Ask HN: How do you programatically deal with forum/blog trolls?

1 points by typicalrunt ↗ HN
Trolling is the age-old annoyance on the Internet, but the latest feature request I received for a forum that I am the technical lead on has me scratching my head.

Scenario: users are creating new accounts en masse just to troll other users. They post inappropriate text/links/pics that disturb other users.

Proposed Solution: the admins/mods would like to see general users* unable to post any links/pics. They can still create threads and post replies, but that data is just going to be text. In order for them to elevate their privileges, they must show some entitlement (i.e.: purchases on their account).

My take: This smacks of elitism because I need to purchase something before I have privilege, but from a technological perspective it also ties me to another service that I don't control. Accounts may be easy to create, and entitlements less so, but I don't think that will stop trolling (possibly just the image trolling).

So what tricks have you used to stop trolling on forums/blogs/etc.?

Edit: I neglected to mention that this forum runs 5,000-10,000 concurrent users throughout the day and over 3k new accounts per day, making any sort of manual moderation of users difficult.

2 comments

[ 3.9 ms ] story [ 18.6 ms ] thread
- "Cave" feature, where the troll can post as many messages as they want, but those messages aren't seen by anyone other than the troll - Duplicate account detection - if the same IP address appears on more than one account, the account is auto-flagged
Accounts can look like they're coming from the same IP address all the time. If people access the website from inside a NAT, they'll have the same IP.

Cave is like a hell-ban, and it works when there aren't too many users on the system. The forum I work on peaks at 10k concurrent users with over a million accounts. It would make for an administrative nightmare (which is probably why the feature request tries to automate a solution).