Account breaches have significant negative impacts at the individual, organization, and nation-state levels. For example, an account breach caused the recent Colonial Pipeline incident, costing the company millions and leading to fuel shortages and panic across the eastern seaboard. It is well known that a key factor in breaches is weak and reused passwords1. Many authentication systems have been proposed to improve the security of passwords1. However, most of these systems never see real-world usage, and uptake is far from universal when they do. The long-term goal of our research in this area is to (i) identify the impediments to adoption and correct usage of authentication systems and (ii) to design and demonstrate the ability of systems designs to address these impediments, helping users to leverage robust authentication systems to protect their online accounts.
Understanding the Adoption Process
There is a critical knowledge gap regarding the process by which users adopt authentication systems, preventing system designs from addressing this process and thereby decreasing the likelihood of supplanting traditional password-based authentication. To address this gap, our overall object is to model the adoption process for authentication systems, enabling system designs that address users’ needs throughout this process. The rationale for this research is that designing authentication systems to support the adoption process will improve the security and usability of those systems and increase the likelihood that they can supplant passwords.
We began our investigation of this area by conducting a laboratory study investigating the setup process for YubiKey and a empirical study observing users in their first two weeks using a YubiKey2. These studies revealed significant problems with the setup process for hardware security tokens. However, these studies also demonstrated that users generally enjoyed day-to-day usage. We have also conducted observational interviews of password manager users, having them demonstrate and explain how they configure their password manager, create accounts (including credential selection/generation), log into accounts, and update accounts. Results from this study strongly suggest a multi-phase process by which users had adopted their manager, with phases including setup, acclimation, gradual replacement of existing passwords with generated passwords, and steady-state usage.
These studies demonstrated a need for longitudinal studies to flesh out the adoption process model for authentication systems. To this end, we plan to conduct large-scale (200+~participant), long-term (1-year), longitudinal studies, observing users as they adopt an authentication tool into their daily lives. Initially, we will conduct studies for hardware security tokens (e.g., a YubiKey) and password managers, using the collected quantitative and qualitative data to model the adoption process for each type of tool. By comparing these models, we will derive a standard model describing the adoption process of authentication systems or demonstrate that it is not possible to create a unified model. Finally, we will design, prototype, and evaluate adoption process-aware systems, measuring their ability to improve security, utility, and usability during adoption and day-to-day usage. This research plan is creative and original in that it breaks from the status quo of treating adoption as a binary process—users do or do not adopt the tool—instead modeling it as a process consisting of distinct phases, each with their respective behaviors and transitions.
Exploring Security and Usability Designs
Even when adopted, security-critical functionality of new authentication systems is underutilized3, leaving users vulnerable. To address this issue, our overall objective is to identify design principles that can improve the security, utility, and usability of authentication systems. Moreover, we will quantify and explain the ability of competing system designs to encourage correct usage of these systems. The rationale for this research is that it will increase the understanding of underlying issues, identify generalizable design principles, and promote the design of more usable and secure password managers.
We have already evaluated the security of over twenty password managers on desktop4 and mobile systems5, examining the full password manager lifecycle—password generation, storage, and autofill. Our results identified numerous security issues with these browsers. One of the most damaging and persistent is the ability of malicious apps, websites, and browser extensions to steal credentials as they are autofilled. To address this issue, we are currently exploring how to build a trusted pathway into the browser, allowing for credentials to be safe from theft. We plan to extend this research, exploring other ways browser-supported authentication can improve the security and usability of authentication systems.6
We have also conducted usability studies of various web authentication systems7, including an in-depth systematization of the use cases design paradigms for password managers.8 Based on this research, we have identified the following research plan to increase the utility, utilization, and usability of password managers. First, we will quantify the ability of entry-aware password generation—generating passwords that are easy to enter on devices where the password manager is unavailable—to increase users’ willingness to use generated passwords. Second, we will quantify the extent to which prioritizing suggestions from password health checks improves users’ compliance with those suggested actions. Third, we will describe the unique processes, needs, and challenges faced by parents and children using password managers and quantify the extent to which system designs addressing this reality improve usability and security. This research plan is creative and original in that it breaks from the status quo of high-level usability assessments, instead quantifying and explaining the ability of system designs to address specific usability issues and improve the utilization of security features. Furthermore, it will be the first research investigating password health checks and multi-user password manager usage.
-
Verizon. 2021. Data breach investigations report. https://www.verizon.com/business/resources/reports/dbir/. ↩ ↩2
-
Reynolds et al. 2018. A tale of two studies: The best and worst of YubiKey usability. In Proceedings of the 38th IEEE Symposium on Security and Privacy. IEEE. ↩
-
Lyastani et al. 2018. Better Managed Than Memorized? Studying the Impact of Managers on Password Strength and Reuse. In Proceedings of the 28th USENIX Security Symposium. USENIX. ↩
-
Oesch and Ruoti. 2020. That was then, this is now: A security evaluation of password generation, storage, and autofill in browser-based password managers. In Proceedings of the 30th USENIX Security Symposium. USENIX. ↩
-
Oesch et al. 2021. The emperor’s new autofill framework: A security analysis of autofill on iOS and Android. In Proceedings of the 37th Annual Computer Security Applications Conference. ACM. ↩
-
Ruoti and Seamons. 2017. End-to-end passwords. In Proceedings of the 20th New Security Paradigms Workshop. ACM. ↩
-
Ruoti et al. 2015. Authentication melee: A usability analysis of seven web authentication systems. In Proceedings of the 24th International Conference on World Wide Web. Internet Society. ↩
-
Simmons et al. 2021. Systematization of password manager use cases and design paradigms. In Proceedings of the 37th Annual Computer Security Applications Conference. ACM. ↩