Stuff:Votorola/p/mA:Guidance of multi-winner assembly elections

This page describes how multi-winner assembly elections based on party-list proportional representation and closed lists may be augmented by the introduction of continuous open primaries. The details are expected to vary depending on jurisdiction and electoral rules. Variations are described further below for each jurisdiction (currently only Germany). What follows is a description of the idealized, standard practice.

Standard practice

The standard practice begins with a continuous, non-stop primary for the purpose of nominating candidates for the assembly. The primary is open to both independent and party nominees. It is open to all primary toolsets that are able to share voting data via a mirroring inter-network. The continuous results are therefore expected to provide the most accurate available prediction of the membership of the next assembly. They take the form a list of nominees ranked in order of popularity, with the leading nominees being the predicted membership of the next assembly. The entire predicted membership is taken as the primary candidate list (unfinalized) for a network of formal parties known as "open parties". This nomination practice is shown in figure N.

[N] Nomination. An aggregate snapshot of a continuous primary nomination. Each node (disc) represents n voters. The voters are using three different primary toolsets (green, red, blue), each hosted by a different open party (i1, i2, i3). Experienced legislative drafters (shown with texts) receive most of the primary votes. The current leading nominees (from Ah with 30n votes, to Ii with 4n) are taken as both the predicted members of the next assembly, and the tentative candidate list for all three open parties. The assembly has 18a seats. legend

The open parties are formally political parties, but actually just primary toolsets or toolset hosts. They have no political identity, neither individually or collectively. All open parties share more-or-less the same candidate list, which is just the expected membership of the assembly as predicted by each toolset. The predictions are more-or-less identical because they are based on mirrored votes. In a parliamentary system, the open parties also share the same formal party leader, who is just the leading root of the open executive primary. The same leader might be shared with a traditional closed party, as well. The open parties cannot be politically opposed to any of the closed parties. Nevertheless they are technically opposed to the party system as a whole. They are expected to undermine that system and to replace it with an inter-network of primary toolsets. Already the open primary enabled by those toolsets is predicting the membership of the next assembly with an unrivalled accuracy (figure N above). Now the closed parties must nominate their own candidate lists, but their choices are largely restricted to these same candidates who are predicted to win seats.

[CPN] Closed party nomination. Two traditional (closed) parties nominate their candidate lists for the election. Each restricts its list to party supporters of course (red, blue), but otherwise tends to favour those supporters with more votes in the open primary (Ah, Bh, and so forth), giving them higher placements in the list. legend

The open parties too must finalize their candidate list before the election. This they extract from the primary results, yielding all candidates from all parties who are predicted to be elected to the new assembly. The order of candidates, however, is changed in this final list. Those candidates who are already placed on closed party lists, or likely to be placed, are moved to the end of the list. There they are grouped by party, larger parties first. The effect of this reordering is that the supporters of closed parties may now shift their votes to an open party with the assurance that the results will be unchanged. This effect can be seen in the election of figure E, below. Closed party H is to elect 6a candidates (Ah. Bh, Dh, Eh, Gh, Hh), while the open parties among them are to elect 9a candidates (Ai to Ii). If a sufficient number of H voters decide to switch their support to open parties on election day, then the last of H's candidates (Hh) will not be seated by H. Instead Hh will remain on the open party list (not crossed off) and eventually be seated as the final candidate by open party i3. The vote shift by H's supporters therefore has no effect on the election. The implication is that party H itself has no effect; it might as well not exist. The same is true for all the other closed parties. They no longer appear to serve any purpose. Their candidates, members, voters and other supporters are therefore apt to be "siphoned off" by the open parties. Carried far enough, this could eliminate the entire party system. After all, the open parties are not parties at all; they are just primary toolsets dressed up to look like parties.

