3 ms·
To be somewhat fair to Meta; the "violation of unclear rules part" part is an appropriate response for spam detection. Usually any entity dealing with the web
by noirscape 3y ago
To be somewhat fair to Meta; the "violation of unclear rules part" part is an appropriate response for spam detection.
Usually any entity dealing with the web don't want to reveal how their spam detector works because doing so could in turn reveal information on how to circumvent it[0].
The issue is that Meta thinks they're large enough that they can apply this to every single one of their business customers and all other parts of their policy as well - normally a company will be inclined to keep the businesses it's partnered with happy because losing their business can have a knock-on effect that could result in a bigger knock-on effect.
The sad thing is that Meta is probably correct in this assessment - losing one business wouldn't hurt them, regardless of spending. Even if you put a few thousand a month on the line, it's not worth for them to consider you. Heck even if you tell your friends about it - Meta doesn't give a shit.
The true Kafka-esque nightmare here though is that even just knowing someone at the company is apparently not even enough to get it properly checked out and fixed (even corporations like Google who have similar black box policies to most users have the means to do that). It seems Facebook runs on "computer says no" logic when it comes to cases they don't have data on anymore. That is truly disgusting.
[0]: This in part because most spam detection is ultimately still just checking heuristics; revealing the heuristic simply reveals what stat to adjust for. There's very few "sophisticated" detection methods - even the big email providers still just rely on heuristics combined with an internal trust score that they gain by checking how users respond to email.
- sideshowb 3y agoOr they could, you know, spend some of their billions on employing humans to review these cases.
- dcanelhas 3y agoWould that generate even more billions in revenue, though?
- intended 3y agoMeta does spend on humans to review cases though. Which is why it seems really strange that this was a lifetime ban, and so quickly at that. Maybe it didnt even get to human review?
- Hackbraten 3y agoAccording to one of the screenshots in the article, both the ban and the appeal were decided on by technology.
- ilovetux 3y agoBut then where would all the money to burn on VR product pipe-dreams come from? /s
- concordDance 3y agoThe spammers can produce far more cases than the humans can review. Welcome to the world of automation.
- bombcar 3y agoYou can put review on a system requiring additional ID/information/escrow payment. A genuine business will be willing to put down some information and a cash payment for a review; spammers won’t have the time or resources. You run the risk of a perverse incentive to trigger it for revenue but make the fee refundable or something.
- SpicyLemonZest 3y agoCan you? I don't know that I've seen anyone try, but I would expect the PR blowback for "pay us $1 to appeal" to be worse than "you're just not allowed to appeal".
- johnnyanmac 3y agoThis is b2b, so it probably wouldn't really have that much PR. Besides, I think for modern society, paying $1 rather than presenting your government ID is an easier ask. There's a few easy checks to prevent perverse incentives: 1. promise (yeah, an impossible task already) a human will review your appeal 2. if the strike was a false positive, refund that appeal cost.
- pdimitar 3y agoThis is a solved problem for a long time now: require upfront payment, even if it's just $1.00. Many website owners have reported that this has killed 95% to 99% of all spam they were receiving. Genuine business wouldn't mind paying their $970 bill by first paying $1 and then paying $969.
- concordDance 3y agome looks at twitter blue bots