Reality-aware Sybil-Resilient Voting

Gal Shahaf, Ehud Shapiro, Nimrod Talmon

Research output: Working paper/PreprintPreprint


We consider an electorate composed of both honest and fake voters, aka sybils, and investigate the resilience of its democratic decision making against two types of sybil attacks: Sybils enforcing decisions on the honest voters, and sybils blocking decisions by the honest voters. We follow Reality-aware Social Choice and use the status quo as the anchor of \emph{sybil resilience}, which we characterize by \emph{safety} -- the inability of sybils to change the status quo against the will of the majority of the honest voters, and \emph{liveness} -- the ability of the honest voters to change the status quo against the will of the sybils. We show that a reality-aware supermajority, in which a simple majority plus half the sybil penetration rate is required to change the status quo, is safe. Interestingly, a reality-aware supermajority is similar to byzantine failures in its tipping point: Below one-third sybil penetration, it assures both safety and liveness, as a sufficiently-high majority of the honest voters may effect a change of the status quo. Above one-third, it assures safety but not liveness, as sybils, while unable to force a change to the status quo, may block any change to it. The proposed reality-aware supermajority rule is safe only if all honest voters participate in the vote. Hence, we explore two approaches to maintain safety while mitigating partial voter participation: One is to further increase the reality-aware supermajority as a function of the participation rate. Another is default (forced) delegation of uncasted votes in support of the status quo. Both ensure safety but burden liveness by increasing the supermajority of active honest voters required to change the status quo. We analyze and compare them.
Original languageEnglish
StatePublished - 2018


Dive into the research topics of 'Reality-aware Sybil-Resilient Voting'. Together they form a unique fingerprint.

Cite this