They have compiled a great list of nutty captchas caught in the wild. The best is certainly the Russian one, as many others have agreed. Captchas are particularly interesting to me since it looks like I’m going to be working with Dr. Luis von Ahn next semester on the GWAP project (games with a purpose).
Luis was one of the inventors of the CAPTCHA and is currently working on reCAPTCHA, which not only prevents bots from accessing protected pages, but solves the problem of difficult OCR scanning. The idea is that if the state-of-the-art OCR cannot read text, no bot will be able to decipher the text either. Two words are presented to the user, one which is known and one which is unknown. As long as the user gets the known word correct, they pass the reCAPTCHA. The unknown word is then given to other users until a high degree of confidence for that word has been established. This is human computation at work to improve the process of digitizing books.
