Measuring Identity Confusion with Uniform Resource Locators

Uniform Resource Locators (URLs) unambiguously specify host identity on the web. URLs are syntactically complex, and although software can accurately parse identity from URLs, users are frequently exposed to URLs and expected to do the same. Unfortunately, incorrect assessment of identity from a URL can expose users to attacks, such as typosquatting and phishing. Our work studies how well users can correctly determine the host identity of real URLs from common services and obfuscated "look-alike" URLs. We observe that participants employ a wide range of URL parsing strategies, and can identify real URLs 93% of time. However, only 40% of obfuscated URLs were identified correctly. These mistakes highlighted several ways in which URLs were confusing to users and why their existing URL parsing strategies fall short. We conclude with future research directions for reliably conveying website identity to users.