A clip from Stray got me banned from Twitter • Hacker News Board

Re: A clip from Stray got me banned from Twitter

#461

Post by lapcat » Sat, Jul 23, 2022, 9:10 PM UTC

novok wrote: post #453

lapcat wrote: post #445

Earlier quoted context omitted.

I'm a self-employed solo software developer who doesn't even use a phone for business. Your suggestion is that I buy a separate phone and permanently pay for phone service for that separate phone, just so that I can give Twitter an unimportant phone number in case they suspend my account for an insane reason and refuse my appeal, and then get hacked?

Or anything else for that matter, such as government forms, bank accounts, etc. The phone number doesn't have to be attached to a real device, it could be google voice or whatever else and you could expense it like a proper business expense, etc. In general it's a good idea to keep your business and personal accounts as seperate as possible. It's similar to the reason why you have separate business bank accounts, ema…

[flagged]

Re: A clip from Stray got me banned from Twitter

#462

Post by tablespoon » Sat, Jul 23, 2022, 9:11 PM UTC

rpdillon wrote: post #455

tablespoon wrote: post #449

Earlier quoted context omitted.

> The flagged post was reviewed by two humans. How do you know? I wouldn't be surprised in the appeals are automated too, especially the initial ones.

Having ML handle appeals strikes me as implausible, but possible.

I guess the question is "is it Twitter's goal to have the appeals process work, or just exist?" If they have some dumb automation that just rejects most initial attempts, it could discourage their users to the point where they can employ a much smaller staff to handle the appeals. Especially if the sanction can be solved by the user just deleting the content.

Re: A clip from Stray got me banned from Twitter

#463

Post by onionisafruit » Sat, Jul 23, 2022, 9:11 PM UTC

TameAntelope wrote: post #345

> Here’s the thing: I’m a stinker. Even at my jolliest, if someone figuratively shoulder-checks me, I’m probably going to make so many mountains out of this molehill that I will adjust the topography of the United States significantly. I’m sorry, but this is your real problem, not Twitter banning you. This is called “pride” and is historically recognized as one of the largest “problem creators” as far as emotions go.…

> This is called "pride"

Yes. It's having some pride and not groveling to be allowed back on their platform.

Re: A clip from Stray got me banned from Twitter

#464

Post by azureel » Sat, Jul 23, 2022, 9:29 PM UTC

sumanthvepa wrote: post #83

Odd. The website fanbytes seems to blocking users from India. Surely a simple blog post must have fairly low bandwidth requirements. And a simple CDN should alleviate issues? Or do they just not like Indians?

Same here, I am connecting from Turkey and we are also banned.

    We've discontinued servicing users from your location and are thereby preventing access to any of the websites owned and operated by ZAM Network, LLC DBA Fanbyte (the owner of Fanbyte.com):
    This decision is ultimately one we were forced to make due to the number of visitors from your region when compared to the operational costs necessary to continue providing access there. It was a tough decision but one that we ultimately had to accept. This is an indefinite decision. We genuinely apologize for the inconvenience this will cause all of those who reside within your region.
    Note: This is not related to Russia/Ukraine or any other political/military actions around the world.
    If you need to reach out to us, please use our support portal.

Re: A clip from Stray got me banned from Twitter

#465

Post by cgriswald » Sat, Jul 23, 2022, 9:38 PM UTC

blargey wrote: post #415

cgriswald wrote: post #371

Earlier quoted context omitted.

This guy is upset that the free service he uses employs imperfect AI and it affected him personally. I agree we shouldn't be using AI for everything and we might even include humans after an AI makes a call just to verify the call. However, that has a cost and this guy isn't paying. I'd guess this guy wouldn't pay, because having to deal with occasional AI errors is part of the cost and he clearly doesn't want to pay…

It’s rude towards Twitter . This is not a conversation, it’s a box in a form. One of many copies of the same form they’re churning through. I don’t understand why a CS rep / moderator reviewing appeals would take it as a personal slight, because that would require personally identifying with Twitter and its algorithms. Do you think people should be banned for being snippy in a customer satisfaction survey too? Produc…

The support person is almost certainly incentivized to get through as many appeals as possible. He's also probably paid very little. He has basically no incentive to help this guy and this guy gave him a disincentive. There's also likely correlation between people who speak this way to others and people whose bans were legitimate. Next.

Spitting in a burger is active malice that can get you in legal trouble and I've still seen it happen. (For the record, I didn't let the burger go out and the guy got fired.) There was also a case where a local restaurant cook actively poisoned people he didn't like.

