[O'ReillyNet:Comment Filter Problem] A Little Too Strong, Maybe?

by M. David Peterson

Sylvain just pinged me to say he had left a comment on my recent "IRC'aholic" post. I refreshed the page expecting to see his comment, but no luck.

I checked my email to see if the system had marked it as "must be approved".


An investigation into the comment section for that entry via the MT admin interface showed no sign of the comment. It wasn't until I accessed the master comment section and clicked the "Junk" tab that I was able to locate what he was referring to. In reading his post, there is nothing strange about it that would lead me to believe the system would justifiably assume this to be junk, so I'm a bit bewildered by what might be going on.

I have noticed what I would term a *SIGNIFICANT* drop off of comments over the last week. Like basically *ZERO* as opposed to what is usually a fair level of activity. With Sylvain's comment being labeled as "Junk" by the system this left me wondering how many other valid comments had been labeled junk and began to click through the listing of junk comments to see what I could find.

The problem? 10 pages worth of comments and I'm only about 6 or 7 hours back into the junk comment history, a history which contains about 24,000+ comments. Obviously, this could take a while.

Before I spend too much time rummaging through the "Junk" pile looking for lost treasure, has anybody else been noticing the same problem? Have you left a comment that provided an "Awaiting Moderator Approval" screen, never to see your comment approved? If yes, I would suggest leaving a comment to let me know such that the folks @ ORA can gain a better feel for what might be going on, but I'm not sure the message will get through.

As such, if you have noticed this problem, please let me know in private email such that I can provide this information directly to the folks @ ORA such that they can avoid having to dig through 24,000+ comments themselves looking for clues.

Thanks in advance for your help!

NOTE: If you could provide a searchable field such as the URL or email address you use this should make the task of locating valid comments quite a bit easier. Thanks!

Quick Update: I should note that this shouldn't be seen as a problem from the ORA side. Sarah Kim came to the rescue a few weeks back, implementing the Akismet comment spam filter plugin, which it seems probably just needs a bit of tweaking to the settings to find the right balance. This kind of thing is normal, expected behavior, *ESPECIALLY* when you are dealing with a variable that can be seen as "typeless" from the standpoint that the comment spammers are always trying new ways to get around the system, so attempting to determine what is legitimate and what is junk is a constant battle between good and evil.

Duck-typing, generics, and to a lesser degree, reflection and constructor overloads are fine when you have a finite set of data types you are dealing with. But when these same data types are changing on an hourly basis, attempting to make a determination in regards to what is what is a task that makes me ill just thinking about. With this in mind, any help you might be willing/able to provide will be of great value, without a doubt.