June 5, 2007

CAPTCHAs are those squiggly words you sometimes see on web sites that you are asked to read and type in to prove you are a human being. CAPTCHAs only work as far as a computer cannot reasonably be used to identify the word automatically, so CAPTCHA software and CAPTCHA cracking software are typically in an arms-race situation.

reCAPTCHA adds a brilliant twist: The images of squiggly words are actually scans of real words from real printed pages that have been run through a computer text recognition program and could not be identified successfully by the computer. By typing in the word, you demonstrate you are a human and you help the scanning effort correct the word.

Of course, to serve as a CAPTCHA, the answer has to be known by the system, so what you actually see is two words, one of which is known. Your answer for the unrecognized word is used to build confidence until the word is considered recognized based on multiple answers.

(Thanks Matt.)

post a comment...