This article describes and demonstrates a novel solution for electoral reform called Representative Proportional Voting (RPV) - similar to the open-list proportional representation system used by Finland - that it is practical to implement. It is also more effective in holding elections where every vote contributes to the formation of a government body. Below is the overall results of the 2021 Canadian general election under the presently used First Past the Post (FPTP) electoral system compared to Representative Proportional Voting (RPV).

I’ve included this table at the top to highlight how ineffective the current electoral system is at giving voters a voice. From a numerical point of view, by analyzing each district, FPTP resulted in 52.15% of all votes (8,883,902), NOT GOING TOWARDS AN ELECTED MP SITTING IN THE HOUSE OF COMMONS!. In contrast, under RPV, 99% of votes went towards an elected candidate.

Introduction

The 2021 Canadian general elections had just passed and I was curious as to how proportional the election results were. I noticed that the Liberals were over-represented. I then began a thought experiment on what an ideal democratic electoral process would look like for Canadian parliamentary elections. When the Liberal party led by Justin Trudeau came into power in 2015, one of their promises was electoral reform of the First Past the Post system currently in use. However, as the last elections have shown us, the Liberal party is substantially benefitting from First Past the Post and it is not strategic for them to follow through on their first campaign promise.

Party List Proportional Representation

A lot of Canadians demand electoral reform without suggestions as to how, while others criticize common proportional voting system (e.g. party-list) for lacking “local representation.” There are also others asserting or even bringing up ranked ballots as the holy grail for democracy, but I digress.

Ranked Ballots

Ranked ballots work great when it’s just for one localized governance, such as mayor or a university student association. If it were applied to federal elections we would still not have proportional representation. Rather, we would see an amalgamation of parties most closely aligned (vote A if not vote B). Ranked voting is strategic voting and minority voters would still be unrepresented.

RPV Overview

Representative Proportional Voting (RPV) is similar to the open-list PR system used in Finland with two caveats being that independent candidates still have a chance and that the huntington-hill method is used to appropriate seats rather than the D’Hondt method. This new system is applicable to provincial/territorial & federal elections.

Voters will vote for their preferred candidate and if they prefer an independent candidate, they can further rank these independent candidates and rank a party-affiliated candidate last, A run-off is held for votes going towards independent candidates are collapsed until the last independent candidate remains. An independent candidates can be elected if they win the First Past the Post manner. Otherwise, they are eliminated and the rankings go to party-affiliated candidates that were ranked last. The votes going towards each of the remaining party candidates are considered as party votes and are pooled for the region. The remaining seats are distributed proportionally based on which districts had the highest percentage support for each party (note that highest percentage support does not necessarily mean the popular vote in that district).

An example to illustrate my point. Say that there is party A that would win the popular vote in every district, and party B which got the rest of the votes for each district. The seats party B and A are entitled to is determined in a round robin fashion of selecting the district that had the greatest percentage support for said party. The order is least seats to most seats to ensure that the smaller parties get districts they had high support in. Otherwise we might get a less than optimal distribution.

In the case of a province or territory that only has one seat, a ranked ballot with runoff would suffice.

One possible criticism is that there will be ridings where the candidate that is elected didn’t get second or third either. This is a valid tradeoff since the current situation is that your vote would be meaningless if the representative of the other party got even one more vote than the party-affiliated representative you voted for. With my proposed RPV, all votes will matter even in hindsight. So the upsides are that: independent representatives are favoured, one of the running candidates will win the riding, every vote counts towards the governing body rather than only for a singular riding, and the democratically elected body is proportional to how voters voted.

To incentivize and protect independent candidates, we give independents priority when picking representatives for each riding. If they were to be included in the Huntington-Hill method for proportional seat distribution, it would be possible for no seat to be allocated to them since party’s would have a greater proportion of votes. This is why I kept the same voting procedures / logistics.

I’ll probably hear people complaining about a non-FPTP candidate winning. No candidate should be entitled to represent a riding without 50% of support and a ranked ballot is not guaranteed to reflect the federal wide policies voters want. Therefore, RPV is a valid Canadian-centred compromise. I will now dive into the technicalities as well as the practical results if RPV was the system used in the last election.

RPV Algorithm

It is simply amazing that computer science can even be applied to politics. I’ve outlined the algorithm for RPV below. Note that this version of the algorithm and the election results distribute seats on a federal level rather than on a region level. I will update the algorithm and results soon. It incorporates many concepts such as FPTP, Huntington-Hill (fun fact is that it is used for the electoral college points in the U.S.A.), Hare Quota, and even ranked ballots (for niche situations).

1. Just like how it is presently, each person will vote for a candidate in their riding with some exception.
   • Voters voting for an independent candidate are allowed to rank their preferences for independent candidates with their final choice optionally being a party-affiliated candidate. This ensures that their vote won’t be lost if no independent candidate won.
   • Provinces & territories that only have one seat in the house of commons should be using ranked ballots. In this manner, voters of each province & territory get to choose best how their seats will be distributed
2. Once all the votes are counted per riding, determine independent candidates that won via the current FPTP method. Eliminate and use the ranking system to determine who the most popular independent candidate is. If they got a higher vote than the party candidates, then they win the riding according to FPTP. Otherwise, eliminate this independent candidate to get the number of votes for each party. Note that party candidates are not elected on a FPTP basis, which is why the party vote count is what is important.
3. Remove the ridings that were won by independent candidates from all available ridings.
4. Use the Huntington-Hill method to determine number of seats to allocate to each party (independents are not considered a party).
   • We will use the Hare quota to determine whether a party deserves a single seat; The total votes (minus independent votes) divided by the total seats available (after removing independent won seats).
   • Initially allocate one seat to each party that meets this quota and discard all parties and independents that did not meet this quota.
   • Allocate seats to whichever party has the highest seat priority until total seats have been allocated.
   • The seat priority is calculated using where V is the the number of votes that party received, and n is the number of seats that party has been allocated so far.
5. Assign seats based on highest proportion of votes per riding.
   • Sort parties in ascending order based on number of seats allocated and prefer parties with higher votes to break ties
   • Iterate over the parties in the previous order until all parties have had their seats assigned to a riding.
     • For each party that has unassigned seats, assign a seat to an unelected riding with the highest percentage voting for that party.
6. Use a list or graph of all ridings with their assigned representative to keep track of seat assignment / election results.

