[IAP] Forms of Personalised Proportional Representation

[Marcel Eðo Pairescu Tafial, UrGP]

This post is for l'Institüt del Agroïcismeu Political (the Institute for Political Nerdery)

During the last Cosă term, there has been growing cross-party support for the adoption of a voting system, which would enable voters a direct choice for their MCs, rather than leaving that choice to party functionaries who are in control of candidate lists, while preserving the current proportional allocation of seats. Such systems have become known as Personalised Proportional Representation in Talossa, or PPR for short. There have been multiple proposals for implementing PPR, each coming with advantages and disadvantages, which I want to showcase here.

1. Mixed-member Proportional (MMP)

MMP is the voting system in use in Germany and New Zealand, though there are differences in implementation.  At its most basic, voters under MMP have two votes: one for a local representative for their constituency, and one for a nation-wide party list. Voters are free to "split their vote", i.e. they can vote for a local represenative from one party and for a party list of another. The total number seats that a party receives are determined by the party list vote, and then those seats are "filled up" first by constituency winners, and the rest by party list members.

There have been proposals to implement MMP since 2014, most commonly in form of 8 constituency seats (one per province), with the remaining 12 for list candidates. Due to voters' ability to split their votes and the possibility of strongly localised support for parties, it is possible that a party may end up with more constituency winners than total seats. These are called "overhang seats". In New Zealand, overhang seats are added to Parliament without compensation, meaning that parliament size increases and proportionality is damaged. As of September 2026, New Zealand Parliament has 120 regular seats and 3 overhangs. In Germany, in the case of theoretical overhang seats, candidates enter the Cosă by the margin at which they won, meaning that overhang seats won by the narrowest margins are ignored. Parliament size thus stays proportional and at its canonical size, but some constituencies are left without a local representative ("orphaned"). As of September 2026, 23 out of 299 constituencies (7.7%) are orphaned, mainly in urban areas.

1.1. Advantages

MMP has been used in Germany since 1949, and in New Zealand since 1996, so it is a system that has been proven to work for high-stakes elections long term. The introduction of a constituency vote gives voters a direct choice as to who would represent each province without caveats and without strict adherence to any one party.

1.2. Disadvantages

MMP would require a drawing up of constituencies for the purposes of Cosă elections. While it seems straightforward enough to make every province their own constituency, considering the stark differences in province populations, doing so could disproportionately favour smaller provinces at the expense of bigger ones.

The stark differences in province activity is another hurdle to implementing MMP. It is not unimaginable that at least some constituencies would have candidates running unopposed, which in a sense is undemocratic, and it would also hurt the proportionality of the final results.

2. Open-list Proportional ("Kosovo system")

Open-list Proportional voting is used in dozens of countries worldwide on various levels of government (including in Kosovo, hence the colloquial Talossan name), and is perhaps the method most similar to what is already in use currently. At its most basic, voters vote for a party list, and then for a single candidate on that list that they like the most. Voters cannot split their vote. Seats are distributed by party vote, and assigned to party list members by the number of candidate votes they received. Ties are broken by party list order.

2.1. Advantages

The "Kosovo system" is by far the easiest to implement, as it is the closest to the pre-existing closed-list proportional system. The ability to pick the order in which list candidats enter the Cosă makes sure that only candidates that are popular with the electorate become MCs, rather than candidates who are most loyal to their party and thus would be listed at the top of their party list.

2.2. Disadvantages

While voters may reorder party lists by voting for their preferred candidate within one such list, they ultimately still have to choose between pre-packaged slates of candiates and may only choose one such slate, which means the actual extent of personalisation is limited. For example, voters who prefer some candiates of one slate and some of another slate have no means of expressing this preference.

3. Single Transferable Vote (STV)

STV is a voting system used for elections to the Irish Dáil and the Australian Senate, and makes use of ranked ballots. Voters rank candidates from highest to lowest preference, with 1 being the highest rank. In the tabulation, candidates with the lowest number of 1st rank preferences are eliminated, and their votes are then distributed to their next highest preference if any. Candidates who have reached the quota (the number of votes needed to guarantee a seat) are elected, and votes exceeding the quota are likewise redistributed to their next highest preference. This continues until all seats are filled, or failing that, if the number of non-eliminated candidates is the same as the number of seats left open.

While it is technically possible for all seats to be elected at once using nation-wide STV, this would make voting and tabulation exceedingly difficult. In practice, countries using STV are divided into constituencies, which then each elect a small number of representatives, usually between 3 and 7.

3.1. Advantages

STV has been in use in Ireland since 1922, and in Australia since 1949, which, like with MMP before, proves that it has been proven to work high-stakes elections long term. STV lets voters make a direct statement on which candidates they prefer most, and enables them to freely vote for candidates they like across party boundaries.

3.2. Disadvantages

Again like with MMP, the necessity of constituencies is potentially questionable. Additionally, if each province was its own constituency, they would each only elect 2 to 3 MCs. Such small constituencies are known to result in problematic, unproportional results, so to guarantee proportionality, constituencies would need to cover at least two provinces and elect between 5 and 6 MCs. As such, the exact form of STV constituencies would need to be deliberted and determined beforehand, and potentially changed as populations change per province.

The use of ranked ballots has produced some controversy in the past, so before STV can be enacted, one would have to make sure that voters are well enough informed as to how they work.

