Studying CAPTCHA Complexity with Eye Tracking

A CAPTCHA is a widely used security mechanism to prevent automated access to webpages by malicious users such as internet bots. Introducing elements such as distortion, distracting backgrounds and noise are some ways to make it harder for computer vision algorithms to break them. However, making CAPTCHAs complex by increasing the intensity of these properties has a direct adverse impact on their legibility to humans and hence usability. We conduct an eye tracking study on participants to understand how ease of comprehension of CAPTCHAs varies across increasing levels of complexity. We seek to find out where to draw the line in increasing complexity for CAPTCHA generation, in order to present ‘good’ CAPTCHAs on websites that are balanced in terms of their usability and security against internet bots. CR Categories: H.1.2 [User/Machine Systems]: Human Factors, H.5.2 [Information Interfaces and Presentation]: User Interfaces— Evaluation/Methodology