• Soupbreaker@lemmy.worldEnglish
    3·
    8 hours ago

    It’d be great if people just stopped submitting to these algorithms. Every time I hear someone on a podcast talking about “their” algorithm, as if it were some benign, cutesy thing, I want to puke. Just quit the corporate bullshit and come absorb depressing content on Lemmy like the rest of us, jeez.

  • chicken@lemmy.dbzer0.comEnglish
    5·
    12 hours ago

    Ok so how do you actually evade algorithmic censorship then? I assume the embedding still is based on transcribed text, so word choice should matter on some level even if it isn’t the be all end all right? The metric of other content preferences of people who prefer your content does seem harder to get around though.

    Anyway here is the link to the paper the video mentions: https://concetticontrastivi.org/wp-content/uploads/2023/01/1369118x.2016.1154086.pdf

    • lepinkainen@lemmy.worldEnglish
      3·
      6 hours ago

      Nope. There are studies with vector databases that show that even language doesn’t matter, the words start grouping together automatically based on relevance just by the way the math works.

      In theory your could try inventing a fake language so weird that it doesn’t match anything existing, but at that point just start encrypting your stuff

      • chicken@lemmy.dbzer0.comEnglish
        2·
        5 hours ago

        the words start grouping together automatically based on relevance just by the way the math works

        Sure but isn’t it still the words that are grouping together? The guy in the OP video seems to be claiming that the fact that he used certain words does not matter, which does not make sense to me, since the depth of understanding these algorithms have of what is being said is still somewhat shallow.

        I would guess that it should be possible to engineer a sentence that communicates a particular message, but is phrased in such a way that it targets a location in vector space that is not associated with that message (until the other parts of their system make that association).

  • otacon239@lemmy.worldEnglish
    33·
    19 hours ago

    I’ve said it before and I’ll say it again.

    Self-censorship is the worst kind because you’re not even initially trying to get the original message out. You’re doing the advertiser’s work for them when you make your posts friendly to the algorithm.