dreadedmonkeygod . net

Traceability

Wow. Some bonehead managed to get a bit of comment spam through my comment filter, which had so far been Chuck Norris to spam's McCaully Culkin.

Let's just take a moment to savor that image...

Anyway, they got one past the filter, which makes me wonder: was my anti-spam servce unreachable for some reason, or did the service actually think this post wasn't spam?

This brings me to an important point about automated processes: traceability. In the event of a problem, you want to know what decisions were made and why. I'm sure I wrote my error-catching code to detect a failed connection to the anti-spam service, and I think I had it give the comment the benefit of the doubt. I may have to reverse that. Either way, I'll be adding another field to my comment database: why it was marked as spam/ham. That way I can specially flag comments for manual review when the spam filter couldn't be reached. (Or, if I were smart, my system would try again whenever those comments were about to be displayed.)

Post a Comment

Name:
Email (Never, ever displayed.)
URL:
Remember me next time.
Comments (Sorry, no HTML allowed. Space paragraphs with a blank line.):