
Brief Announcement: Statement Voting and Liquid Democracy Bingsheng Zhang∗ Hong-sheng Zhou School of Computing and Communications Department of Computer Science Lancaster University, UK Virginia Commonwealth University, USA [email protected] [email protected] ABSTRACT the issues. The cost of identifying the best voting strategy is high, The existing (election) voting systems, e.g., representative democ- even if we assume that the voter has collected accurate information. racy, have many limitations and often fail to serve the best interest What’s worse, misinformation campaigns often inuence the voters of the people in collective decision making. To address this issue, to select certain candidates which could be against the voters’ own the concept of liquid democracy has been emerging as an alterna- interests. In the past decades, the concept of liquid democracy [6] tive decision making model to make better use of “the wisdom of has been emerging as an alternative decision making model to make crowds”. Very recently, a few liquid democracy implementations, better use of collective intelligence. Liquid democracy is a hybrid of e.g. Google Votes and Decentralized Autonomous Organization direct democracy and representative democracy, where the voters (DAO), are released; however, those systems only focus on the can either vote directly on issues, or they can delegate their votes functionality aspect, as no privacy/anonymity is considered. to representatives who vote on their behalf. Due to its advantages, In this work, we, for the rst time, provide a rigorous study of liquid democracy has received high attentions since the spread of liquid democracy under the Universal Composability (UC) frame- its concept; however, there is no satisfactory solution in the form of 1 work. In the literature, liquid democracy was achieved via two either paper-voting or e-voting yet . Is it possible to introduce new separate stages – delegation and voting. We propose an ecient technologies to circumvent the implementation barriers to liquid liquid democracy e-voting scheme that unies these two stages. At democracy? the core of our design is a new voting concept called statement vot- 2 A NEW CONCEPT ing, which can be viewed as a natural extension of the conventional We could approach the above problem via multiple angles. In this voting approaches. We remark that our statement voting can be paper, we propose a new and clean concept: statement voting. State- extended to enable more complex voting and generic ledger-based ment voting can be viewed as a natural extension of traditional non-interactive multi-party computation. We believe that the state- candidate voting. Instead of dening a xed election candidate, ment voting concept opens a door for constructing a new class of each voter can dene a statement in his or her ballot but leave the e-voting schemes. vote “undened” during the voting phase. During the tally phase, CCS CONCEPTS the (conditional) actions expressed in the statement will be carried out to determine the nal vote. Single Transferable Vote (STV) is a ! •Security and privacy Public key encryption; Mathematical special case of statement voting, where the voters rank the election foundations of cryptography; Distributed systems security; candidates instead of naming only one candidate in their ballots. The ranked candidate list together with the STV tally rule can be KEYWORDS viewed as an outcome-dependent statement. Roughly speaking, the E-voting, UC, Liquid democracy, Statement Voting statement declares that if my favorite candidate has already won or has no chance to win, then I would like to vote for my second 1 INTRODUCTION favorite candidate, and so on. In terms of liquid democracy, the vote delegation can be expressed as a target-dependent statement, Elections/Referendums provide people in each society with the op- where a voter can dene that his/her ballot is the same as the target portunity to express their opinions in the collective decision making voter’s ballot. Of course, the target voter can also state whether process. The existing election/voting systems can be mainly divided he/she is willing to be delegated in the ballot. To obtain the basic into two types, direct democracy and representative democracy. Al- intuition, let’s rst leave privacy aside and consider the following though the former one treats every voter equally, it is not scalable; toy example. therefore, the latter one is widely used in most countries. However, Example: Each ballot is in the form of ¹ID; action; targetº. If a voter representative democracy has many limitations and it often fails to V is willing to be delegated, he/she sets ID := V ; otherwise, sets serve the best interest of the people. For example, to make correct i i ID := ?. If V wants to delegate his/her vote to another voter V , decisions, the voters have to invest tremendous eort to analyze i j then the ballot is B := ¹ID; delegate; Vj º. If Vi wants to directly ∗This work is partially supported by EPSRC IoT Security Research Hub, EP/N023234/1. 