A Glossary of Voting Terminology

Transcription

1 A Glossary of Voting Terminology

2 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 2/27 Structure Terms from actual elections Requirements Attacks Cryptography Determining the winner Some academic systems of renown

3 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 3/27 Actual election terminology Voter credentials

4 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 4/27 Actual election terminology Ballot

5 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 5/27 Actual election terminology Ballot box

6 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 6/27 Actual election terminology Booth / Voting Booth / Pollbooth

7 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 7/27 Actual election terminology DRE = Direct Recording Electronic (voting machine) Diebold (USA) Nedap (NL)

8 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 8/27 Actual election terminology VVPAT = Voter Verified Paper Audit Trail

9 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 9/27 Actual election terminology HAVA = Help America Vote Act

10 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 10/27 Actual election terminology chain of custody

11 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 11/27 Requirements eligibility only individuals belonging to the group may vote. democracy only eligible voters may vote, and they may only vote once. accuracy - result depends on all cast votes, - result depends on nothing more than cast votes, - result depends on cast votes as they were cast. fairness no intermediate results.

12 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 12/27 Requirements - verifiability universal verifiability given the set of cast votes, anyone can verify that the announced result is correct. individual verifiability a voter can verify that her vote counts for the correct candidate. eligibility verifiability anyone can verify that the set of cast votes originates only from eligible voters.

13 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 13/27 Requirements - privacy anonymity no observer can learn how a voter voted. receipt-freeness the voter cannot prove how she voted. coercion-resistance (JCJ05) receipt-freeness + resistance to: - forced randomised voting, - forced abstention, - voting in the voter s stead.

14 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 14/27 Requirements - end-to-end End-to-end verifiability: cast-as-intended a voter can verify that her input to the process matches her intent. recorded-as-cast a voter can verify that the record of her vote matches what she gave as input. tallied-as-recorded anyone can verify that the announced result matches the public records of votes cast. counted-as-cast a voter can verify that her vote counts in favour of the candidate for whom she cast it.

15 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 15/27 Attacks ItalianLuxembourgian attack. chain voting.

16 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 16/27 Attacks gerrymandering.

17 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 17/27 Attacks Family voting.

18 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 18/27 Attacks chain voting.

19 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 19/27 Cryptography blind signatures: deblind(sign A (blind(msg,k))) = sign A (msg). homomorphic encryption: enc(msg a,k) enc(msg b,k) = enc(msg a msg b,k). - RSA78 - ElGamal85 - Paillier

20 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 20/27 Cryptography commitments. proofs: - (interactive) Zero Knowledge Proof (ZKP) - Designated Verifier Proofs (DVP) Fiat-Shamir heuristic: Make interactive proofs non-interactive.

21 Cryptography Plaintext Equivalence Test: enc(msg a,k) =? enc(msg b,k). Plaintext Inequivalence Test: enc(msg a,k) <? enc(msg b,k). SecVote 2010, 3 sep 2010 Hugo Jonker - p. 21/27

22 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 22/27 x Cryptography Mixnets w x y z x y 3 Mix 1 Mix 2 Mix 3 3 w 3 y 3 x 3 w 3 w y z z z adapted from [HS00]

23 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 23/27 Cryptography Randomized Partial Auditing / Checking [JJR02]

24 How to fill in the ballot / determining the winner Plurality voting (single winner) FPTP = First Past The Post winner = candidate with most votes. Instant Runoff / Alternative Vote Approval voting Range voting Condorcet Winner = pairwise most preferred candidate. Borda count rank candidates, most preferred wins. SecVote 2010, 3 sep 2010 Hugo Jonker - p. 24/27

25 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 25/27 Determining the winner Arrow s Theorem No system such that: - if every voter prefers A to B, then the group prefers A to B. - if no voter s preference between A and B is changed if C is added, then the group s preference between A and B also remains unchanged. - no single voter can determine the group s preference.

26 Some influential systems Cha81 theoretical BT94 receipt-free JCJ05 coercion-resistant practical Cha04 end-to-end verifiable time Theoretical: Chaum81 FOO92 CFSY96 CGS97 - Helios RF / CR: BT94 SK95 HS00 JCJ05 - Civitas End-to-end: Chaum04 Prêt à Voter Punchscan Scantegrity (I, II) Code Voting SecVote 2010, 3 sep 2010 Hugo Jonker - p. 26/27

27 SecVote 2010, 3 sep 2010 Hugo Jonker - p. 27/27 Done! Thanks for your attention!