Another device for sound auditing using the Benaloh Challenge

E-voting systems often require this two conflicting properties:

• Cast-as-intended: The voting device should be trusted to accurately cast the voter’s intended choice.
• Receipt-freeness: The voting device should forget the randomness used during voting to prevent the voter from proving their vote to a coercer.

The Benaloh Challenge offers a theoretical solution to this problem, allowing voters to either cast a ballot or audit the voting machine.

flowchart TD
    VM{Benaloh Challenge} --> E1([Audit]) --> Audit["`
        Reveal randomness
    `"]
    Audit --> Audit2[The voter checks if the vote corresponds to what they intended]
    VM --> E2([Cast]) --> Cast["`
        Forget randomness
    `"]
    Cast --> Cast2[The encrypted vote is recorded]

In theory voters should audit randomly and regularely, ideally by selecting a candidate at random. However, in practice, few voters perform audits, and selecting a unwanted candidate feels counterintuitive. This makes this solution poorly practical.

To address these limitations, we propose a user-controlled audit device that interfaces with the voting device and follows a theoretically sound protocol for auditing.

The protocol could be as followed:

• The audit device audit the voting device N times (N chosen at random) with random candidates. Every audit should pass.
• The N + 1 time, the voter cast their ballot with the intended candidate.

This would give us a probabilistic proof for cast-as-intended, while receipt-freeness still holds.