Handling trolls with multiple accounts over VPNs

If you feel it’s so simple, feel free to propose it as a plugin in marketplace with a budget.

1 Like

Could pull some estimate out of a hat, that how complex this kind of implementation would be? I am not even sure if cookie tagging is the right approach, there might something more clever, but it is a topic worth discussing as IP logs are nowadays white noise and disposable emails are available for free (for exp. Outlook.com offers email aliases).

Moderation is a very time consuming part of community management. Improving the available tools would actually be something of real value. Maybe we could crowd fund it as a community, if there is interest. As a non-profit our financial resources are very limited, but with allies we could pull something off.

1 Like

see Build a browser fingerprinting plugin

1 Like

A neat feature would be to require SMS verification on registration. So one phone number per account. A new mobile phone number is way harder to get than another mailbox.

Not really. There once was a time that I created digital ocean and mailgun accounts for my clients and used some app to generate phone numbers to recieve a text. It was tedious, but neither difficult nor expensive.

6 Likes

Tedious would probably put some sort of a speedbump in front of your “bored asshole” types, at least. Can’t imagine phone number verification would ever go into core, though, too much variance in service provider, et al. Probably wouldn’t be hard to build a plugin for it, though, if anyone was feeling frisky.

6 Likes
  • Enable must approve users in settings for all new users. Reject all suspicious sign up that comes from VPN, temp-email or anonymous name.

  • Require restrict approval for all new users.

  • Whitelist certain e-mail providers to avoid chances of using temp-emails.

  • Blacklist those e-mail from which troll is generally coming.

  • Set min trust value for message

image

  • Require approval/Block based on certain offending keywords

  • Blacklist offensive username.

  • Block VPN server IP by AS Num. For example, in screenshot I have shown DO.

  • Ask for legal Identity and Address proof as image URL hosted by them in Sign up form under text field. :fire:

4 Likes

I’ll echo this being something that would be useful in trying to detect, and deal with, banned members coming back as new accounts.

It would be very helpful if the system would flag (simply for attention) mods when/if any account logs in with an IP address that is the same as any other account. A match not just for Registration and Last IP, but for any IP that’s ever been recorded.

Could have whitelist to allow admins to take out IP’s that would result in too many false positives.

But, otherwise, having this sort of automatic notification of common IP address for all IP’s ever recorded for all accounts, would be very helpful in letting mods know which accounts to follow up with special attention.

Very little can be done when a troll is not only motivated enough to keep coming back, but is fastidious enough to cover his tracks meticulously.

But trolls seem to pretty much always slip up at some point.

Currently, it requires mods manually checking Last IP frequently and being lucky enough to catch the IP address commonality in real-time. We catch people like this quite a bit, but I imagine that it’s just the tip of the iceberg.

5 Likes

So we’re currently dealing with a problem user who creates a fake account, then gets blocked, then changes their IP with a VPN, creates a new email and then creates a new account.

I have a different suggestion, which I’m not sure is feasible:
A setting that blocks ‘anonymous’ IP address. IP’s from VPN services for example.

Discourse currently uses MaxMindDB to do IP lookups and I see that MaxMindDB has an anonymous ip database. I honestly don’t know anything about the MaxMindDB integration though.

If this were possible it would at least stop users from creating multiple accounts using a VPN to mask their IP. Should cut down trolling a lot.

4 Likes

Ohhh interesting. As a first step can we show whether or not Maxmind tagged that IP as anonymous here @nbianca? Could you take 30 minutes (when you’re back, of course) and see if it’s even possible?

However @RobinTS if you mean a different or additional maxmind database, it’s unlikely we would pull that in. I was wondering if “anonymous” is a designation in the existing database, like the geolocation.

9 Likes

This linked database is paid and “contact us” pricing level, so not something we could ever ship.

12 Likes

Ah that’s a pity. I’ve also seen some APIs out there that you can use to check for pretty much the same thing. I’ve been using this one to manually look up the IP addresses of new users. I’m sure there are others out there that may be better. Sounds more like plugin territory now though.

3 Likes

It may have to be noted that some legitimate users use VPNs (I don’t know exactly the proportion, but I do know I am usually one of them). Totally blocking these IPs would have side effects for them :frowning:

4 Likes

We did a small project a while back in php where we determined for market research purposes where if the verbatim comments where ‘positive’ or ‘negative’. It worked a treat for our custom Net Promoter Score (NPS) results.

Would there be a way to profile the new user from the language they use? Something like … to flag users who have a degree of the same language as specific silenced users.

I personally feel this would be quite viable, provided the user types enough content over a reasonable amount of time. If they only post terse sentences, it’d be tougher.

(Note that I specifically mean “match one unknown user to an existing user based on the words they use, how often and in which order they use them”)

3 Likes

The perspective API plugin will already analyze and store a score for posts, and a data explorer query can give you the average score for users and filter only the recent ones.

If someone tries this and proves it successful we can look into automating it on the plugin.

6 Likes

As @Falco said, that is a paid solution. Instead one could use these IP lists to ban VPN users (or we could add that one to our core to identify anonymous IPs).

6 Likes

Today, I logged out of Meta and wanted to log in again (I had initially written my password on a piece of paper. I now put it somewhere, and I just wanted to make sure I had correctly entered it by doing a copy/paste log in test). Anyway, I got a message “You cannot log in as Mevo from that IP Address” (the one I used for weeks to post, and just logged out with :wink: ). It is through a VPN. So are some VPN addresses blocked now ? Is it Discourse as a whole, or only here on meta ?

Your IP is blocked for some reason, probably because we had problems with a user at that IP before.

5 Likes

Ok, but it’s a VPN used by a LOT of people, so you’re also blocking all potential people wanting to connect using that VPN. It’s interesting to understand with that little example how you tend to run into some problems like this one, using VPNs (1 of the user does something, and the IP shared by a bunch of people gets banned from services :wink: )