Analyzing Accountability in (E-)Exams

Many companies and universities increasingly use electronic exam (e-exam for short) systems: for example Cisco’s CCNA certifications or language tests such as IELTS or TOEFL are organized using computer systems, and also at ETH Zurich electronic exams are held regularly [8]. This naturally raises the question of the security of such systems. From the point of view of the organizing institution, cheating is a major concern, but candidates (i.e., students) are typically more concerned with fairness and transparency of the process. In particular the latter concerns are often ignored by current systems. Typically the system is assumed to be trusted, and the candidates can only partly verify whether everything went well. Yet, as recent scandals have shown [2, 5], (e-)exam systems cannot always be trusted entirely. In security protocol design the notions of Verifiability [1, 7, 3], Accountability [7] and Auditability [6] have been known for some time. A verifiable protocol protocol allows the participants to verify, after the protocol’s execution, whether the outcome of the protocol is correct. When verifiability fails, accountability additionally allows to identify which party is responsible for this failure. Finally a protocol is auditable if it logs sufficient evidence to convince an outside judge of the absence of errors. Some preliminary work on verifiability in e-exams exist [4], however accountability and auditability have not yet been studied in this context.