5 ms·
Here we are, deep down in the dystopia of Automatic Content Classification by robots. Far less important of an example, but I was just put in "Facebook jail" f
by cmrdporcupine 4y ago
Here we are, deep down in the dystopia of Automatic Content Classification by robots.
Far less important of an example, but I was just put in "Facebook jail" for 24 hours for posting a picture of my son at the beach in his bathing suit with no shirt. Y'know, as one does at the beach. I can only assume it's because my son has long hair and the Convolutional Neural Network or whatever decided he was a girl and therefore I'm a pervert.
Sadly it was the "appeal" I submitted that got me blocked. Presumably by a "human", but who knows.
Before I appealed they were simply going to not show the picture. Appealing got me in "trouble." That might be even worse than the original misclassification. On top of that, if I was actually a "community standards" violator who posted potential child exploitation imagery, I'm not sure how a 24-hour ban on activity on Facebook is of any use, either? Except I'm terrified to imagine a world where Meta might have called police on me based on the output of a neural network image classification.
Others have said but I'll say it again: This kind of business doesn't scale ethically. You can't have billions of people on a bulletin board. It doesn't work. Moderation is essential to modern communication. But you can't do moderation automatically or at scale and in a universal way.
Very dark patterns emerge the moment you go FAANG scale and toss algorithms tuning for advertising and "engagement" into the mix, and attempt to do so with the help of computers.
"Sad" as it is, we will need to "retreat" back into smaller forums and BBSs where communities self-police.
Facebook has infiltrated so many aspects of society. Want to interact with the parents from the local school your kids go to? You have to do that on Facebook. Event announcements? Keeping in touch with your distant aunt? Facebook.
If something like Facebook is really a universal utility, it will have to be put under public administration; like the post office. But clearly, that isn't going to happen and would have other problems.
I am going to have to find some other way to engage with old friends and family.
Frank Herbert had some intuition with his whole "Butlerian Jihad" thing.
- jquery 4y agoWe don’t just need human moderation, we need due process. These companies control too much of our digital lives and make so much money from us, but have zero regard for us as soon as it’s inconvenient to investigate a case, because that would cut into their insane margins. I was permabanned for Twitter recently merely for getting the attention of some large influencers who disagreed with me (no actual rule was broken, but I received enough reports that my account was nuked). I appealed multiple times but Twitter’s appeal process is a sham for us little people. I doubt any humans ever looked at my account.
- root_axis 4y agoThey only control what we voluntarily give them. Social media controls nothing of importance in my life.
- BlargMcLarg 4y agoEmphasis 'we', not 'my'. These guys are already making shadow profiles out of info given by friends, corporations, etc. Not participating is also causing red flags in certain circles. Withstanding peer pressure is one thing, having your identity made up or flagged out of your control is another. This is a slippery slope that should be tackled before it gets to that. The only people not affected indirectly, are the people who will die without children or younger cohorts as friends.
- root_axis 4y ago> These guys are already making shadow profiles out of info given by friends, corporations, etc. True, it's a shady practice indeed, but signing up and giving them even more information directly from the source is obviously far worse than the fraction of signals they can extract from your friends. > Not participating is also causing red flags in certain circles. There's no accounting for the peculiarities of social groups, you could say the same thing about refusing to smoke weed or drink alcohol, it doesn't mean those vices are vital, and social media is no different.
- BlargMcLarg 4y agoI don't think you fully grasp what the slope is sliding to. There are enough token anecdotes of companies doing background checks on social media, and will actively flag individuals for having zero presence. We also have social credit score horror stories. Your answer doesn't work anymore when lack of participation is considered wrong. We should be blocking that instead of assuming things will just work out forever as long as individuals guard their identity. Again, this goes beyond just standing up against peer pressure.
- saurik 4y agoIf you are only using Facebook for its original purpose of keeping up with old friends and family, then it doesn't actually have the scaling problem: if one of your friends or someone in your family starts posting a ton of racist bullshit you either confront them about it or drop them as a friend (preferably bilaterally), and either result is actually better for society than having Facebook attempt to present a skewed view of them that tries to just pretend they aren't posting all of the horrible stuff in the first place (whether by blocking it from being posted, quickly removing it once posted, or running some complex ranking algorithm that does a good job of hiding it). It is only when you start having strangers talking to strangers that you run into a need for moderation, and even there you should be able to scale by sharding and punting the problem to others: if you have a group--similar to a real-world club--the moderation is on you, as the issues in your community shouldn't leak to people who haven't joined your community (and if people leave your community because you fail, all the better). The only real issue is that Facebook wants to--for the increased engagement, and thereby ad revenue--run a ton of recommendation algorithms that shove content from people you have no affiliation to in your face constantly (which one might notice should already be considered antithetical to the design of a social network) which leads to a ton of stranger-to-stranger interactions that are entirely "on Facebook" to ensure are clean.
