Friday, January 15, 2010

Re-Captcha-ring Old Texts One Word at a Time

Frustrated by typing captchas--those distorted, blotched, wiggly, struck-through words to prove you are a human user of a website?  They do help keep sites from spammers and their automated emails.

And I've noticed I'm recognizing real words more than I used to.  Perhaps this has something to do with a hidden task at our fingertips that we are performing without knowing it.  The new captchas--called re-captchas--are actually words from archival texts that computers have difficulty transcribing digitally.

At a rate of 20 million-a-day, according to ScienCentral News, Internet users are solving the mysteries by picking out the letters from noisy speckles, blotches, and lines.  Captcha inventor, Luis von Ahn of Carnegie Mellon University, here in Pittsburgh, notes in this video how much time per day is used by humanity while sorting out these digital keys, part of the motivation for turning this task into something doubly useful.

So where the ink is bleeding through, where the bookworm has had its lunch, where mold had left its mark, or air had darkened the page, re-captcha is sorting the wheat from the chaff with my help. It's nice to know that next time I work out a captcha riddle I may be doing my part to save an ancient text.

No comments: