Ask HN: How do you deal with high-volume spam?
3 comments
I have no experience with this subject, but clicked through in order to think about how I might want to attack such a problem.
At a glance, it looks like dzone.com there doesn't even try as the new page is full of accounts which shovel in half a dozen or more links per minute. I'd think that some basic means of rate limiting would be a start, a base cooldown which grows if an account continues to bang into it.
From there, further limit and eventually blackhole new links from accounts whose previous submissions fail to gain traction.
I'd also think about how to score the trustworthiness and relationships of accounts to qualify submissions. I'd expect that spam accounts wouldn't spend much time promoting stories other than those of their own network. On the other hand, promotion by a trusty account could indicate that a link and the submitting account is less spammy.
At a glance, it looks like dzone.com there doesn't even try as the new page is full of accounts which shovel in half a dozen or more links per minute. I'd think that some basic means of rate limiting would be a start, a base cooldown which grows if an account continues to bang into it.
From there, further limit and eventually blackhole new links from accounts whose previous submissions fail to gain traction.
I'd also think about how to score the trustworthiness and relationships of accounts to qualify submissions. I'd expect that spam accounts wouldn't spend much time promoting stories other than those of their own network. On the other hand, promotion by a trusty account could indicate that a link and the submitting account is less spammy.
One thing I use on a much lower volume site is very simple and looks at the volume of posts people are making, on my site spammers post a lot more than regular people, that lets me automatically flag them and not show their posts to anyone but themselves / other spammers.
IMO, I don't mind the 'new links' section. Seems like spam will naturally filter out anyway, since it would probably not get any upvotes.
It is a very high trafficked site and is the target of a what seems like a million spammers, which is obvious given the entire 'new links' section is all garbage.
Their signup process requires a huge amount of information (home address, phone, etc), but that still wouldn't stop anyone from filling in bogus info..
They also moderate the first submission from new users, but that wouldn't stop someone from just submitting a legit first link, and then just submitting spam after their approved.
I just would really like to learn some of the REAL spam techniques that would help in this situation (a site people submit links to), that are a little more involved than just a captcha or some basic device.