Reading Distorted Type

Science 12 September 2008: Vol. 321. no. 5895, pp. 1465 - 1468.

In last week’s issue of Science magazine, Luis von Ahn and his colleagues write about CAPTCHAs, that is Completely Automated Public Turing tests to tell Computers and Humans Apart. An example is seen above, in which an image containing several distorted letters is presented to online users before they purchase tickets or join social networks. Each day, 100 million of these distorted words are decoded and retyped by you and me.

Their research explores whether this human intervention can be used to help such projects as Google books’ digitization of library collections. When the optical character recognition machines cannot decipher particular words, CAPTCHAs could be used to solve the distortion. Therefore, every time you order something online, Princeton and other libraries would benefit.

The complete text can be read at: