Infrastructure

We're currently using the StopForumSpam blacklist to bock IPs from making POST requests (eg, new posts/comments or sign up/sign in) at the HTTPd level.

While it did help reducing slightly the amount of spam it's not ideal, because:
- the block is IP-based not username-based, so one can't contribute through Tor or similar;
- the HTTPd doesn't distinguish between account creation, login, post submission, new comment.

A better solution would be to integrate the antispam solution with Askbot, maybe as follows:
- IP-based ban for account creation;
- run new posts and comments of low-karma users through the antispam (bayesian filter plus IP blacklist); if the spam score is high enough, send the post to moderation (or reject depending on configuration), otherwise post immediately.

If a bayesian filter is integrated with Askbot moderators will also need to be able to train it, for instance with a “Mark as spam” button next to each post.