[E] Election. An aggregate view of an election. The open parties have already nominated their own candidate list, having placed the closed-party candidates last (red and blue) and reordering them by party size. Together the open parties receive 50% of the votes and take 9a of the seats; the closed parties taking the other 9a seats. The assembly is elected as originally predicted in the open primary (figure N). legend

The assurance that votes on election day may shift from closed to open parties without unseating anyone depends on the distribution of seats among the open parties and closed parties. For a closed party H, the number of seats that may shift from its smaller closed rivals to the open parties that are larger than H is limited. It must not exceed the combined size of the open parties that are smaller than H. Otherwise H will unseat candidates from its smaller rivals.

[S] Safe for all vote shifts. This shows the election of figure E (left) and a vote shift scenario (right). Parties are ordered by votes received. Elected candidates are marked 'x'. All of party J's votes have shifted in the right scenario to open party i1, which gains 3 seats (+). Nevertheless J's candidates are seated by the smaller open parties i2 and i3.

H i1 i2 J i3 ---- ---- ---- ---- ---- Ah x Ai x Ai Ai Bh x Bi x Bi Bj x Bi Ci x Ci Ci Dh x Di x Di Di Eh x Ei x Ei Ej x Ei Fi Fi x Fi Gh x Gi Gi x Gi Hh x Hi Hi x Hj x Hi Ii Ii Ii x Ah Ah Ah Bh Bh Bh Dh Dh Dh Eh Eh Eh Gh Gh Gh Hh Hh Hh Bj Bj Bj Ej Ej Ej Hj Hj Hj

i1 H i2 i3 J ---- ---- ---- ---- ---- Ai x Ah x Ai Ai Bi x Bh x Bi Bi Bj - Ci x Ci Ci Di x Dh x Di Di Ei x Eh x Ei Ei Ej - Fi + Fi Fi Gi + Gh x Gi Gi Hi + Hh x Hi Hi Hj - Ii Ii x Ii Ah Ah Ah Bh Bh Bh Dh Dh Dh Eh Eh Eh Gh Gh Gh Hh Hh Hh Bj Bj x Bj Ej Ej x Ej Hj Hj Hj x

Figure S summarizes a worst case scenario of open vote shifting for election E. Here all of party J's votes have shifted to open party i1, which is now larger than party H. In general, a shift like this across a larger party is potentially unsafe. But the results are unaffected in this particular case, because J's candidates are seated by the smaller open parties i2 and i3. Owing to the presence of this buffer of smaller open parties, even the most extreme of vote shifts is safe in this election.

[U] Unsafe for extreme vote shifts. This shows an election in which unsafe shifts are possible (left) and an extreme shift scenario (right). All of party J's and K's votes have shifted to open party i1, which gains 4 seats (+). Although 3 of the closed party candidates are seated by the smaller open parties i2 and i3, the fourth (Ek) is not. Instead party H seats an extra candidate.

i1 H J i2 i3 K ---- ---- ---- ---- ---- ---- Ai x Ah x Ai Ai Bi x Bh x Bj x Bi Bi Ci x Ci Ci Di x Dh x Di Di Ei x Eh x Ej x Ei Ei Ek x Fi x Fi Fi Gi Gh x Gi x Gi Hi Hj x Hi x Hi Ii Ii Ii x Ah Ah Ah Bh Bh Bh Dh Dh Dh Eh Eh Eh Gh Gh Gh Hh Hh Hh Bj Bj Bj Ej Ej Ej Hj Hj Hj Ek Ek Ek

i1 H i2 i3 J K ---- ---- ---- ---- ---- ---- Ai x Ah Ai Ai Bi x Bh x Bi Bi Bj - Ci x Ci Ci Di x Dh x Di Di Ei x Eh x Ei Ei Ej - Ek - Fi x Fi Fi Gi + Gh x Gi Gi Hi + x Hi Hi Hj - Ii + Ii Ii x Ah + Ah Ah Bh Bh Bh Dh Dh Dh Eh Eh Eh Gh Gh Gh Hh Hh Hh Bj Bj x Bj Ej Ej x Ej Hj Hj Hj x Ek Ek Ek

Figure U summarizes a worst case scenario for a different election, one in which unsafe open vote shifts are possible. Here the buffer of smaller open parties (i2 + i3) is insufficient (3 seats) to absorb the vote shift (4 seats) that stretches across closed party H and into the larger open party i1. Consequently Ek is not seated by the open parties as predicted in the primary. That would require that the distribution of open parties be better balanced, or the size of the vote shift reduced.

In Germany

The German parliament (the Bundestag) is elected by the additional member system with closed party lists. A fixed number of seats (S) are reserved for the direct election of district members (Wahlkreisabgeordneten), one for each of the S electoral districts (currently 299). Additional members are then seated from party lists (Landesliste), thus maintaining overall proportionality. This brings the regular size of the assembly to 2S or 598 seats. This number may be lower owing to the election of independent or non-party district members, or higher owing to overhang seats. The elector has two votes on the ballot: one for a district candidate (left in figure FEB) and one for a party list (right).

[FEB] German federal election ballot.

If the SPD candidate in figure FEB (Dr. Bartels) wins the seat for Kiel district, then overall party proportionality is maintained by seating one less candidate from the SPD list than would otherwise be seated. But if the seat were instead won by an independent candidate (if any were on the ballot), then the total number of seats would be reduced by 1. It is therefore possible to reduce the Bundestag to just S seats (299) by electing independent candidates alone. This unprecedented result would eliminate the parties from politics, save the expense of several hundred seats, and simplify the overall electoral process. The open primaries might then be implemented purely as single-winner primaries, one per district. But if this is a good end to work toward, it is not a practical means of achieving that end. The present situation works greatly to the disadvantage of independent candidates. They cannot make use of dual candidacy. They are confronted by powerful parties and strong party candidates who receive top placement on the ballot. So severe are these disadvantages that no independent has been elected to the Bundestag since 1949.

The ideal practice would therefore be a combination of single-winner primaries for the district seats and multi-winner primaries for the additional list seats, both linked by dual candidacy through the open parties. German election law, however, does not allow multiple party nomination as required by the standard multi-winner practice. The standard practice must therefore be modified at the election stage (figure E). The largest open party (most used primary toolset) of each state (Land) becomes the open party for that state. It submits a single list of all candidates who consent to being nominated. Likewise it nominates the leaders of the state's district primaries, the nomination for each going to the highest vote receiver who consents to it. Nominations are offered regardless of political affiliation. This is the crucial point. The open party wants to attract the best candidates of the closed parties. The law forces candidates to segregate on party lines in the final list, but has no effect in the open primary; all candidates (party and independent) are there co-joined and the electoral strength of each is revealed in relation to the others. All but the weakest closed-party candidates will have the option of being elected by accepting the closed-party nomination as usual, or by accepting the open-party nomination instead. The consequences of enabling this option are explored in figure NB.

In scenario (2), candidate Eh accepts the open-party nomination instead of her usual H nomination. She is replaced by Gh (H list), and displaces Fi (i1 list). This has the strange effect of helping the left, while simultaneously hurting the left party (H). It helps the left because 5 left candidates (Ah, Bh, Dh, Gh from the H list, plus Eh from the i1 list) are now predicted to be elected, but it hurts the left party because its primary electoral strength drops as a consequence from 63 to 60n; it now has weaker candidates. Worse still, a former candidate has demonstrated her electoral independence and called into question the purpose of the party. If party H is no longer the means of getting elected on the left, then what purpose does it serve? What is to prevent other candidates from following in the footsteps of Eh and accepting open party nominations?

[NB] Nomination for Berlin seats. A hypothetical open primary and several nomination scenarios (1 to 3). The primary results (0) are predicting that Berlin's 12 list-allocated seats will go to candidates Ah through Fi. Their primary votes are shown at left in multiples of n. The default scenario (1) splits the nominations among the parties (H, i1, J) according to known preferences, with only the independent, non-aligned candidates accepting the open party nomination (i1). Scenario (2) shows candidate Eh instead declining her party nomination in favour of the open party, while (3) shows a larger number of candidates (Bh, Dh and Bj) doing the same. The primary electoral strength of each party is shown at bottom for all three scenarios.

