Evaluative Satisfaction System (Pacifica)
The Evaluative Satisfaction System (ESS) is a mixed electoral system under which representatives in a multi-member body are elected on two tiers via a single round of evaluative voting. Representatives are first chosen in constituencies as the lower tier by using a majority approval vote; the overall composition of the body is then made to approach proportionality on the upper tier using a best-loser algorithm.
ESS is a semi-proportional electoral system, since parties might win more constituency seats in the lower tier than they would be proportionally entitled to in total, and the system does not compensate this by awarding other parties leveling seats to make a purely proportional result on the upper tier possible again. This issue typically affects elections where there are too few upper-tier seats in comparison to constituency seats.
The system was first developed for and used in the 2006 Council of the Union election with the intention to reduce incentives for strategic voting. It has drawn both praise and criticism: Supporters laud that its straightforward ballot layout makes casting one's vote accessible and ballots hard to spoil unintentionally, as well as that the multiple different tallies from each ballot make the election relatively resistent to strategic voting; critics however complain that this very fact also renders elections intransparent by making it hard for voters to precisely see how their vote will influence ‒ and, even after polls close, how it actually did affect ‒ the results.
Procedure
Under ESS, each voter casts a single evaluative voting ballot: Any number of candidates they approve or disapprove of may be marked, with the option to avoid marking a candidate altogether to express indifference. Candidates may be linked to a party list or stand as independents.
Lower tier
In the first step of the election procedure, representatives are determined on the constituency level as the lower-tier seats. Each constituency elects at least one representative, but there is no theoretical limit on the number of representatives a constituency could elect. There isn't a strict need to have multiple constituencies, either ‒ for small-scale elections, having just a single multi-seat constituency is equally viable, but the number of candidates standing can grow quite quickly with an increasing amount of constituency seats, thus limiting the practicability of single-constituency elections.
The exact method used to elect representatives is evaluative voting: Voters mark both all candidates they approve of and all they disapprove of, with the option to abstan from evaluating a candidate altogether. Approvals and disapprovals are tallied up for each candidate, and the candidate(s) with the highest net approval ‒ calculated by subtracting the number of disapprovals they received from the number of approvals received ‒ win(s) the constituency's seat(s).
Upper tier
The second step, electing the upper-tier compensatory seats, works in a slightly more complicated way. First, the parties eligible for receiving compensatory seats need to be determined ‒ the Tepertopian implementation requires that a party's candidates cross a certain approval threshold, for which the lower-tier approvals are considered. If parliamentary fragmentation isn't too large a concern, an implementation could also just consider all parties eligible.
With the eligible parties determined, ballots are re-counted to determine the electorate's so-called satisfaction with the candidates of the eligible parties. Each voter's satisfaction is divided equally between all candidates they approved and who stand for eligible parties ‒ approvals towards independent candidates or candidates standing for ineligible parties are ignored. The satisfaction received by each candidates is then pooled for the associated parties to determine the overall satisfaction shares of the parties. Those relative shares then constitute the share of seats the party is ideally entitled to at-large.
Once the shares stand, the compensatory upper-tier seats are handed out to the parties iteratively: Until all seats are filled, the most-underrepresented party (determined by the difference between its satisfaction share and current total seat share ‒ including those won on the lower tier) sees its least-disapproved candidate elected. Since upper-tier seats rely on unelected lower-tier candidates, parties have an incentive to stand candidates in as many constituencies as possible.
Example
The constituency of Examplington encompasses the entire eponymous city, for which the three-seat municipal council is to be elected using ESS. Two parties (green and yellow) and one independent vie for the votes of the ten voters. Two of the three seats are to be elected on the lower, the remaining one on the upper tier.
The ballots come in as follows:
Ballots Candidate
|
4× | 3× | 2× | 1× |
|---|---|---|---|---|
| . A | ✓ | |||
| . B | ✓ | ✗ | ✗ | |
| . C | ✗ | ✓ | ✗ | |
| . D | ✓ | ✗ | ✓ | |
| . E | ✓ | ✓ | ||
| Approval = ✓ | Disapproval = ✗ | ||||
Lower tier
According to the results above, the candidate's net approval scores are as follows:
| Candidate | Approvals | Disapprovals | Net Approval |
|---|---|---|---|
| . A | 4 | 0 | +4 |
| . B | 4 | 3 + 2 | -1 |
| . C | 3 | 4 + 1 | -2 |
| . D | 3 + 1 | 2 | +2 |
| . E | 2 + 1 | 0 | +3 |
As candidates A and E have the highest net approval, they are elected as representatives on the lower tier.
