Using CAPTCHAs to do something useful
The purpose of CAPTCHAs is to prevent automated use of resources that are supposed to be attached to humans. They provide a test that is relatively easy for a human to perform, but expensive for a computer to perform. You can see an example on this blog when you attempt to comment on a post.
The folks at Carnegie Mellon University have come up with reCAPTCHA which provides a free CAPTCHA service that uses people’s CAPTCHA solving abilities to parse digitized text scanned books. The process takes two words (one known to the service, and one unknown to the service) and transforms them to a CAPTCHA-esque form. The CAPTCHA test requires the human to get the known word right, and uses the response to the unknown word to help the parse the scanned text.
It seems like an interesting and innocuous use of human computing power for a reasonable end. So some time in the coming decade, you may see these CAPTCHAs on this site.
As a side note: since one of the words in the CAPTCHA test is unknown, it looks like it’s possible to skew the scan results. I tried a couple of the reCAPTCHA examples and noted that it was possible to guess which word was unknown and provide a bogus value instead of the proper reading. Since it is still necessary to get one of the words right, the value of the CAPTCHA doesn’t seem to be diminished. The folks at CMU may want to implement a voting scheme to ensure that their results aren’t polluted by idiots like me.
Hat tip: Genome.