(0) all --- 30 Ah 29 Ai 14 Bh 14 Bi 14 Bj 13 Ci 10 Dh 10 Di 9 Eh 8 Ei 8 Ej 8 Fi --- 6 Gh 5 Gi 4 Hh 4 Hi 4 Hj 4 Li ---

(1) H i1 J --- --- --- Ah Ai Bh Bi Bj Ci Dh Di Eh Ei Ej Fi --- --- --- --- --- --- 63 + 82 + 22 = 167

(2) H i1 J --- --- --- Ah Ai Bh Bi Bj Ci Dh Di - Eh Ei Ej - --- --- --- Gh --- --- --- 60 + 83 + 22 = 165

(3) H i1 J --- --- --- Ah Ai - Bh Bi Bj - Ci - Dh - Eh - Ej - --- --- --- Gh Hh Hj --- --- --- 49 + 94 + 12 = 155

Scenario (3) therefore explores a more extreme shift of nominations. By displacing non-aligned candidates on the i1 list, the left increases its predicted seat count by 50% (from 4 to 6). Not to be outdone, the right also increases its seat count by 50% (from 2 to 3). That is the most they can do. Perhaps it is also the least they can do, as neither is likely to tolerate the other gaining an advantage in seats. But the resulting shift in electoral strength is also extreme. The closed parties (H and J) lose 24n primary votes between them, while the open party gains 12n; the open party takes the cream of the candidates and leaves the closed parties looking hollowed out. This puts some strain on the primary prediction as it concerns the balance among parties. At least some of the primary voters currently supporting candidates Bh, Bj and Dh are likely to follow suit on election day, shifting their votes to the open party. The extremes of this candidate-oriented vote shift are explored in figure FC.

[FC] Following the candidates. Candidates Bh, Bj and Dh have shifted their nominations to the open party in both (3) and (4), but the two scenarios show different election results. Scenario (3) shows the electors using their list votes (right of figure FEB) to elect the predicted seating of the parties, while (4) shows them instead electing the predicted seating of the candidates.

(0) all --- 30 Ah 29 Ai 14 Bh 14 Bi 14 Bj 13 Ci 10 Dh 10 Di 9 Eh 8 Ei 8 Ej 8 Fi --- 6 Gh 5 Gi 4 Hh 4 Hi 4 Hj 4 Li ---

(1) H i1 J --- --- --- Ah Ai Bh Bi Bj Ci Dh Di Eh Ei Ej Fi --- --- --- --- --- --- 63 + 82 + 22 = 167

(3) H i1 J --- --- --- Ah Ai - Bh Bi Bj - Ci - Dh - Eh - Ej - --- --- --- Gh Hh Hj --- --- --- 49 + 94 + 12 = 155

(4) H i1 J --- --- --- Ah Ai - Bh Bi Bj - Ci - Dh Di Eh Ei Ej Fi --- --- --- --- ---- -- 39 + 120 + 8 = 167

If all electors follow the candidates as shown in scenario (4), then the extra seats on the left and right that were sought in (3) are erased. Instead the candidates predicted in the primary (0) are elected. The nomination shifters (Bh, Bj, Dh) benefit from safer placement on the i1 list, with Di, Ei, Fi being elected to the tail. They also benefit by escaping the collapse of the party system. Nothing remains to prevent Ah, Eh and Ej from following them and making the collapse total. A more likely outcome, however, is something in between scenarios (3) and (4). At least some electors continue to vote on party lines, if only from habit. and it will probably take multiple election cycles to break with tradition and effect a total collapse of the party system. The open parties and multi-winner primaries would then be shelved, as they would no longer serve any purpose. The districts would elect all independent candidates. The Bundestag would then automatically be reduced to 299 seats; 299 for the districts, and zero for the parties. The hearts of the parties are instead transplanted to the executive primaries (left and right), whence the strongest is invited to lead the open party (i1) into the election.

Addenda

See also

• Stuff:Votorola/p/guidance of assembly elections - in general

Property settings

• Issue type (Category:Assembly)
• Place (Stuff:Decision guideway)
• Place (Stuff:System gap)
• Purpose (Stuff:Votorola/p)