Foreign spammer problem documented
 

This is a graph of new user accounts. A few weeks ago we changed the settings of the CAPTCHA which people have to fill out on registration.

http://cellar.org/2008/newusr.png

Clearly, the CAPTCHA had been broken and automated bots were used to create new user accounts.

And clearly, too, at some point the new CAPTCHA will be broken, and the one after that...

They want to love us long time.

It looks like you changed the Captcha in the middle of Week 19.

So did you personally come up with new settings, or did vBulletin release a patch of some sort, or what?

I wonder what caused the drop in Week 12. Chinese New Year?

The settings were always in there, but I wasn't aware that they'd make this much of a difference. Once the spam problem became intolerable I started looking into what people were doing to prevent it.

Started thinking about captchas, all "build a better mousetrap." Except I don't want the sort of people who want a better captcha at my door.

Came across reCAPTCHA, where CMU gives you two words (one known, one unknown), and in turn uses you to fill the gaps in the OCR software they have processing old books.

I feel as though it might be 'nifty social work' at the cost of 'extreme security.' But it's free.

Wow. What a brilliant idea. That's really cool, Skunks.