When we noted shortcomings with Google’s own CAPTCHA security protection recently, several readers suggested the Big G should shift to using reCAPTCHA. Now Google has gone one step better and actually acquired the protect-your-system-against-machine-registrations provider.
The neat trick with reCAPTCHA is that it provides a two-word code to break, one of which is an unrecognised word from a printed text that’s been scanned in but which OCR software hasn’t confidently recognised. If you get the machine-generated word right, reCAPTCHA assumes you’ve probably also cracked the missing word. Unknown words matched by multiple users can then be added to the scanned text of books or newspapers. That’s a clever way of contributing to scanning efforts (and a logical step for Google given its ongoing Booksscanning efforts), but doesn’t necessarily mean that the resulting CAPTCHA is any easier to read.




















Peter Vo
Thursday, September 17, 2009 at 5:04 PMI never realised that typing in captchas contributed to improving OCR technology. Awesome! A nice example of crowdsourcing.
Craig
Thursday, September 17, 2009 at 5:47 PMMan, what won’t these guys buy! :) Still reCAPTCHA is an awesome little project and a good addition to there stable, that’s for sure.
Stephen Gore
Saturday, September 26, 2009 at 8:50 PMGo Google. ReCAPTCHA is a great project for sure.