Spam Defense/Filtering

[Arthmoor]

So in recent weeks my forum has had a wave of spammers who have been able to freely post blatantly obvious spam on the site. Word filtering is absolutely ineffective to preempt them, and in the most recent case, isn't stopping them at all despite the terms being added to the post filter. In the support thread I filed, it was made clear your current anti-spam system only deals with user accounts, NOT with the content of the things they're posting. I'd say that's proven to be more or less useless. What IPS needs is preemptive/proactive spam defense, not reactive flagging and content deletion.

To that end, I'd like to very strongly suggest that IPB set up a partnership with the Akismet anti-spam service that's widely used by Wordpress blogs and a number of other sites and applications. Their system works by analyzing the actual content being posted and will return a yes/no type of response as to whether they think it's spam or not. From my own experience, they're rarely wrong about that. I'm using it on a bug tracking site I wrote myself, along with an obsolete forum package I still keep online for an old MUD.

A properly implemented Akismet setup will flag spam and store it in a temporary place for moderator review. From there, the content can be checked. If it's not spam, the moderators can click a button to indicate this, and a false positive is sent to Akismet for the system to learn from it. If it is spam, you can either delete it on the spot or let it sit in the pool until a specified expiry time (on my other 2 sites, I use the suggested 30 days). If a post actually does manage to hit the forum, the moderators can click a button there to have it reported as spam and their system learns from that as well. So it's very adaptable. What's even cooler, is the content scanning is done in real time so there's no having things sitting in some queue for days before it's evaluated by their servers.

There's a great deal more useful information to be found at https://akismet.com/ such as developer support and API documentation etc. I'm sure they could explain the gory details a lot better than I can. In the end though, this would be a wildly more effective way of dealing with spam than what we currently have.