The dynamics of strategic voting: pathways to consensus and gridlock

Abstract: The outcomes of democratic elections rest on individuals' decision-making that is driven by their varying preferences and beliefs. Individuals may prefer consensus to gridlock, or gridlock to consensus, and information may be fractured via echo-chambers. To understand the role of these factors in whether or not elections reach consensus, we develop and explore a computational model in which voters have varying party affiliations, preferences, beliefs, and voting strategies. Voters may change their voting strategies either by imitating others or reconsidering their strategy individually. Preferences are orderings of the following election outcomes: a voter's party winning a super-majority, the opposing party winning such a majority, and gridlock. Voters beliefs and decisions are shaped by their social networks, and thus are heterogeneous in the population. We observe a "tipping point" phenomenon wherein the voters' initial strategies and randomness impact whether the minority party voters vote to create gridlock or consensus. A positive feedback loop secures such voters into one behaviour or the other. Consensus is reached by the minority party evolving to prefer consensus, which in turn drives the majority to also prefer consensus due to the influence of social learning. Further, consensus is promoted by an uneven distribution of party affiliation, and undermined when it is even. We also find that a moderate prevalence or strength of echo-chambers can boost consensus, since they can quell voters' desires for gridlock.