Abstract
Single Transferable Vote (STV) counting, used in several jurisdictions in Australia, is a system for choosing multiple election winners given voters' preferences over candidates. There are a variety of different versions of STV legislated and/or applied across Australia. This paper shows some of the unintuitive properties of some of these systems.
Introduction
Single Transferable Vote (STV) is the cornerstone of proportional representation in Australia, utilized for the Federal Senate as well as various state and local government bodies. While the general principle of STV—redistributing surplus votes from winners and preferences from eliminated candidates—is widely understood, the specific mathematical "fine print" varies significantly across jurisdictions. This paper examines how specific legislative choices regarding rounding, vote exhaustion, and redistribution formulas lead to outcomes that violate fundamental democratic intuitions.
Methodology
The authors conducted a rigorous logical and mathematical analysis of two primary legislative frameworks: the NSW Local Government (General) Regulation 2005 and the Australian Federal Senate rules. To validate their findings, they utilized ConcreteSTV, an open-source tool designed to implement these complex rulesets with high precision. The analysis was applied to historical election data, including the 2013 Victorian Senate and 2016 Tasmanian Senate elections, to determine if theoretical anomalies manifest in real-world counts.
The Three Fallacies of Australian STV
The research identifies three specific "fallacies" where the mathematical implementation of the law produces bizarre results.
1. The Negative Tally
In New South Wales local government elections, the formula for redistributing surplus votes involves a denominator that subtracts "exhausted" votes (ballots that no longer have valid preferences). Because the total tally is rounded down at each step of the count while exhausted votes are not, it is mathematically possible for this denominator to become negative. This results in a negative "transfer value," which effectively assigns a candidate a negative number of votes during redistribution.
2. Voting Alchemy (Value Inflation)
In the Australian Senate, when a candidate is elected, all ballots in their pile are assigned a new "transfer value" for redistribution. The authors discovered that ballots which entered a candidate's pile with a very low value (e.g., 0.06) can be "reweighted" to a higher value (e.g., 0.14) when that candidate's surplus is moved. This "Voting Alchemy" means a vote can spontaneously become more powerful as the count progresses, violating the principle that a vote should never exceed its initial value of 1.
3. The "Loser Wins" Paradox (Non-monotonicity)
Non-monotonicity is a known quirk of STV where receiving more votes can actually cause a candidate to lose. The authors demonstrated this using the 2013 Victorian Senate race. If candidate Helen Kroger had "given away" 2,595 of her votes to a rival (Joe Zammit), she would have altered the elimination sequence. This change would have knocked out the eventual winner, Ricky Muir, early in the process, ultimately allowing Kroger to win the seat herself.
Key Results and Impact
The study reveals that these anomalies are not merely theoretical edge cases:
- Frequency: "Value Inflation" (Fallacy 2) was observed in approximately one-third of recent Australian Senate elections.
- Magnitude: In the 2016 Tasmanian Senate count, individual ballots increased in value by over 125% (from 0.064 to 0.144) during a single redistribution step.
- Strategic Vulnerability: The 2013 Victorian Senate example shows that a candidate could have overcome a massive margin of 87,000 votes simply by strategically "losing" a small number of votes early on to manipulate the elimination order.
Conclusion
While some properties, such as non-monotonicity, are inherent to the STV algorithm, others—like negative tallies and value inflation—are "bugs" introduced by specific legislative wording and rounding choices. These findings suggest that future electoral legislation should be drafted in close consultation with mathematicians and computer scientists to ensure that the laws are unambiguous and avoid these problematic mathematical outcomes. Ensuring that the mathematical reality of the count aligns with democratic intent is essential for maintaining public trust in the electoral process.
Summary updated