Margen
For years, every time you typed those crooked words from a captcha you weren't just proving you were human: you were transcribing old books and newspapers, word by word.
The system was called reCAPTCHA and it was devised by Luis von Ahn at Carnegie Mellon University. The move was elegant: it showed you two words. One the system already knew, and it served to verify you. The other came from a scanned text that optical recognition software couldn't decipher. By typing it, millions of people were solving, without realising it, what the machines could not.
A study published in the journal Science in 2008 documented the invention: with this method, the historical archives of The New York Times and mountains of old books were transcribed. Billions of small anonymous favours, ten seconds each.
Source: Science (2008)