• ProdigalFrog@slrpnk.net
    link
    fedilink
    English
    arrow-up
    16
    ·
    edit-2
    5 hours ago

    The Fediverse has a lot more safeguards in place, in particular the ability to require a message to register an account, such as my instance requires, weeds out 99% of bots.

    We can also defederate from instances that become overwhelmed from bots if they have lax sign-up requirements (already happened a few times), which vastly limits their ability to take hold.

    The bigger problem for us, I think, is the fight against bot scrapers. Anubis is keeping them at bay for now, but it will likely be an ongoing cat and mouse game until the AI bubble bursts.

    • Upgrayedd1776@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      3
      ·
      9 hours ago

      the moderation and comment moderation history is nice but it is sort of hidden. that would be nicer if it was more intuitively inline. Also did you mean “bot scrapers” rather than “not scrapers”. And I have been toying with the idea, what about ai or bot supported pipelines, accept that they are part of the visitors, provide it optimized data, and then restrict ui/ux heavy processes to stricter rate limits per second, etc…sort of like a robots txt v2 or something

      • ProdigalFrog@slrpnk.net
        link
        fedilink
        English
        arrow-up
        3
        ·
        6 hours ago

        It’s less hidden for admins. Whenever we click on a user’s profile, there’s a button that can take us directly to their individual mod-log history, which can help us quickly see if there’s a history of AI or bot spam behavior that was spotted by other admins/mods.

        Also did you mean “bot scrapers”

        Agh, auto correct got me. Yes, I meant bot scrapers.

        As for the robots.txt v2 idea, I’m not sure that would be very effective or popular amongst admins, as I think most would prefer not to give it any data if possible, and even with reduced rates, there would probably still be enough bots and queries to present a significant issue.

        I think there have been some experiments with ‘trapping’ bots in a recursive loop that they don’t realize is a loop, and therefore can’t escape, but I’m not sure how effective those have been.