Toons, your post has been cleared and I apologize for the delay and no doubt your frustration.
Most forums use some form of spam filter. We've been using Defensio in production for several months. All told, it's been decent, but not great. We have had some significant problems with it generating errors and frustration in getting support from Defensio technicians. One of the most visible drawbacks, at least from a new member's perspective, is that Defensiot experiences a large number of false positives. We are actively looking for alternative systems or means to improve this one. We recently conducted a study of about a dozen alternative spam filters, including the key-word filter EpicSki relied on when we were a vBulletin site. Defensio had the largest percentage of false positives. I found it interesting that the Key-work filter scored second-worst in this category. Members that submit posts only to get a message that they will have to wait until the post is approved, certainly feel frustrated if not offended.
False positive results are an important metric for consideration and we need to focus on minimizing them. At the same time, the system does get human oversight and we try to correct problems quickly. It is certainly less arbitrary than some systems which block IPs or email account registrations based on databases such as Stop Forum Spam, or Project Honeypot, which seemed to prevent some members from even logging in.
Spam is a significant problem in most forums, and the ability to defeat spammers and bots is a never-ending battle and results in inconvenience. We are continuing to try to make those improvements, and are actually working very hard on that project at this time. Keep in mind that the newest members are subject to the highest level of filter scrutiny. Most members that have posted here over several months never see the intercept.
With regard to your questions:
- *Are the fraudulent accounts being setup by individuals or computers?
Yes. we have over 2000 banned spam accounts on file. Some spam bots have hit us with over 250 posts or wiki entries in less than one hour. The damage can take considerable time to clean up if they penetrate the defenses.
- *Are the spam messages you are seeing being sent by individuals or computers?
- *Could you integrate CAPTCHA's somewhere? When joining? Posting? PMing?
CAPTCHA is a sub-routine included in filters like Mollum. Our study found it's designed on the assumption that most spam comes from bots, which is true of a fairly small percentage of spam posts on EpicSki. The vast majority of spam that blogs see is submitted by bots. However, this isn't true for sites on the Huddler platform. Almost all of the spam submitted to our sites is submitted by humans, with registered accounts. As such, CAPTCHA is not a useful tool for discerning spam from ham on our sites.
- *What stops members from forwarding offensive/spam PM's to a moderator?
We have a post reporting system, and it is effective is bringing spam and posting behavior that violates our site terms to our attention. We just don't think members should have to be the primary spam filter on the site.
- *Are moderators able to delete posts?
Absolutely. We also can move posts to unseen forums where we can discuss issues and even return the post to the forum should it be deemed appropriate.Edited by Cirquerider - 12/30/09 at 12:16am