Hacker News Reader

Comment by hkchad | Hacker News Reader

https://cointelegraph.com/news/fake-spacex-youtube-channels-...

The first one I saw was a Rip of a SpaceX livestream. I watched it for a bit then noticed all the BitCoin references and got confused then realized it was a scam account. How they get promoted basically to the front page is the issue.

View on HN

A solution as simple as just hiring someone for $10/hour to periodically scan youtube for crypto keywords and disable the scam videos and accounts would have prevented millions of dollars of theft.

I think this is a good solution, but maybe not viable. Going past Crypto this could apply to so many issues needed to be reviewed on YouTube. But at that scale, the human cost is something YouTube refuses to pay.

it would be a stopgap measure until the algos are refined to filter out the videos without the need for human intervention. YouTube is being sued by Brad Garlinghouse for not taking action fast enough. My guess is my $10/hour solution is cheaper than $1000/hour lawyers.

Those lawyers already spent the money writing a ToS that says they're not liable for shit.

What are examples of keywords that would have a low false positive (and ideally, low false negative) rate, such that a minimum wage worker could efficiently disable videos with little context or time?

it is not the keywords but the content of the videos. anything that involves a livestream and a pitch to send X to get 2x-10x back. very easy to train someone to identify what these scam look like with zero low false positive rate

Why would you need a human to do a job that a computer can do 100x faster?

> "What are examples of keywords that would have a low false positive"

"Bitcoin"

Reply on news.ycombinator.com