Plugin idea: emailrep.io
Emailrep.io is currently a free service that evaluates the reputation of an email address.
I'm thinking the plugin could do a few things:
- High reputation: auto registration plus bonus (configurable) reputation points (e.g. skip some of those minimum thresholds)
- Medium reputation: auto registration but no bonus points
- Risky reputation: goes to the registration queue for manual processing
Could be useful. But... one must remember that whenever anything is automated via AI that one is also opening up potential for that AI to be gamed.
I would be opposed to any automatic reputation points. Reputation should be earned via feedback from the community at large, wh/imo is generally the best vetting.
I've not done much with the "Stop Forum Spam" nor "Honeypot" plugins. What would emailrep.io bring to the party?
Just take a look at it. Throw a few emails in the search box and see what you get. Sure it could be gamed but the attacker would have to know about it and the spammers I've been seeing are going for quantity not quality. They're not going to set up LinkedIn, Facebook, Twitter, Gravatar, and other accounts. It even checks account breaches as a data point.
Definitely makes sense to integrate this as an additional verification vector in the spam-be-gone plugin.
Thoughts @akhoury ?
Our spam problem is only 5-7 spam accounts per week so I find that Spam filters have high recall but low precision; meaning a lot of false positives.
Does spam-be-gone put accounts in a moderation queue if they fail registration?
@djensen47 I did take a quick gander - looks like it drags in a bunch of Google depends. Not everybody is keen on the Goog these days.
Be that as it may; You did not actually answer my question. I don't have a big issue w/forum spam and the stuff I do have enabled seems to be coping. Knock on wood. The few addresses I did check were flagged by each, the bad was missed by each, although emailrep.io seems to be a fair bit more aggressive in giving somewhat dubious reps to known clean addresses. Hence my initial impression based on an admittedly very short look does not leave me particularly enthused as I'd prefer to risk false negatives than false positives.
So, I would be curious to know just what "niche" this targets. Not that having more options is not a good thing. I would be curious if you have done any compare/contrast evaluation on your end cuz I am open to new data. And if it's cool mostly because it's new and offers promise then that is okay too.
Edit: Heh, now I neglected your question and that you need to ask it pretty much answers mine about the testing on your end. I'll have to refresh my memory on that and get back to you, else probably some staff has it off the top of their head sooner...
@julian If so, please give me a knob for opting out of the additional stuff. Which seems to be nodebb's approach in general w.r.t. stuff like this and just dandy by me.
djensen47 last edited by
Which seems to be nodebb's approach in general w.r.t. stuff like this
In general, yes. We try to not muck around with existing installs too much, but plugins are a little more fast and loose with things like this. Will keep it in mind