Upper tier
For the purposes of demonstration, Examplington qualifies all parties for compensatory seats without any restrictions. The independent E however does not play any role for upper-tier allocation, since only party-affiliated candidates are considered ‒ the two voters whose only positive approval went towards E are consequently completely removed from upper-tier calculations. As such, the satisfaction shares divide as such:
Ballots Candidate
|
4× | 3× | 2× | 1× | Candidate Satisfaction | Party Satisfaction | Party Share |
|---|---|---|---|---|---|---|---|
| . A | 0.5 | 2 | 4 | 50% | |||
| . B | 0.5 | 2 | |||||
| . C | 0.5 | 1.5 | 4 | 50% | |||
| . D | 0.5 | 1 | 2.5 | ||||
| Total: | 8 | 100% | |||||
In terms of party satisfaction, the parties are split evenly ‒ which makes sense, considering that there were four ballots from voters only supporting the green party, three only supporting the yellow party, and one that supported a yellow party candidate (and the ‒ irrelevant ‒ independent). These vote shares determine the further procedure of the compensation algorithm, which runs the following steps once per seat to fill ‒ in this case, only a single time, since only one seat remains to be filled on the council: First, the current share of seats each party has is determined, and then, the party with the greater difference between entitled and current seat share is awarded the seat.
| Party | Current Seats | Current Share | Entitled Share | Difference |
|---|---|---|---|---|
| Green | 1 | 1/3 = 33% | 50% | 12% |
| Yellow | 0 | 0/3 = 0% | 50% | 50% |
As the yellow party is more under-represented than the green party, it is awarded the seat. Within the party, the seat goes to the least-disapproved candidate from among the constituency candidates ‒ in this case, candidate C from the yellow party received 5 disapprovals for a total disapproval rate of 50%, and candidate D received 2 for a rate of 20%. As such, the final seat is filled by candidate D.
This means that the municipal council of Examplington will have A from the green and D from the yellow party, as well as the independent E.
Differences to Other Voting Systems
MMP
Unlike the Mixed-Member Proportional system, ESS does not natively accomodate for overhang seats ‒ since the system works on an iterative, relative share basis with a fixed number of seats, other parties are actively harmed by a party winning more seats on the lower tier than it would be entitled to on the upper: If a party would be entitled to 50% of the seats, but its competitor already won a total of 60% in constituencies, a 100-seat parliament would be split 60-40 in seats under the ESS, while MMP would still award the party its absolute 50 seats for a final 60-50 seat split. Some versions of MMP would furthermore fully compensate overhang seats with leveling seats for the other parties, which would result in a fully proportional 60-60 seat split in the former example. Effectively, the ESS sacrifices this full proportionality in order to avoid ever-growing seat numbers stemming from overhang and leveling seats.
Furthermore, most MMP implementations separate the candidate and the party vote, allowing voters to support local candidates without giving votes to the associated party (split-ticket voting), whereas ESS only exposes the candidates to voter approval and automatically converts their performance to party support to avoid decoy lists. However, ESS gives voters more options for influencing the order in which compensatory seats are awarded to candidates by using their disapproval rate, whereas MMP often features a closed list, with the political parties fixing the order in which they want to have candidates elected in advance.
AMS
ESS is very similar to the Additional-Member System, which works the exact same on the overhang seat issue. Strictly speaking, ESS could be considered a variant of the AMS adapted to the use of Evaluative Voting. However, AMS too often features split-ticket voting and closed party lists.
MMM
Mixed-Member Majoritarian systems differ from ESS in the way the proportional seats are handed out, since under MMM, both constituency and proportional seats are awarded completely separately: A party may win every single constituency seat and still receive its share of the proportional seats. ESS on the other hand considers the totality of the seats to be elected for its proportional component; accordingly, the seats a party has won on the constituency level do play a role in the allocation of compensatory ("top-up") seats.