Screening Automatically Suspected Comment Spam
Short, concise description of the idea
Optionally, use an automatic filter to look for suspected spam. Post these comments screened so the journal owner/community maintainer can delete or unscreen as appropriate.
Full description of the idea
As automatic spam-detection generally does not endanger a service provider's legal status, use this in helping fight comment-spam too.
This would be an optional setting for folks who are getting enough spam to warrant the trouble; those who don't like the idea wouldn't have to use it.
When non-friends (or non-members, for communities) comment, optionally run the comment past a spam-filter to see if it gets tripped. If it gets tripped, post the comment screened initially.
Once the owner/maintainer sees the screened comment, they can make the call whether to send it on to the great spam-filter in
the sky the Abuse department, to just plain delete it, to leave it screened, or to unscreen it for public viewing.
I say "for non-friends" because presumably you have not actually added any spammers to your friends list. (Remember, unwanted contact from another party that you know is not spam, and if they keep contacting you it's time to serve bancakes, not time to irritate Abuse by wasting their time.)
- Less visible spam.
- Less search engine indexable spam.
- An interim step between 'allow comments from registered users' (unscreened) and 'screen comments from non-friends'.
- The filter on latest.bml is actually pretty good and is likely being trained in what's normal for LJ rather than the rest of the internet. (Major props to engineering and our friends at 6A.)
- Someone getting spam would still have to delete it.
- Any form of filtering can come off as creepy.
- False positives could alienate legitimate commenters.
- Some comments from spam users appear merely nonsensical and out of context until you look at their journal which is spam central and you realize that IT'S A TRAP.