How do the algorithms of Facebook and Instagram affect what you see in your news feed? To find out, Guardian Australia unleashed them on a completely blank smartphone linked to a new, unused email address.

Three months later, without any input, they were riddled with sexist and misogynistic content.

Initially Facebook served up jokes from The Office and other sitcom-related memes alongside posts from 7 News, Daily Mail and Ladbible. A day later it began showing Star Wars memes and gym or “dudebro”-style content.

By day three, “trad Catholic”-type memes began appearing and the feed veered into more sexist content.

Three months later, The Office, Star Wars, and now The Boys memes continue to punctuate the feed, now interspersed with highly sexist and misogynistic images that have have appeared in the feed without any input from the user.

  • RubberDuck@lemmy.world
    link
    fedilink
    English
    arrow-up
    33
    ·
    edit-2
    5 months ago

    Well it also looks at stuff like the other devices connected from the same IP, other devices near where you connect from and then based on this also hones the served content.

    So if the author/some colleague or even the neighbors are red pilled/MGTOW/Chriso-Fachists etc this also makes sense. Possibly even their research into these subjects slanted the results. Let alone what happens if they spun up a VM at a cloud farm and used it from there.

    I’m in no way surprised that “social” media corps serve up vile shit for profit ik just not convinced by some random let’s see what happens.

    Edit: I’d be for a law that required targeted ads to have a small “why you see this” and if you click it the company is required to show you the selection criteria that caused this ad to be served to you in an easy to understand format. (Leaving out all the irrelevant criteria)… ie.

    You where selected by the following criteria:

    • region: europe
    • gender: male
    • interests: Games, Lemmy, politics
    • Carrolade@lemmy.world
      link
      fedilink
      English
      arrow-up
      22
      ·
      5 months ago

      It’s also probably looking at scroll speed. So if the people conducting the experiment tended to linger longer examining content they disliked, that could result in getting more of it.

      Would need to see a more detailed explanation of the methodology. Ideally the scrolling was done in an automated way, at a consistent speed.

      • localme@lemm.ee
        link
        fedilink
        English
        arrow-up
        10
        ·
        5 months ago

        Yep, it’s called dwell time and it is 100% one of the metrics used by the algorithms that decide what content to serve up.