Also called Instant Run-off Voting (IRV). Note that SVDSA uses a variant: SCottish single Transferable Vote (SCTV).
Problems With Existing Plurality Voting
We aren’t talking about the system of representation (parliamentary or presidential, bicameral or not, electoral college, number of parties …), just about the method of electing representatives.
Both the US and India (one of them is the world’s largest democracy) have “first past the post” voting — of all the candidates, the candidate who gets the most votes (a plurality) wins, even if that is a minority of the total votes cast.
The US has a plurality-based voting method which lazy political analysts are able to continuously miscast as a “majoritarian democracy” only because of the rigid two-party system in place — with just two candidates, the plurality winner will necessarily have the majority, simply due to a lack of competition. In a 2 party system, plurality voting leads to a disregard of the wishes of a sizable minority in the district. Furthermore — at the level of the assembly of representatives — it can easily lead to “minority rule”: A party may lose badly in a minority of seats and lose the overall majority vote, but win by small margins in a majority of districts and win governing control. A two-party plurality encourages gerrymandering and inequitable representation via an entrenched “minority rule”.
In a multi-party system, plurality voting leads to horse-trading, “spoiler candidates” etc., and most importantly, the apparent disregard for the voters of anybody but the winning candidate, which could easily be the majority of voters.
Wikipedia on Ranked Voting: lots of details of RCV, IRV and multifarious variants.
A lovely paper on the ideal number of representatives: Basically, N, where N is the size of the population to be represented.
What are general attributes of an ideal fair voting system?
(Just so we know what we mean by a “good” voting system.)
This is a list of four attributes by Kenneth Arrow 1963, from New Scientist:
- Voters should be able to express a complete set of their preferences
- No single voter should be allowed to dictate the outcome of the election
- If every voter prefers one candidate to another, the final ranking should reflect that
- If a voter prefers one candidate to a second, introducing a third candidate should not reverse that preference.
Note that there are mathematical theorems that prove that no conceivable voting system can meet all 4 criteria. So there can be no “best” because all voting systems involve trade-offs between the above criteria.
However, there are voting systems that reduce the odds of missing any of the criteria, and the fact that there are exceptional distributions of favorability rankings that can upend one or other of the rules should not be a reason to reject voting systems that are vast improvements over the plurality voting which manifestly and repeatedly violates the above 4 common sense criteria for fairness.
Ranked Choice Voting as an Alternative to Plurality voting
RCV has been around since 1870 and over the last few decades has regained attention as an attempt to meet the above criteria and bring us closer to an ideal fair voting system. It avoids many of the problems of plurality voting mentioned above.
Australia seems to be the earliest to use IRV at the Federal level, India uses IRV or some variant for the upper house in state legislatures and for Presidential elections, and Pacific nations seem to have adopted it quite extensively.
How Ranked Choice Voting Works
Single Winner per District
Every voter ranks the candidates in descending order of preference. The counting is an iterative process. Count the first place votes for each candidate. If a candidate has more than the majority of votes, they win and the counting stops. If no candidate meets the threshold of 50%, the candidate with the least first place votes is dropped (run off the board, so to speak). The votes on these ballots are then assigned to each ballot’s second ranked candidate still in the running and added to their total. The process stops when a candidate’s cumulative vote count exceeds 50%.
Multiple Winners per District
What if there are to be 3 representatives to be elected from the same district? The same process as above is used, with one minor modification: the threshold for winning is changed -in this case- from 50% to 100%(3+1)=25%.
If there are n representatives to be elected, then regardless of the number of candidates, the threshold is 100/(n+1)%.
Consider an election in which there are three candidates for two seats. If the voters have no overall preference, each candidate will get a third, or 33% of the votes. However, if the voters have any aggregate preference at all, one candidate will get fewer than 33% and at least one will get more than 33%; the latter can be safely declared a winner. Clearly 25% is too low a threshold: with only three candidates for two seats, all three can conceivably reach 25%, which would yield three “winners” for two seats. So the minimum % to win has to be 100/(n+1)%. Why can’t we insist on an even higher threshold, for example a “majority”, i.e. 50%, in order to win? In the case of two electees, only one can exceed this threshold and we won’t be able to fill the seats. Similarly, at most n-1candidates can exceed 100/n%, which would leave us one electee short. So 100/(n+1)% is the maximum minimum.
What about the waste of excess votes for a winning candidate?
