Hacker News Reader

Comment by can16358p | Hacker News Reader

Same for Instagram and Twitter. The formats of the obvious spam is super simple to detect on both platforms for years.

Yet I haven't seen ANY progress on either platform. Assuming that they are much more competent than a random dude like me as a whole tech company, the only possible explanation is that they have incentive not to prevent spam.

Perhaps: spam = more notifications delivered = more app opens = more potential feed engagement since they've opened the app even if from a spam notification = more ads shown = more revenue = happier shareholders short time... all at the expense of losing long-term trust to the platforms.

View on HN

Have you been working on actually detecting them or are you assuming they’re easy to detect? Did you go digging for false negatives or false positives in other languages? Considering that not all special characters are printable, these platforms don’t limit their users to ASCII, random foreign words are common in other languages, the false positive rate has to be pretty much zero, etc etc etc I’ll bet it’s a whole lot more sophisticated than you imagine.

I'm assuming they're easy to detect yet I'm super confident about the patterns I'm seeing that I'd bet money that it can be implemented.

False positives should ideally be zero yet algorithms don't have to ban right away.

If there's a relatively small team of people reviewing those spams manuallu after the algorithms flag, I don't see any problems with the approach.

I of course get there are many different corner cases, Unicode characters that look like other letters etc. to bypass the filters yet they're also very easy to implement a filter that would probably detect many spams right away even with many of the corner cases. But again, a human team would still review these tweets anyway giving no room for false positives.

About 90% of the public Youtube spam I see on YT still follows this ultra-simple template: one person asks something on the "top comment", another one replies with a faux-legitimate message.

The question is often "does anyone know how to pirate videos?", "...how to invest in crypto", "...how to get fake instagram followers", life coaches, etc.

It's not really custom messages with special chars, it's something that I could detect by using Ctrl+F in the browser. It's not something they'd need special tools on their end to catch. The only thing that is changing is the name of the users.

So yes, I'm gonna agree with GP that "the formats of the obvious spam" are super easy to detect.

Reply on news.ycombinator.com