A few weeks ago it was reported that a botnet operator had cracked the Microsoft Windows Live Mail CAPTCHA system. CAPTCHAs are the pictures you see, typically of distorted words, when you sign up for accounts from various providers. Now it seems that this same botnet operator has successfully cracked about 20% of Google’s CAPTCHAs.
I would imagine that the technology and techniques that are being developed to crack CAPTHCA would have a direct application with OCR. If on the other hand it doesn’t, maybe companies like Google should take a look at their employee’s hand writing. They could then take the people with the worst lowest OCR success rate and use their hand writing instead of generating something with a computer.
Of course I’m no CAPTHA or OCR expert so I might be completely off base here. I think I’ll do some poking around on the net to see what I can find out.