Conversation

Notices

1. Embed this notice
   Hrefna (DHC) (hrefna@hachyderm.io)'s status on Thursday, 04-Jan-2024 00:42:00 JST Hrefna (DHC)

   We really need better safety tools and better feed algorithms.

   • Embed this notice
     Hrefna (DHC) (hrefna@hachyderm.io)'s status on Thursday, 04-Jan-2024 00:41:52 JST Hrefna (DHC)

     in reply to

     • Clifford Adams
     • Jenniferplusplus

     @jenniferplusplus

     This is one reason I reject the framing of that there "is no algorithm" here.

     I know what they mean but if we treat what is there as an algorithm, if we acknowledge that, then it isn't a far step to say "and so we can dictate what the algorithm should be doing for us."

     If we fight it because we treat reverse chronology as the One True Way™ then we lose out on a lot of ability to make improvements.

     @wikicliff

   • Embed this notice
     Jenniferplusplus (jenniferplusplus@hachyderm.io)'s status on Thursday, 04-Jan-2024 00:41:53 JST Jenniferplusplus

     in reply to

     • Clifford Adams

     @wikicliff @hrefna honestly, a super effective and easy thing would be to just sort replies any way except reverse chronological. The algorithm I would suggest is 1. posts OP replied to; 2. posts from people the OP follows; 3. posts the OP faved.

     Paul Cantrell repeated this.
   • Embed this notice
     Clifford Adams (wikicliff@fosstodon.org)'s status on Thursday, 04-Jan-2024 00:41:54 JST Clifford Adams

     in reply to

     @hrefna

     I'm not planning to implement this soon (if ever), but think it would allow more open discussion while preventing a lot of abuse.

     OK--now all of you can tell me why this can't work. 😜
     5/5

   • Embed this notice
     Clifford Adams (wikicliff@fosstodon.org)'s status on Thursday, 04-Jan-2024 00:41:55 JST Clifford Adams

     in reply to

     @hrefna

     More thoughts:

     * Perhaps in low-risk situations the original poster could also click-through and approve replies.

     * One would set up accounts with screening services and link them to a fedi account. The original poster's server would forward the messages for approval.

     * I think this could be done with only changes to the original poster's server, not the reply-guy server. It would be nice if the reply server knew of the screening to warn the person replying.

     4/x (almost done)

   • Embed this notice
     Clifford Adams (wikicliff@fosstodon.org)'s status on Thursday, 04-Jan-2024 00:41:56 JST Clifford Adams

     in reply to

     @hrefna
     Reply limits are useful. Followers-only replies is a great idea to require a little effort, but it has some issues IMO:

     * With open following it is too easy for bad actors to follow just for a reply, while *also* being too much of a barrier for many good actors who want to contribute.

     * Approval-required follows are way too much effort for most people. Even if promptly approved, it is likely too late for a good actor who wanted to make a quick positive comment.

     2/x

   • Embed this notice
     Clifford Adams (wikicliff@fosstodon.org)'s status on Thursday, 04-Jan-2024 00:41:56 JST Clifford Adams

     in reply to

     @hrefna

     I am suggesting a new screening option for reply limits. This would be in addition to easy rules like "mentioned people may reply" and "people followed by the original poster may reply".

     The basic idea would be that the original poster somehow lists (maybe on their profile?) one or more screening services that they trust. When a stranger replies, the original poster's server would hide/suspend the message until a screening service approves it.

     3/x

   • Embed this notice
     Clifford Adams (wikicliff@fosstodon.org)'s status on Thursday, 04-Jan-2024 00:41:57 JST Clifford Adams

     in reply to

     @hrefna
     The examples you list sound good--I hadn't considered limiting notifications before. (Probably because I turn off all real-time notifications.) I like extra clicks (like a USEFUL CW--I disable them because almost all are just annoying).

     I can't quite picture a reasonable UI for blocks affecting others in my social network (other than a shared blocklist). I am slowly working on ideas for score-based ranking of notifications and replies, but only with local data.
     1/x

   • Embed this notice
     Hrefna (DHC) (hrefna@hachyderm.io)'s status on Thursday, 04-Jan-2024 00:41:58 JST Hrefna (DHC)

     in reply to

     • Clifford Adams

     @wikicliff I'm always up for a T&S rant.

     Nope, not direct private messages, just ordinary messages.

     The ability to limit replies would be one example, another would be deprioritizing the message so that I don't get notified and have an extra click to show it. Another is that by having me block them others in my social network would be less likely to see their posts.

   • Embed this notice
     Clifford Adams (wikicliff@fosstodon.org)'s status on Thursday, 04-Jan-2024 00:41:59 JST Clifford Adams

     in reply to

     @hrefna
     Were those direct/"private" messages?

     I'm wondering what kind of safety tools you have in mind? I've been thinking about possible chosen-3rd-party screening to help protect frequently targeted people (in both public and direct/semi-private messages). In my idea, a person could choose the level of screening they want from filtering severe abuse only all the way up to filtering out non-constructive or negative messages.

     I'll spare you the full rant this time unless you want it.