1All the existing liquid democracy implementations, e.g., Google Votes and Decen- Permission to make digital or hard copies of part or all of this work for personal or tralized Autonomous Organization (DAO) do not consider privacy/anonymity. This classroom use is granted without fee provided that copies are not made or distributed drawback prevents them from being used in serious elections. Here, we note that for prot or commercial advantage and that copies bear this notice and the full citation straightforword blockchain-based solutions cannot provide good privacy in practice. on the rst page. Copyrights for third-party components of this work must be honored. Although some blockchains such as Zerocash [1] can be viewed as a global mixer, they For all other uses, contact the owner/author(s). implicitly require anonymous channels. While in practice, all the implementations of PODC ’17, Washington, DC, USA anonymous channels suer from time leakage, i.e., the user’s ID is only hidden among © 2017 Copyright held by the owner/author(s). 978-1-4503-4992-5/17/07...$15.00 the other users who are also using the system at the same time. Subsequently, the DOI: 10.1145/3087801.3087868 adversary can easily identify the user during quiet hours. ¹ vote º vote for vi , then the ballot is B := ID; ;vi . Suppose there Functionality FLiid are seven ballots: B := ¹V ; delegate; V º, B := ¹V ; vote;v º, 1 1 7 2 2 2 The functionality FLiid interacts with a set of voters ¹ ; vote; º ¹?; vote; º ¹ ; delegate; º B3 := V3 v3 , B4 := v4 , B5 := V5 V4 , V := fV1;:::; Vn g, a set of trustees T := fT1;:::; Tk g, an B6 := ¹?; delegate; V3º and auditor Au, and the adversary S. Let Vhonest, Vcorrupt and Thonest, B7 := ¹V7; delegate; V3º. Here, the eective vote of B1 is dened Tcorrupt denote the set of honest/corrupt voters and trustees, by B7, which is further dened by B3; note that B3 votes for v3; that respectively. FLiid is parameterized by an algorithm TallyProcess, 2 » ¼ means, B7 votes for v3 by following B3. Now let’s consider B6: B6 a table Tab, and variables result, T1, T2, T3, and Bi for all i n . follows B3; however, B6 is not willing to be followed by anyone; as Initially, set result := ;, T1 := ;, T2 := ;, T3 := ;; for i 2 »n¼, set a result, B6 also votes for v3. Finally, let’s consider B5: B5 follows Bi := ;. B4; however, B4 is not willing to be followed by anyone; as a con- Preparation: sequence, B5 is re-dened as blank ballot, ?. After interpreting the delegation statements, the nal votes are ¹v3;v2;v3;v4; ?;v3;v3º. (1) Upon receiving input ¹InitialTrustee; sidº from the Careful readers may wonder why this type of natural voting idea trustee Tj 2 T, set T1 := T1 [ fTj g, and send a noti- has never appeared in the physical world. Indeed, it is typically cation message ¹InitialTrusteeNotify; sid; Tj º to the S not available in the real life. Dierent from the toy example, in the adversary . (2) Upon receiving input ¹InitialVoter; sid; ηº from the reality, the voters care about privacy and anonymity. To ensure voter Vi 2 V, if jT1 j < k, ignore the input. anonymity, the voters are not willing to leave their identities in Otherwise, send ¹InitialVoterNotify; sid; Vi º to the ad- the ballots. If no identities (or equivalences) are included in the versary S. If jTcorrupt j = k, then additionally send a mes- ballots, then it is dicult for voters to “follow” other voters’ choices. sage ¹DelLeak; sid; Vi; ηº to the adversary S. The election committees might assign each voter a temporal ID to Upon receiving ¹VoterID; sid; Vi; wi º from S, achieve anonymity, but a voter needs to obtain the target voter’s if Vi 2 Vcorrupt, then set Tab»i¼ := wi ; temporal ID in order to delegate his vote. This requires secure peer- otherwise, if Vi 2 Vhonest and η = 0, then set Tab»i¼ := ?; 0 f gλ to-peer channels among all the voters, which is not practical. In else, generate random temporal ID wi 0; 1 , and set » ¼ = 0 “our design” paragraph in the Introduction, we will present the rst Tab i : wi . digital construction for implementing full-edged liquid democracy. Voting/Delegation: (1) Upon receiving input ¹Delegate; sid; Vj º from the voter 3 MODELING LIQUID DEMOCRACY VOTING Vi 2 V, if jT1 j < k, ignore the input. Other- = ¹ » ¼; delegate; » ¼º We for the rst time provide a rigorous modeling for liquid democ- wise, record Bi : Tab i Tab j ; send a message ¹ExecuteNotify; sid; V º to the adversary S.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages3 Page
-
File Size-