Improve filtering of spam suggested edits

Question

I've been seeing an increasing number of spam edits in the review queues, replacing the entire body of the post with filler text such as a complement (to get past spam filters), then including a link formatted in BBCode (rather than Markdown). An example of such an edit can be found here and is as follows:

Hello, just wanted to say, I liked this article. It was helpful. Keep on posting! [url=(...)/cinture-burberry.html](...)/cinture-burberry.html[/url]

The URL in the above text points to a spam page on a compromised website. Can we make the filter smarter so that these edits are completely blocked?

Answer 1

This has been implemented. There's a check that runs before an edit is even accepted to see what percentage of the post has been changed. We tied this check into the spam system where we now catch bots that nobody ever saw at work, because the system was rejecting 98% of what they were submitting. This particular event now feeds the spam scores for the origins of the bot / human / monkey submitting these, and has made a serious dent in it.

We're also tracking posts that are oddly targeted for link placement. If an edit you suggest anonymously to a post that is a known spam target is rejected as spam, you're immediately dealt a blocking penalty in the system.

Not yet done cinching the corners, but we're keeping a lot more at bay than we were two months ago when the system was rolled out.