0

How are categories (eg: Porn) maintained? Can this be abused?

I just tried following a link to journalofcontroversialideas.org. It didn't work. I checked my NextDNS logs and it was blocked as a porn site? Huh? Take a look it: it's not porn.  This is from its Wikipedia page:

"The Journal of Controversial Ideas is a cross-disciplinary, peer-reviewed academic journal that aims to allow academics to publish using pseudonyms if they request it. Established in 2018 by philosophers Francesca Minerva, Jeff McMahan, and Peter Singer, the journal began accepting submissions on 20 April 2020. According to McMahan, the journal is needed because of the fear among academics about publishing articles that support certain contentious positions."

I tried two random URL category lookups and neither mark it as porn:

The bigger questions here, as far as I'm concerned, are:

  • Where does NextDNS you source URL categories?
  • Do they do any manual checking?
  • How easy is it for someone to abuse a category to block sites they don't "like"?
  • What isthe process for requesting a recategorisation?

I would hate to think that NextDNS is just an extension of cancel culture, providing suitably motivated people with a way to block what they don't like. So a satisfactory response  / explanation is essential if I'm keeping my account.

5 replies

null
    • Hey
    • 2 yrs ago
    • Reported - view

    I've read about this before and I'll forward what they said on another thread.

    They're using AI for other things now that aren't purely threats and they've from what I've heard started using AI on Porn/Parental Control as well.

    I've seen some that were saying they're losing control or it's some political thing as it blocked some site that was associated but it's a logical thing for an AI to make mistakes and have a few flaws at the start.

    I remember threat protection itself blocking a few sites earlier on and that's not an issue as of now, so it's the process of the AI maturing to an extent where it has little to no false positives.

    I personally like it as yeah it might have a few false flags now but as time passes by, it's going to be a feature that truly helps with proactive responses in terms of parental control instead of reactive lists.

      • Paul_Rafferty
      • 2 yrs ago
      • Reported - view

      Hey I don't accept that this is an AI screwup. Porn is very specific. If it miscategorised an image-heavy site, fair enough. Or one with videos. Or... I mean, anything remotely resembling Porn. But check out that site. It's a boring text website that discusses philosophy. It's FAR more likely that somebody doesn't like the content, than AI screwed this up. Either way, I'm not happy letting NextDNS act as my DNS / filter until I know more about how it works and how this site got categorised as porn.

      Is there any official NextDNS support? Or do we all have to 'support ourselves' here on a forum? Because if I'm paying, I want a NextDNS employee to chime in.

      • NextDNs
      • 2 yrs ago
      • Reported - view

      Paul Rafferty there are many parameters used to detect porn websites and millions of domains that fall in that category. We are constantly working on improving our algorithm to avoid false positives without lowering our protection. The idea our staff would spend their time adding domains they don't like in the porn filter can't be serious.

      • Paul_Rafferty
      • 2 yrs ago
      • Reported - view

      NextDNS At no point did I suggest that your staff are doing anything nefarious. What I am suggesting is that, depending on where/how you source domain category info, there is the potential for this type of abuse. I worked with a large corp that got added to a popular adblock list because of a coordinated campaign on social media. It interrupted their site for a subset of users for a week before the adblock maintainer 'eventually agreed to unblock'. In the meantime, the company was at the mercy of some dude out there on the internet, possibly in a foreign country (making litigation difficult). I've seen entire communities banned from Reddit because of coordinated campaigns by SJWs who simply didn't like the ideas they were discussing (and no, there was no hate or rule-breaking, just mass reporting and eventual ban). Since I can't see the inner workings of your system, I can't know that it's immune from this type of abuse.

       

      So, is there an answer to the question: How do you categorise sites? How can anyone trust these category filters without knowing at least the underlying mechanisms? How do I report a false match? How do I follow up on the report?

    • Mark
    • 2 yrs ago
    • Reported - view

    Another site that is being blocked as porn but is not is: http://thepostmillennial.com/

    This is a right/conservative website. I hope this will be corrected as well. 

Login to reply

Content aside

  • 2 yrs agoLast active
  • 5Replies
  • 218Views
  • 5 Following