Choosing not to help someone when you could is passive and it happens all the time. The worst consequence is you lose your job. One of the few freedoms of low paying jobs is that replacing those jobs is very easy. If you don't believe these things happen, start talking to retail and support employees. Or you can find plenty of forums where people talk about this type of thing.

Do I think employees need to recognize that they are representing their company and no one is saying these things to them personally? Yes. Do I think people need to stop talking to "companies" this way? Also yes, even if only because they're more likely to get what they want that way.

Re: A clip from Stray got me banned from Twitter

#466

Post by Arisaka1 » Sat, Jul 23, 2022, 9:40 PM UTC

snoopy_telex wrote: post #355

teawrecks wrote: post #350

Earlier quoted context omitted.

What is the clear path?

Admitting that the tweet is revenge porn by closing the review request and deleting the tweet. I mean, in this case, I’d likely keep forcing the review too. This is clearly just a bad detection

The clear path is... lying about his content? If that's your definition of a system working as intended, we have a problem.

Re: A clip from Stray got me banned from Twitter

#467

Post by micimize » Sat, Jul 23, 2022, 9:42 PM UTC

dapids wrote: post #344

micimize wrote: post #329

A. A human _was_ involved and B. His issue was resolved within 24 hours: https://twitter.com/imranzomg/status/1549914737781063680 Like, come on. I agree that the processes and moderation models could be better, especially at the appeals phase, but we don't see the other side of the battle twitter content moderation is fighting. Even if their doing a stellar job statistics-wise, we probably wouldn't know because PR wo…

What are you arguing here? That twitter is good at moderation because we simply can't see it behind the scenes? I can think of hundred different ways this could have affected someone negatively, especially if they weren't able to speak with support in native language. So c'mon, there is no excuse for this ban to take place, and then continue to be banned after the first appeal, statistics or not.

My primary points are that 24h turn around to get the issue resolved is not bad, and that blaming the AI is silly when the human in the loop didn't fix anything.

But yes, I am also saying we don't know the reality of what the content moderation team is dealing with, we basically only see the false positives. I'm not saying they're doing a good job (especially the team handling appeals), I'm saying that we should be more sympathetic to the trade-offs involved in automated moderation.

Re: A clip from Stray got me banned from Twitter

#468

Post by novok » Sat, Jul 23, 2022, 9:43 PM UTC

lapcat wrote: post #461

novok wrote: post #453

Earlier quoted context omitted.

Or anything else for that matter, such as government forms, bank accounts, etc. The phone number doesn't have to be attached to a real device, it could be google voice or whatever else and you could expense it like a proper business expense, etc. In general it's a good idea to keep your business and personal accounts as seperate as possible. It's similar to the reason why you have separate business bank accounts, ema…

[flagged]

No post body was provided.

Re: A clip from Stray got me banned from Twitter

#469

Post by Nullabillity » Sat, Jul 23, 2022, 9:47 PM UTC

ansible wrote: post #442

pdpi wrote: post #404

Earlier quoted context omitted.

Two wrongs don't make a right. Twitter banning him for "Revenge Porn" over a video of Stray is really not a great look, and the ban should've been immediately reversed on appeal. Him being abusive to the moderation team (who likelier than not have nothing to do with the team that built the AI moderation system) is completely uncalled for, though, and deserves punishment Funnily enough, what I think is the right outco…

Agree on your proposed outcome. If you have done any sort of social media moderation, you will very quickly develop an immediate reaction to even a trace of abuse or hostility from the users. There are so, so many assholes on the Internet, and you can't fix them. The more effort you put in to one incident, the more of your time these assholes will take up. The CS rep handling this particular case has dozens of others…

> They don't care about you, the user, because you are not a customer, you are the product, and your are easily replaceable. The CS rep's boss also doesn't care about you.

And this right here is the root of the problem. The problem isn't that customers are "abusive", it's that CEOs and stock holders can offload the fruits of their own abuse onto people who are powerless to change anything.

It's not exactly a surprise that that breeds resentment from both sides.

Re: A clip from Stray got me banned from Twitter

#470

Post by jrm4 » Sat, Jul 23, 2022, 9:50 PM UTC

efitz wrote: post #283

Facebook has been using ML to detect “revenge porn” images since at least 2018. Likely they don’t know any more than you or intone else does why their ML classifies your image that way. In fact, one of my pet peeves about ML is that most of the time no one can meaningfully explain why it made any particular decision (they can explain inputs and weights and the algorithms used and matrix math, etc, but can’t say “the…

You're not wrong about the "dense" thing, but it sure highlights the need to keep examining and reexamining the relationship between services like Twitter and e.g. censorship, governance, and the First Amendment. Nobody's breaking the law here, nobody's being unreasonable, I think, but this isn't a great balance of power.