What is a “wasted excess vote”? Let’s say there are three (or more) candidates for two seats, so the winning threshold is 33%. At some stage, the first winner emerges, Candidate A, with say 40% of the vote, and further suppose that candidate B has 31% and Candidate C has 29%. It then looks like the following should happen: A is no longer in the running, C is the current lowest ranked candidate, they are dropped and their voters’ votes transferred to the next choice — B. So B would be the second winner.
But then it seems as if the voters for A in the previous round don’t have any say in the next winner. For example, let’s say that of the 40% who voted for A, 84% ranked C second and only 16% ranked B second. Many of the 40% (the 7% in excess of the 33% threshold) might feel cheated with respect to their second choice — if they had known A would win by such a large margin, they could have ranked C first without risking A’s win, possibly leading to C and A being the two winners instead of A and B.
In order to avoid the earlier pathological outcome (the less preferred candidate B winning the second seat and ACB voters being aggrieved), the excess of votes above the winning threshold for any candidate are then allocated to their next-ranked candidate, in proportion to the excess — in the example above, for all the 40% who voted for A in first place, their next ranked candidate will get 7/40th of a vote. If 84% of the 40% ranked C second, that would amount to a transfer of (7/40)*(0.84)*40% = 6% of the total to C, pushing C to 29+6 = 35% and relegating B to 31+1 = 32% of the total.
So, before the lowest current candidate is run-off and their votes transferred, the most recent winner’s excess votes are proportionally allocated to the next ranked candidate. This, essentially, is “Single Transferable Vote” (SCTV).
Ballotpedia — Encyclopedia of American Politics on RCV — Look for the section “How ranked-choice voting works”
What It Looks Like in Practise at DSA
Step by step discussion of the Results from the last Officer election.
There were elections for 5 officers, and 39 valid ballots. So first let’s establish the winning threshold. This is 39/(5+1)=6.5 which is rounded up to 7. Let’s look at Round 1:
Candidates A, C and F have exceeded the threshold by 4, 2 and 3 votes respectively. If we were doing straight-up IRV, and there are only 6 candidates, we would be tempted to drop Candidate E. That would have implied (A,B,C,D,F) being declared the winners (or E’s next place votes being transferred). Under SCTV however, A is declared a winner, and for the 11 people who ranked A first, their second ranked candidates still in the running get 4/11th of a vote for the next round. A, C and F are treated simultaneously, there is no need to treat them sequentially.
After transferring 4/11th,2/9th and 3/10thof the second rank votes from the ballots which ranked A,C and F first, we get
So B is declared a winner, their votes are transferred to E.
Only at this stage is a losing candidate, D, identified, their next ranked votes transferred to the only remaining candidate in the running, E, who then surpasses the threshold and is also declared the winner for the last of the 5 positions.
Why Ranked Choice Voting is a Good Alternative
A list of the pros and cons, directly from the League of Women Voters of Vermont:
PROs of IRV/RCV
Promotes majority support — Voting continues until a candidate has a majority of votes.
Discourages negative campaigning — Candidates who campaign negatively against another candidate risk losing the next place votes of those who favor the latter to the former.
Provides more choice for voters — Voters needn’t worry about the spoiler effect.
Promotes a diversity of political viewpoints being represented, even in single representative elections, on average.
Minimizes strategic voting — Voters needn’t feel compelled to vote for ‘the lesser of two evils”.
Provides an outcome more reflective of the majority of voters than either primaries (which leads to extreme candidates “playing to their base”) or subsequent run-off elections (which typically have far lower turnout).
Avoids the reduced representation due to unevenly distributed drops in voting rates for subsequent elections, especially for populations that already face barriers to exercise their vote (the elderly, selectively disenfranchised populations, overseas citizens or military personnel.)
Saves cost compared to running primary elections or run-offs. Because America, everything has to be about money.
CONs of IRV/RCV
It is new — A certain percentage of people don’t like change. This isn’t really a “con” of IRV, it is just resistance to change.
It will require education about how it works — Well of course.
The ballots and the counting of the ballots will be more expensive — if it is done by hand. Modernize US voting already!
The “vetting” is less clear — This is not so much a “con” of IRV as it is an “American exceptionalism” based misguided defense of primaries.
You could still fail to get a candidate with a majority — If enough voters did not complete the ranking of all candidates. This can be fixed by treating the “missing” lower rankings as abstentions.
There are a bunch of other specious arguments against RCV, some of them seem motivated by resistance to change, others, from conservative what-passes-for-thinkers are based on “oh, this might hurt my guy” and “if it ain’t broke, don’t fix it, and it ain’t broke for me”.
CampaignLegalCenter — RCV A nice graphic and a short video.