Google’s research into Socially-Adjusted CAPTCHAs

CAPTCHAs (Completely Automated Public Turing test to tell Computers and Humans Apart) attempt to distinguish humans from computers by presenting tests are are simple for humans but complex for software to solve.

Examples of text-based CAPTCHAs include distorted words or phrases, simple arithmetic (e.g. “10+3=?”), answers to simple questions (e.g. “What is the capital city of England?”), and questions based on images (e.g. “How many dogs do you see?”).

These systems can be tedious to populate with tests and certainly aren’t foolproof – Ars Technica reported in early 2008 that one in five requests originating from bots against www.gmail.com’s text-based CAPTCHAs succeeded.

Google recently released research on “Socially Adjusted CAPTCHAs“, circular images rotated by a random angle that users need to orient correctly to pass (multiple results from testers calibrate a proper angle for the image to be acceptable, hence “socially adjusted“).

This CAPTCHA implementation offers major advantages over other image-based systems in that these images can easily and automatically be generated by software, but not easily solved.

You can read Google Research’s post on the subject here or download the original research paper here.

Related posts:

1. Google Chrome Frame changes Internet Explorer into Google Chrome!
2. Uninstall Google Software Update on Mac OS X
3. Geekology has 10 Google Wave invites to give away!
4. Review of the Google Wave Limited Preview
5. Advanced Google search hacks and tricks