44th 2021 General Elections FPTP Versus RPV

I wrote the first draft of this article in Nov-2021, however I wanted to show how the 2021 election would be different if RPV was used rather than FPTP before I shared RPV to everyone.

Methodology: I wrote Python code that downloaded 2021 election polling data, transformed that data to be useful, implemented the RPV algorithm, supplied data to the algorithm, and lastly summarized the results for consumption.

Vote Distribution

Vote distribution for all parties (that got at least one vote) in the 2021 Canadian Election

• There were 338 districts in this election.
• Total number of votes counted was 17,034,224
• Total votes that went to a party is 17,008,619

FPTP Results

My code also aggregated votes that were not cast to the winning candidate, I calculated that 8,883,902 votes or 52.15% of all votes DID NOT GO TOWARDS ELECTING A CANDIDATE!. In other words those votes did not matter in hindsight on top of citizens not going to the polls.

RPV Results

The Hare quota is 50,322. There were 87 seat changes from the FPTP results.

It’s clear that under RPV, we can get a 99% representative government body where the 1% that went unrepresented voted for parties that got less than ~50,000 votes (Hare Quota). For the list of ridings that would have a different representative, you can download a CSV of the compiled results.

Some notable points are that all elected parties would have more seats than the proportional vote because 1% of votes went to very unpopular parties. The liberals lose 49 seats, NDP doubles their seat count, and the PPC ends up getting more than double the seats than what Green would get. This sort of system would force all parties to reach out to every voter even in districts that are considered wins.

Since this is a post-election analysis, numbers for some parties will be under-represented even though they won more seats. For example, it is common for voters to vote Liberals just so that the Conservatives don’t win and it’s possible some people voted Conservative instead of PPC so that the Liberals don’t win. Therefore, under RPV, we could reasonably expect bigger parties to lose even more seats to smaller parties. This logic does extend to parties on the same side of the spectrum such as the Green party. As seen in the CSV provided, we could see Green party getting seats that would have been NDP under FPTP.

RPV would do away with strategic voting to the point where it would be ignorant to vote for a party other than the one you genuinely support. A greater proportional system would also promote bootstrapping of parties as support only needs to be nation wide to win a seat rather than concentrated in one riding.

Under the RPV system, 99% of voters ended up represented. The votes of the 1%, although divided, do matter in aggregate. For example, the Free Party had just under the threshold to get a seat. Every vote really does matter under RPV compared to FPTP, where voting might have seemed like a waste of time in hindsight. Only under an RPV system would I ever support a policy that forced citizens to check into a polling station to vote or explicitly abstain from voting. There would be almost no subjective reasons to avoid voting under RPV as it would be in direct opposition to your interests.

Reproducing The Results

Reproduction is very important. My Python code, the CSV file listing both FPTP and RPV results, and a JSON for further analysis can be found on GitHub.

Conclusion

In conclusion, every voter’s voice matters under RPV. If I was/am ever an MP or MPP, I would propose to use this method for all provincial/territorial and federal elections. I strongly believe that under RPV, Canada will be setting the standard for a democracy where each citizen’s voice is valued due to aggregation rather than the current system where a vote only goes as far as the district.

Extensions

One criticism that I can see for my own method is that independent candidates are given a relatively unfair advantage to win their riding. However, an independent candidate will most likely lack the financial + labour + social resources that party-affiliated candidates will have. Our country is considered a democracy even with the FPTP electoral system, so allowing independent candidates to make use of it in a reformed electoral system is really not a good argument especially when results are more democratic anyways. The majority of Canadians have not experienced something better than First Past the Post in the first place.

The CSV file provided can be used to draw a colored map of each riding as well as to summarize which parties traded seats. For example, Conservatives went down 4 seats in total, but if you look at the CSV, half the seats lost were to the PPC and the other half to the NDP. They gained most of these seats back from the Liberals and some from the Bloc Québécois. Zero seats were lost to the Liberals. Another example is that the NDP lost two seats to the Green party and one seat to the PPC. That one seat is the only seat the PPC gained from a party other than the Conservatives.

My hypothesis is that this type of system would force the mainstream parties to invite smaller parties to debates. Otherwise the smaller party voters would simply not watch the debates and vote for the smaller party anyways. We know that the PPC got votes even with the backdrop of strategic voting, but only under FPTP does 5% round down to 0%. In a proportional system, all parties had more seats proportional than voters’ choices.

There is also a system countries like Norway use, however the main point is to make electoral reform easy, simple, and effective with minimal tradeoffs.

Electoral Misconduct

Like with every law and system, there will be people trying to bend it. One such situation I could forsee far into the future are candidates running as an independents but supported by a party through advertisements/funding. Elections Canada would be tasked at ensuring such faltugiri horsing around would be met with major fines.