PolyBandit: Polymorphic Linux Command Line Simulation
Open Access
- Author:
- Cromity, Steve
- Graduate Program:
- Informatics
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- March 19, 2021
- Committee Members:
- Nick Giacobe, Thesis Advisor/Co-Advisor
Edward J Glantz, Committee Member
Anna Cinzia Squicciarini, Committee Member
Michael Keith Hills, Special Signatory
Mary Beth Rosson, Program Head/Chair - Keywords:
- Polylab
Automatic Problem Generation
Cybersecurity Education
Education Technology - Abstract:
- This research discusses the cybersecurity landscape, shortcomings with traditional, static learning modules, as well as previous advances in cybersecurity instruction, and proposes a new Linux command line instructional simulation called PolyBandit. The system derived its namesake from the popular Bandit Linux wargame by OverTheWire, but appended to “Poly” due to the polymorphic spin the researchers put on the concept. PolyBandit comprises 100 levels that simulate real-world-based challenges requiring a range of fundamental Linux skills, including changing object permissions, directory navigation, and Bash script editing. This proposed system allows for easier assignment generation and creates unique experiences for each student helping limit the prevalence of cheating. It is widely scalable and can benefit Linux users across various skill levels. The researchers conducted a between-groups experiment wherein the researchers examined the use of two versions of PolyBandit, one built on a polymorphic paradigm (changing user-facing messages, directories, and text files according to the user) and one built to be static, similar to traditional lab assignments. Two groups of randomly assigned participants utilized either the polymorphic or static version of the simulation. This study focused on two main items, including the differences in engagement, retention, and performance between the versions as well as the prevalence of the Dunning Kruger effect. Our results did not imply differences in engagement, retention, or performance, but indicated the lack of a Dunning Kruger effect.