The Problem
There are two problems with preferential voting:
- it arbitrarily breaks up large communities into blocks that elect one person each with results that aren't necessarily representative of the community overall; and
- it trends over time towards two parties, making it basically impossible for smaller niche parties to win seats, no matter how popular they are.
If you need a refresher on how Preferential voting works, click here.
Let me give you an example: imagine a city of 300,000 people that's broken into three electorates of 100,000 people each. At the end of counting and after all preferences have been accounted for we end up with these three electorates each voting for Party A at 51% over Party B at 49% meaning three seats are awarded to Party A candidates. But what about the 49% in the wider community who wanted Party B? Just because the Electoral Commission chose to draw imaginary lines to break up their community it means those people are unrepresented. A more proportional outcome would be two seats to Party A and one seat to Party B.
In a first-past-the-post (fptp) election this scenario is even more common because a party can be elected without even getting close to a majority, like the winner of Belfast South in the 2015 UK election with only 24% of the vote. That's part of the reason why Australia switched from fptp to preferential voting in 1918-19.
But at least under fptp a smaller niche party can win a seat if they get enough votes! With preferential voting the winning candidates trend toward broad-based parties over time. This is because it's uncommon for a candidate to receive a majority of first-preference votes so they have to rely on preferences, and since smaller niche parties are more polarising and/or unfamiliar than their well-known and "safer" broad-based counterparts, they're less likely to appear near the top of ballot papers.
Why is this a problem? Because it creates an "Us vs Them" political environment where it's winner-takes-all and no compromise. At best a large party can form factions but then debate happens behind closed doors, and those factions can't break away from the main party without falling victim to the same problems as niche parties. People have strong opinions about crossbenchers and majority vs minority governments and I'm not going to comment on that here but suffice to say that a good democracy happens in the public theatre with a diverse range of perspectives.
The Alternatives
So how do we fix this? There's a few options out there to look at:
- First Past The Post like the UK,
- Proportional Representation which the Senate uses,
- Preferential Bloc Voting which the Senate used in the past,
- Hare-Clark like Tasmania and the ACT,
- and Mixed-Member Proportional which NZ has been using for a few decades.
But they don't quite solve the problem:
- FPTP still arbitrarily breaks up large communities,
- PR uses massive electorates that don't represent individual communities,
- PBV typically awards seats to the same party because it does full recounts,
- HC is pretty good but it's difficult to understand and doesn't carry preferences in-full,
- and MMP isn't appropriate in Australia because it effectively merges the House of Reps (locally-elected representatives) with the Senate (proportionally-elected review body).
To complicate things further, it would be such a political schism to completely upend our ol' faithful preferential system for one of these alternatives. But what if we could just modify the preferential system a little bit for electorates that don't fully cover a community?
Introducing: Popular Consensus Voting.
Popular Consensus Voting
There are 5 elecorates that overlap with the Gold Coast's metropolitan area:
- Fadden (wholly)
- Forde (partly)
- McPherson (wholly)
- Moncrieff (wholly)
- Wright (partly)
Let's take the three that fit wholly within the city's boundary (Fadden, McPherson, and Moncreiff) and merge them together into one electorate that awards three seats. We can do this without any issues because these three are undeniably "Gold Coast" - the people there have a shared identity with shared interests in the same metropolitan area.
At the next election people fill out their ballot just like they do now, numbering the candidates 1-N in order of preference. The only difference that voters will see is that:
- there might be multiple candidates from the same party, but this is fine because you're voting for a person not a party; and
- because of that the ballot is noticeably longer, so it's worth thinking about letting people fill out a minimum number of preferences rather than the whole ballot.
On election night the votes are cast and the count begins. Rather than requiring a full majority at 50% + 1, we set the threshold proportional to the number of seats: in this case 25% [why 25%?]. All the first preferences are tallied up and we find that no candidate has met the threshold, so just like normal the candidate with the fewest first preference votes gets removed and those ballots are redistributed to their second preference.
The second round of counting starts and we find that Candidate 1 has exceeded the threshold so they're awarded a seat. The threshold gets recalculated for the remaining 2 seats making it 33.33%, and all the ballots that had Candidate 1 as their first preference get redistributed to their second preference. Then again just like normal the candidate with the fewest votes gets removed and those ballots are redistributed to their next preference.
The third round of counting starts and this time there's two candidates that have reached the new threshold so they're both awarded seats. Now there's no seats left so the election is complete.
The Explanation
So what does popular consensus change? Well:
- neighbouring electorates with a shared identity are merged and given multiple seats;
- multiple candidates may be listed for the same party (which can happen now but they don't do it because it would be pointless);
- the threshold for winning a seat is proportional to the number of seats available; and
- ballots are redistributed from both ends as candidates get removed.
But isn't that what Hare-Clark does? It's similar but there's two main differences:
- the threshold increases as seats get awarded so preferences matter more than in Hare-Clark elections; and
- surplus votes for a winning candidate don't get redistributed, the ballots themselves do just like when the lowest-ranked candidate gets removed from the bottom end of the tally.
This is important because it treats each round as its own election without throwing away all the context gained so far from preference flows (which is what kills Preferential Bloc Voting).
Plus, from an implementation perspective, it isn't a total rethink of the way votes are counted now which is important for simplicity - nobody wants two completely different voting systems for the same parliament! With popular consensus the count just continues in the way that preferential already does. If an electorate awards one seat then absolutely nothing changes.
Why does the threshold move? This is where popular consensus voting gets its name: the threshold starts low which means that first preferences matter a lot because a candidate can win a seat if voters really like them (this is the "popular" part in popular consensus), but as the seats get awarded and the threshold increases the preference flows to broad-based parties start to kick in and we end up awarding seats to someone in the middle (which is the "consensus" part).
This does a few things:
- smaller niche parties can actually win a seat if they get enough first preferences (ie their policies have motivated enough voters to want reform);
- preferences still matter just as much as they do now so broad-based parties aren't alienated; and
- party factions become a factor if a party is standing two or more candidates because voters can cast a nuanced ballot rather than being stuck with the only option they get.
Critiques and FAQs
What happens to rural electorates? Nothing. Rural electorates can continue to award one seat and follow exactly the same process as they do now - popular consensus doesn't change anything about preferential voting, it just adds support for multi-seat electorates.
This is also relevant for the roll out: the Electoral Commission can trial popular consensus in a few obvious places like the Gold Coast or inner-city Melbourne without committing to a nationwide roll out. Barely anything changes and clear comparisons can be made with a preferential count.
However if it's deemed appropriate or the voters demand it, then rural electorates could be merged the same way. For example Indi and Nicholls in Victoria or Gilmore and Eden-Monaro in New South Wales, both of whom are made up of dozens of small-medium towns and cities over a common region.
What's to stop a party from saying "vote for Person A, then B, then C"? This is the hack that kills Preferential Bloc Voting, but Popular Consensus Voting is largely immune to it because of the moving threshold.
The issue with Preferential Bloc Voting is that each seat is awarded via full recounts, so in a 3-seat electorate you'd run the same Prefential Voting process three times. If Person A wins in the first round, then Person B would win the next, and Person C after that simply because the preferences flow the same way each time and you essentially replay the same election.
Popular Consensus Voting solves this with the moving threshold and redistributing preferences from winning candidates rather than starting again. In the same scenario Person A would win the first seat when the threshold is lowest, and righfully so because enough people really like them. Their votes are redistributed to Person B who may not end up with enough to reach the increased threshold in the second round, meanwhile other candidates are benefitting from preference flows and increasing their share of the vote. Eventually the final round would be between the two most preferred candidates which might not even include Person C, especially when you consider party factions in voter preferences.
But of course a simple fix for all of this is to have the Electoral Commission ban parties from telling voters which order to vote for people in.
Aren't the ballots for winning candidates effectively counted twice when they're redistributed? Yes, but if you flip it around, no.
When there's only one seat remaining, the final round of counting works the same way as it does currently: preferences flow until one candidate has reached a 50% majority. For the winning candidate each ballot is counted once.
For the earlier seats where popularity matter more, then yes, any ballots that went towards both a popular winning condidate and the consensus winning candidate are effectively counted twice. But this is by design: unlike proportional systems where candidates are elected equally, we're electing candidates who receive a high number of first-preferences (they're popular) as well as a candidate that most people can agree on (the consensus candidate); and if there's ballots that overlap between the two types then yes one ballot can go towards electing two people.
