1 / 64

Manipulation and Control for Approval Voting and Other Voting Systems

Manipulation and Control for Approval Voting and Other Voting Systems. Jörg Rothe Oxford Meeting for COST Action IC1205 o n Computational Social Choice April 16, 2013. Introduction. S ocial C hoice T heory voting theory preference aggregation judgment aggregation

abie
Télécharger la présentation

Manipulation and Control for Approval Voting and Other Voting Systems

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Manipulation andControlforApprovalVotingandOther Voting Systems Jörg Rothe Oxford Meeting for COST Action IC1205 on ComputationalSocial Choice April 16, 2013

  2. Introduction SocialChoice Theory • votingtheory • preferenceaggregation • judgmentaggregation TheoreticalComputer Science • artificialintelligence • algorithm design • computationalcomplexitytheory - worst-case/average-casecomplexity - optimization, etc. • voting in multiagentsystems • multi-criteriadecisionmaking • metasearch, etc. ... Software agentscansystematicallyanalyzeelectionsto find optimal strategies

  3. Introduction SocialChoice Theory • votingtheory • preferenceaggregation • judgmentaggregation TheoreticalComputer Science • artificialintelligence • algorithm design • computationalcomplexitytheory - worst-case/average-casecomplexity - optimization, etc. • computationalbarrierstoprevent • manipulation • control • bribery • ComputationalSocialChoice Software agentscansystematicallyanalyzeelectionsto find optimal strategies

  4. Computational Social Choice Withthe power of NP-hardnessvulcanshaveconstructedcomplexityshieldstoprotectelectionsagainstmanytypesofmanipulationandcontrol.

  5. Computational Social Choice Withthe power of NP-hardnessvulcanshaveconstructedcomplexityshieldstoprotectelectionsagainstmanytypesofmanipulationandcontrol. • Question: • Are NP-hardnesscomplexityshieldsenough? • Or do theyevaporatefor single-peakedelectorates?

  6. NP-HardnessShields toProtectElections Elections & Voting Systems NP-hardness shields Manipulation & Control Manipulation & Control in Single-peakedElectorates Proof Sketch: CCAV in Approval

  7. NP-HardnessShields toProtectElections Elections & Voting Systems NP-hardness shields Manipulation & Control Manipulation & Control in Single-peakedElectorates Proof Sketch: CCAV in Approval

  8. Elections • An electionis a pair (C,V) with • a finite setCofcandidates: • a finite listVofvoters. • VotersarerepresentedbytheirpreferencesoverC: • eitherbylinear orders: > > > • orbyapprovalvectors: (1,1,0,1) • Votingsystem: determineswinnersfromthepreferences

  9. Voting Systems ApprovalVoting (AV) • votesareapprovalvectors in

  10. Voting Systems ApprovalVoting (AV) • votesareapprovalvectors in • winners: all candidateswiththemostapprovals

  11. Voting Systems ApprovalVoting (AV) • votesareapprovalvectors in • winners: all candidateswiththemostapprovals winners:

  12. Voting Systems PositionalScoring Rules (formcandidates) • definedbyscoringvectorwith • eachvotergivespointstothecandidate on positioni • winners: all candidateswithmaximum score Borda: PluralityVoting (PV): k-Approval (m-k-Veto): Veto (Anti-Plurality):

  13. Voting Systems PairwiseComparison v1: > > > v3: > > > v2: > > > v4: > > > Condorcet: beats all othercandidatesstrictly Copeland : 1pointforvictorypointsfortie Maximin: maximumofthe worstpairwisecomparison Hi, I am Ramon Llull. In 1299, Icame up with thevotingsystem that these guys nowstudy!

  14. Llull/Copeland Rule For FIFA World Championships or UEFA European Championships: Simply use = 1/3 as the tie value. Difference between the LlullandtheCopeland rule? What happens if the head-to-head contest ends with a tie? Llull: Both get 1 point Copeland0: Both get 0 points Copeland0.5: Both get half a point Copeland: Both get  points, for a rational , 0<<1

  15. Voting Systems Round-based: Single Transferable Vote (STV) v1: > > > v2: > > > v3: > > > v4: > > >

  16. Voting Systems Round-based: Single Transferable Vote (STV) v1: > > v2: > > v3: > > v4: > >

  17. Voting Systems Round-based: Single Transferable Vote (STV) v1: v2: v3: v4:

  18. Voting Systems Level-based: BucklinVoting (BV) v1: > > > v2: > > > v3: > > > v4: > > > v5: > > > • 5 voters => strictmajoritythresholdis 3

