5
devmg
5y

Why do we keep developing algorithms to find us things we are already interested in? I'm tired of never being subjected to anything new, and I'm tired of not getting all the raw data. If you have to add a filter for me, please at least make it optional, or customizable. And make sure you make a matching algorithm distinguish uncertainty from negativity. Not knowing if I like x and just predicting I don't like x just because I've used y is not good enough. It's horrible.

Comments
  • 3
    @instagram For the love of God let me turn your shitty ass feed algorithm off. And fucking fix your suggestions algorithm. I never ever look at anything photography related, yet my suggestions is full of pictures OF CAMERAS. Or let's not forget the time the suggestions were full of memes in different languages, something else I never look at.

    Sorry for ranting on your rant.
  • 1
    @jhh2450 haha I hear you. We're here to rant, aren't we? Taking out all our aggression and turning it into back patting. Love the idea, gotta say.
  • 1
    I have been thinking about that and i think at least they should improve algorithm. I call current algorithm "snowballing effect" which creates bubbles of stuff you might like which is highly changeable and wrong . People need contrast. I understand intention but execution is wrong. I have better algorithm which I am trying out. Instead of promoting "good" content suspend bad one.

    Let's clarify with example. Current system promote more stuff from same category or relative to something you liked/share/open.
    New system would keep all content align unless somebody doesn't like it. You push this content down that it will be less likely shown to others.

    This way you are left with good content overall instead of bubble.

    Any thoughts? Hope you get the idea.
  • 0
    @danielstaleiny I think that makes a ton of sense. In a sense, that's snowballing too, but with a subtraction from the main set. Should be very easy to switch to.

    Facebook should do this for sure, it align with their goals.
  • 2
    For the devRant algo we kind of try to do this. It of course shows you things similar to what you've liked in the past, but also new things that people like in general and, collaborative filtering based on what users with similar tastes to you also like, even if it's not similar to something you specifically liked in the past.
  • 1
    @dfox that makes sense I guess. I guess I'm just so afraid of the filter overall. What bugs me is I have no idea what I'm missing. I realise that what I really lack is the control. There are lots of subtle repercussions that I haven't thought about though, like how is the community affected when we see "bad" posts more often? Will people with minority views get bashed? It's crazy hard to answer. I guess all I really can say is I like devRant so I guess you've done it right. But it could be that people are just great here overall too. It's too much data... :P
Add Comment