Admin Post - Blogger's Spam Filter Goes Berserk

 

Good morning everyone, and welcome to the final day of the Challenge. This is just a quick admin post to make you aware of something that is going on behind the scenes.

Some of you may have noticed that comments you have left have been disappearing and then reappearing. Up until a few days ago it wasn't anything too major - a couple of comments a day were getting caught up by Blogger's spam filter.

At the start of the week, I noticed that there was a marked daily increase in the size of the spam folder, but didn't see any recent spam. Looking through the folder, I discovered that more than the usual number of genuine comments had been mistaken as spam.

On Wednesday it started to go really berserk, and I must have recovered about 100 mis-categorised comments. If I thought that was bad, Thursday was nuts - Blogger marked 449 comments as spam - 18 were actually spam, the other 431 were genuine comments. Friday was a bit calmer, with just 154 comments marked (14 actual spam, 140 genuine comments).

If I thought Thursday was nuts, I really wasn't prepared for yesterday:

• 831 comments marked as spam
  
  
• 11 were actual spam
  
  
• 820 were genuine comments (which I recovered)

My guess is that someone has tweaked the filter to catch more spam, but didn't think to check if it was catching genuine comments by mistake. Imagine the scene:

SWAT team commander: "Ladies and gentlemen of the press - our operation successfully killed all 11 gunmen."

All the reporters: "That may well be true, but you also killed 820 innocent bystanders. Would you care to comment on that?"

SWAT team commander: "........"

Judging by the spam folder this morning, it looks like it isn't going to end any time soon...wish me luck!

Update #1 - having reviewed 33 messages earlier this morning (2 spam, 31 genuine) I've just checked and as of 1100 UTC there are 276 more to review.

Update #2 - I forgot to mention that whilst a lot of the comments being caught are "recent", the majority are much older. It seems to be short comments (one, two or three words) that it is incorrectly marking as "spam".

Update #3 - by the time I got to reviewing the "spam" comments, the number had gone up to 331. While I was reviewing them, it added another 33, taking the total for that batch to 364. Only two comments were actually spam, the other 362 were genuine comments and have been restored. However, just as I finished another 3 comments came into the folder - grrrrrrrrr!

Update #4 - it's not added any since about 6pm UTC. Today's total was 659 comments caught by the filter, 9 of which were spam, 650 of which were genuine comments.