@Felthry sometimes i fantasize about being able to grab posts and just toss them like they're physical objects
and there could be heuristics there like, once you've thrown away x posts by y user, you start seeing less or none posts by y user
there's a lot of potential here :o
@bleak ehhhh I don't like that personally
If I follow someone I want to see all their stuff, not some randomly selected subset of it
@Felthry you do the selecting though
@bleak but if it starts showing you fewer posts from that person, how does it know which ones to avoid?
@Felthry i mean what i was imagining mainly was just once you throw away a certain number of posts from a certain user it just stops showing them to you entirely, and the reduced frequency aspect of it was kind of a spitball i wasn't married to
@bleak That's not really any different from muting them though, and I'd prefer that be done intentionally--the most I'd sign on for is some kind of notification "you seem to be hiding a lot of things from this user, would you like to mute them"
@bleak ehh, I just don't like not having control of what's shown to me honestly. If some opaque algorithm determines what I can and can't see, I can't tell if it's working right because I don't know what it's hiding from me (also there's the possibility of abuse with such a system. less here than on other social media but i've heard bad things about twitter and facebook's algorithms (i never used either though))
@bleak yeah, but at that point I'd prefer the algorithm be exactly what you write--nothing that gets changed based on usage. write it once and it stays that way, basically
@Felthry that'd be cool but i don't know how to code so idk how useful such an interface would be for me. maybe having a 'stop learning' button would be really nice on the thing
@bleak oh i'm pretty crap at coding too, but i can at least read code and get some idea of what it's doing, so i wouldn't have to worry about it doing something suspicious
@Felthry yeah. i think ideally for my thing, if the code was self-modifying you'd be able to look under the hood and tinker with it directly if you wanted to too, with the addition of the actual larger codebase being open source
@Felthry i'm getting the idea that people would want to modify how the thing would even learn, so i'd probably have a space for the user to get under that particular hood as well
@Felthry yeah, my intention for this fantasy of mine is to like, sort of get rid of all the old paradigms and let you write the algo, modify it, or reset it to 'factory defaults' even, yourself, simply by using the interface. if it starts acting up, you can easily shape it back the way you want it, or say fuck it and clear the use history or w/e.
it doesn't need to be opaque or tucked away from the user at all, is what i'm putting down here