow-up
    17
    ·
    edit-2
    1 year ago

    It’s important to note that your password has to be stored someway, no matter what, no matter where. How it’s stored can be varied, from hashed (think encrypted) to cleartext. I’m assuming lemmy is using hashed passwords, so if you’re concerned about your password being available to an instance owner, admin, or potential attackers, then you’ll need to follow safe password guidelines. Changing the concept from passwords to passphrases is a great start.

    Always keep in mind, if the data isn’t stored on your device, you do not technically own that data. You have to trust the owners to be good data custodians and treat the data you give them as if it were their own private data.

    I’ll leave this now internet-ancient sacred image for future passphrase converts.

    • Just to be clear, you’re talking about Lemmy. There are authentication mechanisms in which the instance never has access to your password, in any form.

      I wish I could upvote your comment multiple times for the XKCD comic on this. And to anyone curious, there are many tools which will generate XKCD passwords for you. One’s called “correcthorse”, and there’s another called “correctpony”. A github search for “xkcd password” should turn up several.

      • karlthemailman
        link
        fedilink
        arrow-up
        2
        ·
        1 year ago

        Why can’t the admin just change the Lemmy source code to not hash anymore?

        • Are you asking why can’t a bad admin change the code so that they can more easily steal the password? They could, and this is what OP was saying about trusting the admin. What I was saying was that there are client-side auth mechanisms, where the admin never has access to the password. But Lemmy could also implement OAuth, or a similar federated identity pattern, where (again) the Lemmy admin never has access to any form of the password.

          I’ve never run a Lemmy instance; it’s possible the server software supports SSO but few instances use it.