- cmrdporcupine 4y agoAnd yet here I was, posting a picture of my kids at the beach so my mom and friends could see them, and I ended up banned from participating on Facebook for 24 hours and accused of violating community decency. They're f*cked.
- saurik 4y agoYeah... so I honestly consider this to be a separate (also horrible, for avoidance of doubt) problem than "moderation"? Like, Apple and Google aren't exactly having a moderation problem when it comes to attempts to curtail CSAM stored on their photos platforms (which has led to their automated flagging systems and then need to scale to the entire world of users appeals and escalation, and then the subsequent concern about "are they going to call the police on me when I fail my appeal?!")... it is more of an attempt to deal with awkward regulatory "think of the children" overreach and hostile American law enforcement. Facebook seems to me to be doing the same thing here and failing. Imagine a world where social networks were built by people who simply didn't care about "engagement" at all and weren't being motivated by ad revenue... I think you could design an end-to-end encrypted version of the system where no-one except your friends--and certainly not the network operators--even knew what you were posting in the first place and they would hopefully be able to avoid installing client-side filters for CSAM (but, with stuff like FOSTA and SESTA, maybe not?). This model should even work, I'd think, for Twitter/Instagram-like broadcast models (though the legal implications of the well-known "secret" key being published and accessible to the network might lead to various problems; you might have to go fully-decentralized).
- tomohawk 4y agoThat's life under the Techiban
- swayvil 4y agoIn Science We Trust.
- protomyth 4y agoI sometimes wonder what would happen (it won't) if some fed-up Congressperson drafted a bill that would allow folks to sue Facebook, etc. for libel when it accused you of being a ped0, criminal, or other malcontent. Perhaps they would be a bit less happy to accuse their users of vile things.
- toomuchtodo 4y agoSec 230 repeal? Or something adjacent to that strategy? https://www.politico.com/news/2022/09/08/white-house-renews-call-to-remove-section-230-liability-shield-00055771 https://www.politico.com/news/2022/09/08/white-house-renews-... https://www.whitehouse.gov/briefing-room/statements-releases/2022/09/08/readout-of-white-house-listening-session-on-tech-platform-accountability/ https://www.whitehouse.gov/briefing-room/statements-releases...
- protomyth 4y agoMore in the way of a specific law. I don't think a full Sec 230 repeal is a good idea. I really, really want services to have to specify what you did wrong, and you should have some actual options to deal with these services. They have become utilities and need to have some accountability when they call you something vile.
- tinus_hn 4y agohttps://www.theregister.com/2022/09/16/texas_social_media_law_appeal/ https://www.theregister.com/2022/09/16/texas_social_media_la... Depending on what the Supreme Court decides this might be less far away than you think.
- nradov 4y ago"We" don't need to retreat back into smaller forums. Despite the huge number of false positives flagged by content review systems, they still impact less than 1% of active users. So everyone else will continue using it. Facebook is terrible and unethical in many ways but it's still the fastest, most convenient way to share pictures and updates with friends and family scattered across the world. I don't have enough hours in my day to pursue other options.
- 650REDHAIR 4y agoI’ve flagged 100s of illegal firearms sales on Reddit and FB and not a single one has been taken down.
- nradov 4y agoPerhaps, but so what? At that scale there will be huge numbers of both false positives and false negatives in any content moderation system. If you dig into any popular online classified sales site you can find some illegal items. Criminal activity is quite a different thing than censoring legal content which possibly violates corporate terms of service. If you have evidence of an actual crime then you should report that to law enforcement instead of expecting a private for-profit company to handle the incident. And absent further hard evidence, I am frankly skeptical of your claim. Most people aren't experts on the nuances of firearms sales laws in various jurisdictions, so a post that appears to be soliciting a crime might be entirely legal (or vice versa). I don't really use Reddit, but I've been on Facebook for years and have never seen a post for an illegal firearms sale. Do you at least have some screen snapshots?
- soulofmischief 4y agoIt reminds me of the difference between hunting and killing your own game, and factory farming. Something sinister emerges at those scales.
- soundnote 4y ago> Except I'm terrified to imagine a world where Meta might have called police on me based on the output of a neural network image classification. Don't worry, Google's already doing it. And destroying your digital life if you've put too much stuff in their care. With the characteristic Google customer service of "go fuck yourself". https://archive.ph/W41mf https://archive.ph/W41mf
- autoexec 4y ago> Before I appealed they were simply going to not show the picture. I wouldn't be surprised if Facebook didn't just "not show the picture" and they actually forwarded your info to police or some three letter agency. You're probably flagged already. Probably being watched more closely as a result of your "perversion" and clearing up the matter with Facebook may not matter to anyone else they shared that particular data point with. I agree you should get the hell away from facebook, but this kind of thing won't stop there. Apple wants to scan your personal files and so does Microsoft (last I checked Windows 10 already records and sends to MS the filename of every image you open in their "Photos" app and how long you spent looking at it). When your OS is acting against you, or even your cameras there won't be anything you can really stop using.