  19. Voting Systems Level-based: BucklinVoting (BV) v1: > > > v2: > > > v3: > > > v4: > > > v5: > > > • 5 voters => strictmajoritythresholdis 3

  20. Voting Systems Level-based: BucklinVoting (BV) v1: > > > v2: > > > v3: > > > v4: > > > Level 2 Bucklin v5: > > > winners: • 5 voters => strictmajoritythresholdis 3

  21. Voting Systems Level-based: FallbackVoting (FV) • combines AV and BV Candidates: v: { , } | { , } v: > | { , } • Bucklinwinnersarefallbackwinners. • IfnoBucklinwinnerexists (due todisapprovals), thenapprovalwinnerswin.

  22. NP-HardnessShields toProtectElections Elections & Voting Systems NP-hardness shields Manipulation & Control Manipulation & Control in Single-peakedElectorates Proof Sketch: CCAV in Approval

  23. War on ElectoralControl AV winners: "chair": knows all preferences

  24. War on ElectoralControl AV winner: "chair": knows all preferences andcanchangethestructure of an election

  25. War on ElectoralControl AV winner: "chair": knows all preferences andcanchangethestructure Other typesofcontrol: of an election • adding/partitioningvoters • deleting/adding/partitioningcandidates

  26. NP-HardnessShields forControl Resistance = NP-hardness,Vulnerability = P, Immunity, andSusceptibility

  27. NP-HardnessShields forControl

  28. References: Control • J. Bartholdi, C. Tovey, and M. Trick: HowHard isittoControl an Election?Mathematicaland Computer Modelling, 1992. • E. Hemaspaandra , L. Hemaspaandra, and J. Rothe: Anyone but Him: The ComplexityofPrecluding an Alternative. ArtificialIntelligence, 2007. (AAAI-2005) • P. Faliszewski, E. Hemaspaandra , L. Hemaspaandra, and J. Rothe: Llulland Copeland VotingComputationallyResistBriberyandConstructiveControl. Journal ofArtificialIntelligence Research, 2009.(AAAI-2007; AAIM-2008) • G. Erdélyi, M. Nowak, and J. Rothe: SP-AV FullyResistsConstructiveControlandBroadlyResistsDestructiveControl. MathematicalLogic Quarterly, 2009. (MFCS-2008) • G. ErdélyiandJ. Rothe: ControlComplexity in FallbackVoting. Proceedingsof CATS-2010. • G. Erdélyi, L. Piras, and J. Rothe: The ComplexityofVoter Partition in BucklinandFallbackVoting: SolvingThree Open Problems. ProceedingsofAAMAS-2011.

  29. War on Manipulation I like Spock but I don‘twanthimtobethecaptain!! Copeland : winner v1: > > > v3: > > > v2: > > > v4: > > > assumption: .v4knowstheother voters‘ votes v4 lies tomakehis mostpreferred candidatewin

  30. War on Manipulation I like Spock but I don‘twanthimtobethecaptain!! Copeland : winners v1: > > > v3: > > > v2: > > > v4: > > > Here: unweightedvoters, singlemanipulator . Other types: - coalitionalmanipulation - weightedvoters

  31. NP-Hardness Shields for Manipulation Results due toConitzer, Sandholm, Lang (J.ACM 2007)

  32. NP-HardnessShields toProtectElections Elections & Voting Systems NP-hardness shields Manipulation & Control Manipulation & Control in Single-peakedElectorates Proof Sketch: CCAV in Approval

  33. Single-Peaked Preferences • A collection V ofvotesissaidtobesingle-peakedifthereexists a linear order L over C such thateachvoter‘s „degreeofpreference“ risesto a peakandthen falls (or just risesor just falls). A voter‘s preference curve on galactic taxes lowgalactictaxes high galactictaxes

  34. Single-Peaked Preferences • A collection V ofvotesissaidtobesingle-peakedifthereexists a linear order L over C such thateachvoter‘s „degreeofpreference“ risesto a peakandthen falls (or just risesor just falls). A voter‘s > > > preference curve on galactic taxes lowgalactictaxes high galactictaxes Single-peakedpreferenceconsistentwith linear orderofcandidates

  35. Single-PeakedPreferences • A collection V ofvotesissaidtobesingle-peakedifthereexists a linear order L over C such thateachvoter‘s „degreeofpreference“ risesto a peakandthen falls (or just risesor just falls). A voter‘s > > > preference curve on galactic taxes lowgalactictaxes high galactictaxes Preference thatisinconsistentwiththis linear orderofcandidates

  36. Single-PeakedPreferences • A collection V ofvotesissaidtobesingle-peakedifthereexists a linear order L over C such thateachvoter‘s „degreeofpreference“ risesto a peakandthen falls (or just risesor just falls). • Ifeachvotevi in V is a linear order >iover C, thismeansthatforeachtripleofcandidates c, d, and e: (c L d L e or e L d L c) impliesthatforeach i, if c >i d then d >i e.

