Verification: checking that something is what it claims to be

Verification is checking that something is what it claims to be. In elections the word covers several distinct checks: a voter confirming that a ballot records their choice, the system confirming that a ballot was recorded, and the public confirming that ballots were counted as recorded. Each protects against a different failure, so My Vote Forward always says which kind of verification it means.

Definition

The site glossary defines verification as checking that something is what it claims to be, and notes that in elections it covers distinct checks that must each be named (see the short entry in the glossary). Election standards break it down precisely. The U.S. Election Assistance Commission lists three properties that end-to-end verifiable protocols must support:⁠¹

• Cast as intended: voters can “confirm the voting system correctly interpreted their ballot selections while in the polling place.”
• Recorded as cast: voters can “verify that their cast ballots were accurately recorded by the voting system.”
• Tallied as recorded: a “publicly verifiable tabulation process from the public records of encoded ballots.”

The Voluntary Voting System Guidelines 2.0 (VVSG 2.0) define “voter verifiable” as a feature that “provides the voter an opportunity to verify that their contest selections are being recorded correctly before the ballot is cast”.⁠²

Origin and history

Verification in elections has moved toward checking evidence rather than trusting equipment. Electronic voting raised a specific problem, which Lindeman and Stark state plainly: “computer software cannot be guaranteed to be perfect or secure”.⁠³ The response was the principle of software independence. Lindeman and Stark state it as “an undetected change or error in voting system software should be incapable of causing an undetectable change or error in an election outcome,” citing Rivest and Wack (2006) and Rivest (2008).⁠³ VVSG 2.0 made software independence mandatory and recognized two ways to achieve it: “independent voter-verifiable paper records” and “cryptographic E2E verifiable voting systems”.⁠²

How verification is done

Different checks answer different questions.

• Voter verification at casting. Federal law requires that a voting system let the voter “verify (in a private and independent manner) the votes selected by the voter” before the ballot is cast.⁠⁴ This catches a misread or mismarked ballot, but only if voters actually check.
• Post-election audits. Officials compare a sample of paper records with machine counts. A risk-limiting audit gives statistical evidence that the reported winner is correct.
• Cryptographic verification. In end-to-end verifiable voting, voters receive a receipt that lets them check inclusion without revealing their vote.⁠²
• Compliance checks. Lindeman and Stark stress that if the procedures protecting the records cannot be shown to have been followed, an audit “may be mere theater”.⁠³ See Chain of custody.

A system can pass one check and fail another. Accurate counting of altered paper is still a wrong result; a perfectly verified individual ballot says nothing about whether ineligible ballots were added.

Common confusions and neighboring terms

• Verification vs. authentication. Authentication checks that a user controls an account; election verification checks ballots and counts.
• Verification vs. identity verification. Identity verification is one step of identity proofing, unrelated to checking ballots.
• Verification vs. audit. An audit is an organized form of verification by independent examiners.
• Voter-verified vs. voter-verifiable. A record is verifiable if the voter had the chance to check it; it is verified only if the voter did.

How My Vote Forward uses the term

My Vote Forward never uses “verified” without saying what was checked and by whom. Its Audits and verification chapter explains how voter-verified records, preserved custody, risk-limiting audits, and approved end-to-end protocols work together while keeping ballots secret. For its own published releases, the framework applies the same idea: a release states its method and denominator so others can check it. See Published release.

Unresolved questions

• How many voters actually check their ballots or receipts, and does a check that few people perform provide enough protection?
• How should verification results be explained to the public so that a passed check is not overstated?
• Can a participation system’s published aggregates be made independently checkable without exposing individual answers?