Google finally demotes fraudulent, official-looking site after 8 years
8 replies, posted
https://www.bbc.co.uk/news/technology-46316655
Google says it will stop ads for expensive unofficial Esta services appearing at the top of search results, eight years after the first complaints.
In 2010, the US started charging UK travellers to use the Electronic System for Travel Authorisation (Esta).
Unofficial sites charging five times as much as the US government soon flooded the top of Google's search results, despite breaking Google's ad rules.
Why the fuck did this take so long
This is in the article. It wasn't just one fake ESTA site, it was a lot of them. Before Google finally decided to let their machine learning take care of it, they would just take down whatever ad was reported, then more of the same kind would pop up.
I guess a better question is why does Google not have people authenticating ads on their services to make sure they don't, you know, break the law
It's Google. It's automation first, automation second, automation third... You get the point, until all of them fail so hard and bad they're forced to send a human in to clean up the mess.
I'd assume a similar reason to why Youtube employs algorithms to assist in content moderation; the scale of the issue is just far too large. If they were to genuinely hire enough people to actually authenticate these things, the only way they could make it viably affordable is by paying dirt cheap wages to unskilled people which would ultimately result in mistakes being made regardless
Makes sense, using employees for something that's ramping in scale as quickly as their service is is literally completely impossible.
That's maybe partially true, yeah. Still, they should have switched to machine-learning more quickly. I mean, if it's the case that humans can't analyse stuff, could ISIS put up an advert supporting terrorism? I just think it's not a great idea to not have somebody analyse certain things, with perhaps machine-learning to determine what things should be looked at by people. It wouldn't be perfect, but it'd be better.
This isn't even difficult because it was a legitimate US government URL, unless the website matched that it should've just autoblocked them. This didn't even need Machine Learning or Automation.
Sorry, you need to Log In to post a reply to this thread.