4. Sequential Proportional Approval Voting (SPAV)

SPAV is a form of Approval Voting, modified to allow for proportional representation. Voters are given the choice of approve or disapprove any number of candidates on the ballot. All approvals are worth the same. In tabulation, the candidate with the highest number of approvals is elected, and all ballots that approved of this winning candidate are multiplied by some weight in  subsequent rounds (e.g., ballots with 1 winner are worth 1/2, ballots with 2 winners are worth 3, ballots with 3 winners are worth 1/4, etc). This continues until all seats are filled.

4.1. Advantages

SPAV, like STV, allows voters to directly express which candidates they like and which they dislike, and gives voters the ability to "compose their own party lists", since all candidates they approve of would function as one list in the tabulation. The fact that choices are limited to "approve" and "disapprove", rather than having to rank candidates in order, should make this system easier to understand for voters.

4.2. Disadvantages

SPAV is not used for any high-stakes election (it was last used in Sweden from 1909 until 1921), making this system the least thoroughly tested and thus the most "experimental" of the four presented here. Though there has been a lot of research into proportional approval voting methods in academic literature, nations may be wary of implementing such a system without knowing beforehand that it would be effective.

5. Conclusion

PPR is not one system, but a diverse set of methods that all aim to achieve the same goal through very different means. It should then not surprise anyone, that among those who have campaigned in the past for the adoption of PPR for Cosă elections, each person has their own personal favourite (MMP is historically preferred among reformers, the "Kosovo system" is prefered by S:reu Autófil, STV is preferred by M:çă Schivă, and SPAV is my personal favourite). By writing this overview of where we stand right now, my hope is to encourage a national debate on the merits of each system, so that we can decide collectively which path to take as a nation.

Though strictly speaking a distinct and separate issue, the topic of voting reform has historically been combined and conflated with the topic of Senäts reform or abolition. I may write a separate post on one particular way of preserving the Senäts at a later time.

[Miestră Schivă, UrN-GC]

Judging by the debates on the Democracy Amendment in the last Cosa, there appears to be cross-party support for moving to a system where we vote for individuals as well as (even instead of) parties in the Cosa, while preserving proportionality. It is rare that you see @owenedwards and myself agreeing on something.

My personal preferences are either STV or a minimalist system where you vote for a party list + one candidate on that list. Let's get the debate going and try to get something out of this Cosa.

[Miestră Schivă, UrN-GC]

A few more technical thoughts on STV.

The problem of constituencies is not so much a problem with vote counting - there exist several very good online tabulation systems, like the one we use for the Fiova Praisidïeu. The problem is effective threshold for election.

There was a good hard debate on the 20-seat Cosa which means that from now on a party or independent candidate will need at least 2, maybe 3 votes to get a seat. The problem will become harder if you have a constituency which is half or a quarter of the country.

The threshold for election in an STV system is "total votes/number of vacancies+1". So, if the whole country is a single constituency and there are 105 valid votes, the quota for election on the first count would be (105/20+1) = 5 votes. Of course, you wouldn't need 5 first prefs to be elected - you could be elected on latter counts with many fewer votes than that.

(Australian Senate elections are notorious for this. They used to have a "Group Voting Ticket" which meant that instead of ranking every candidate, you could tick a box to indicate your party's "default" ranking. The overwhelming majority of voters used this box - which mean, for example, the Labor Party's top candidate might get 1 million first preferences, and their 2nd and 3rd ranking might get 5000 and 3000 votes respectively, but those latter two would still be elected because almost all the preferences would go straight down the GVT.)

But anyway, that would mean a party/candidate with 4.76% of the vote would be guaranteed representation. Now let's look at calculations I prepared earlier for ways in which you could divide Talossa up into 2 or 4 constituencies. (Turnout figures are those for the 62nd Cosa; I have used provinces as the basis for constituencies because there is simply no other non-arbitrary method.)



The first option divides Talossa into two constituencies:

• KA+VD+FL+BE, covering 61 citizens and electing 9 MCs
• FI+MA+MM+CZ, covering 75 citizens and electing 11 MCs

The quota for election is, respectively, 4.9 votes (10% of the first-pref vote) or 4.6667 (8.3%).

The latter divides Talossa into four constituencies:

• KA+VD, covering 28 citizens and electing 4 MCs: quota is 4.2 votes (20%)
• BE+FF, covering 33 citizens and electing 5 MCs: quota is 4.6667 (16.7%)
• FI+MA, covering 45 citizens and electing 7 MCs: quota is 4 (12.5%)
• CZ+MM, covering 30 citizens and electing 4 MCs: quota is 4.8 (20%)

So, not only does the percentage of the vote you'd need to get a seat climb higher as the constituencies get smaller, but it would vary between constituencies. 2 or 3 seat constituencies, as Marcel mentioned, would be so hard to win in they might as well be first-past-the-post.

I would just like to finally say that Marcel is absolutely right that STV relies on citizens ranking down the ballot, something that especially American voters have a hard time grasping, and also something which is very sensitive to ballot design (I believe that the last Fiova election was more or less inoperative because the ballot did not make rankings obvious or intuitive). One way around this is the Australian method for requiring more than one ranking for a valid vote - in some cases, ranking every candidate. But when this was briefly the case for Seneschal elections, one MC got extremely angry because it was "anti-freedom of speech" to require him to give a (2) ranking to his disfavored candidate. I'd say this is by far a greater difficulty than vote counting.