  37. Single-Peaked Preferences • A collection V ofvotesissaidtobesingle-peakedifthereexists a linear order L over C such thateachvoter‘s „degreeofpreference“ risesto a peakandthen falls (or just risesor just falls). • Ifeachvotevi in V is a linear order >iover C, thismeansthatforeachtripleofcandidates c, d, and e: (c L d L e or e L d L c) impliesthatforeach i, if c >i d then d >i e. • Bartholdi & Trick (1986); Escoffier, Lang & Öztürk (2008): Given a collection V of linear ordersover C, in polynomial time wecanproduce a linear order L witnessingV‘s single-peakednessorcandeterminethat V is not single-peaked.

  38. Single-PeakedApprovalVectors • A collection V ofvotesissaidtobesingle-peakedifthereexists a linear order L over C such thateachvoter‘s „degreeofpreference“ risesto a peakandthen falls (or just risesor just falls).

  39. Removing NP-hardnessshields: 3-candidate Borda veto everyscoringprotocolfor -candidate 3-veto, Leavingthem in place: STV (Walsh, AAAI-2007) 4-candidate Borda 5-candidate 3-veto Erecting NP-hardnessshields: Artificialelectionsystemwithapprovalvotes, for size-3-coalition unweightedmanipulation Results due toFaliszewski, Hemaspaandra, Hemaspaandra& Rothe (Information & Computation 2011) ConstructiveCoalitionalWeighted Manipulation General Single-peaked

  40. Removing NP-hardnessshields: Approval Constructivecontrolbyaddingvoters Constructivecontrolbydeletingvoters Plurality constructivecontrolbyaddingcandidates destructivecontrolbyaddingcandidates constructivecontrolbydeletingcandidates destructivecontrolbydeletingcandidates Results due toFaliszewski, Hemaspaandra, Hemaspaandra& Rothe (2011) Brandt, Brill, Hemaspaandra & Hemaspaandra (AAAI 2010)achievedsimilarresults forothervotingsystemsaswell (e.g., forsystemssatisfyingtheweak Condorcet criterion) and also forconstructivecontrolbypartitionofvoters. Controlfor Single-PeakedElectorates General Single-peaked

  41. Removing NP-hardnessshields: Approval Constructivecontrolbyaddingvoters Constructivecontrolbydeletingvoters Plurality constructivecontrolbyaddingcandidates destructivecontrolbyaddingcandidates constructivecontrolbydeletingcandidates destructivecontrolbydeletingcandidates Results due toFaliszewski, Hemaspaandra, Hemaspaandra& Rothe (2011) Brandt, Brill, Hemaspaandra & Hemaspaandra (AAAI 2010)achievedsimilarresults forothervotingsystemsaswell (e.g., forsystemssatisfyingtheweak Condorcet criterion) and also forconstructivecontrolbypartitionofvoters. Controlfor Single-PeakedElectorates General Single-peaked

  42. NP-HardnessShields toProtectElections Elections & Voting Systems NP-hardness shields Manipulation & Control Manipulation & Control in Single-peakedElectorates Proof Sketch: CCAV in Approval

  43. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1 9 5

  44. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 Whichvotetypesfrom Wshouldweadd? Especiallyiftheyareincomparable? 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1 9 5

  45. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 We‘ll handle this by a „smart greedy“ algorithm. 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1 9 5

  46. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 Whyare F, C, B, c, f, and j dangerous but theremainingcandidatescanbeignored? 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1 9 5

  47. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 First, each added vote will be an interval including p. So drop all others. 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1 9 5

  48. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 First, each added vote will be an interval including p. So drop all others. 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1

  49. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 Now, ifaddingvotesfrom Wcauses p tobeat c then p must also beat a and b. 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1

  50. A Sample Proof Sketch number of approvals from voters in V for candidates that are 1 Thus, c is a dangerousrivalfor p but a and b cansafelybeignored. 2 votes in Wthatcanbeadded (with multiplicities) 4 7 3 1

More Related