What are these algorithms you talk about? Because many people are not that much into tech the word “algorythm” has become very abstract. A recommendation algorythm is a piece of code that orders things according to specific parameters. What are the parameters that you want your timeline to be sorted by? I want to understand I like my timeline chronological. You want it to be sorted by what is interesting. How do you deterine what is interesting to you?
You want it to be sorted by what is interesting. How do you deterine what is interesting to you?
People don’t need to science it. They can choose different modes from the dropdown, stick with what they like for whatever reason, or play around. Or even ignore the option altogether. Personally, I use ‘New Comments’ as my default, inspect communities with ‘New’, and occasionally switch to ‘Hot’, ‘Active’ and ‘Top X Hours’ when I’m looking for more.
In terms of manipulation, I guess the biggest lever here is to which instance I log in, followed by which communities I subscribe to. This heavily influences the type of content I see, the political leaning, and things like that. How this content is sorted into a feed is a minor decision in comparison.
Is it possible we talk about different things. OP was talking about Microblogging (e.g. Mastodon). Here on lemmy/kbin that is totally different. It wouldn’t work without it. I don’t think this style would really work with microblogging though I have nothing against trying that. I was more thinking about real recommender algorithms like “Things you might like”.
Admittedly, I spent very little time on Mastodon. But as I remember it, there is something like a ‘home feed’. And I also remember only seeing the most recent entries at the top, which is not necessarily what I would have found the most interesting. For example, I think I’d be at least equally interested in entries with engagement, where people talk about the post. Which sometimes requires some time to pass.
You’re right I lost track about the precise topic, sorry for that.
What are these algorithms you talk about? Because many people are not that much into tech the word “algorythm” has become very abstract. A recommendation algorythm is a piece of code that orders things according to specific parameters. What are the parameters that you want your timeline to be sorted by? I want to understand I like my timeline chronological. You want it to be sorted by what is interesting. How do you deterine what is interesting to you?
Currently, we have these: https://github.com/LemmyNet/lemmy-docs/blob/main/src/users/03-votes-and-ranking.md
People don’t need to science it. They can choose different modes from the dropdown, stick with what they like for whatever reason, or play around. Or even ignore the option altogether. Personally, I use ‘New Comments’ as my default, inspect communities with ‘New’, and occasionally switch to ‘Hot’, ‘Active’ and ‘Top X Hours’ when I’m looking for more.
In terms of manipulation, I guess the biggest lever here is to which instance I log in, followed by which communities I subscribe to. This heavily influences the type of content I see, the political leaning, and things like that. How this content is sorted into a feed is a minor decision in comparison.
Is it possible we talk about different things. OP was talking about Microblogging (e.g. Mastodon). Here on lemmy/kbin that is totally different. It wouldn’t work without it. I don’t think this style would really work with microblogging though I have nothing against trying that. I was more thinking about real recommender algorithms like “Things you might like”.
Admittedly, I spent very little time on Mastodon. But as I remember it, there is something like a ‘home feed’. And I also remember only seeing the most recent entries at the top, which is not necessarily what I would have found the most interesting. For example, I think I’d be at least equally interested in entries with engagement, where people talk about the post. Which sometimes requires some time to pass.
You’re right I lost track about the precise topic, sorry for that.