Hacker News new | past | comments | ask | show | jobs | submit login

Well the text is HTML entities to escape ASCII text. "People being end with a person. Like everything in OUNASS ProMo CODE Onas OuNaS oNass cOuPOn DiscOUnt NoON SiVvl non toyou NaMshi"

Here Sivvi, toyou, Namshi, noon, and OUNASS are all brands of shopping websites and you can see their logos in the image.

Clearly this is some sort of keyword spam, though it's hard to tell more than that from your screenshot. It's also not clear why they'd bother to use HTML entities... a bug in the spam code? Or perhaps exploiting some parser differential between different twitter systems? Who can say.




So this is pure speculation, but more people should be aware of parser differentials (same thing as that email thing the other day) so let me say what I mean...

Hypothetically say a website has an internal service to index posts for keywords for search, that just so happens to unescape HTML entities during keyword normalization due to a seemingly harmless bug.

Plus a second internal service to identify keyword spam that _doesn't_ do any HTML entity unescaping (because why would you?)

Then you could end up in a situation where a spammer uses HTML entities to avoid spam detection while still showing up in search results. They hope that the user ignores the nonsense text and just clicks their link based on the image (a list of big shopping brands in the middle east) instead.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: