In June 2023, the AfD won the district administration of Sonneberg. Between the first round and the runoff, the AfD’s vote share rose from 46.7 to 52.8 per cent, a gain of 6.1 percentage points. Its CDU opponent went from 35.7 to 47.2 per cent, gaining 11.5 points. The AfD won, but its eleven-point lead almost halved. Requiring it to compete for support beyond its initial electorate made victory harder. The episode recommends a form of indirect militant democracy: electoral rules that make limited appeal beyond a party’s committed supporters a disadvantage. Single-member ranked-choice voting (RCV, also known as instant-runoff, preferential or alternative voting) gives later preferences a role in deciding who wins. Where an exclusionary party attracts relatively few of them, it must compensate with stronger first-choice support. This gives us a reason to prefer RCV to proportional representation for the chamber that determines who governs, while leaving every party free to compete.

Beyond First Preferences

The mechanism matters for democratic defence where exclusionary parties are frequently a voter’s first or last choice, but rarely an intermediate one. Where they lack a majority of first preferences, their difficulty attracting further support puts them at a disadvantage under single-member majority systems such as RCV or a two-candidate runoff. The pattern can be seen in the AfD’s own record in runoff elections. In Nordhausen’s mayoral election in September 2023, the AfD led with 42.1 per cent against the opponent’s 23.7. In the runoff, it gained three percentage points; its opponent gained 31.2 and won. In Saale-Orla in January 2024, the AfD’s 45.7 per cent grew to 47.6, while the CDU’s 33.3 became a winning 52.4.

These are not merely spectacular reversals selected because the AfD lost. Comparing the official returns for all nine Thuringian district-head contests in which the AfD reached the June 2024 runoffs reveals the same asymmetry in every contest. Its opponents gained more vote share. The unweighted average increase was 8.4 percentage points for the AfD, against 20.4 for its opponents. Admittedly, a runoff at a later date between the top two candidates is not the same as an instant runoff, where voters rank their candidates on the same ballot on the same day. Thus, these figures measure gains between elections, not identifiable transfers between candidates. Turnout and campaigning between the two elections would have changed; in Sonneberg itself, turnout rose from 49.1 to 59.6 per cent. In all nine June 2024 district-head runoffs examined here, it fell. Nor does a two-round election replicate the successive eliminations of ranked-choice voting. But the recurring difficulty of expanding beyond an initial base is nonetheless relevant to the choice of electoral system, whether or not it changes the winner.

Preference-ranking research provides independent support. In Anna-Sophie Kurella’s survey experiment following Germany’s 2025 federal election, the AfD was frequently ranked first or last, but rarely in intermediate positions. Participants ranked parties under a Borda rule, which awards points to preferences, rather than an alternative-vote count. Its simulated results cannot be presented as an RCV forecast. The underlying ranking pattern nevertheless matters: strong first-choice support can coexist with weak appeal as an alternative. Australian experience supplies a closer comparison. Benjamin Reilly identifies the factional One Nation Party’s difficulty attracting later preferences, including at the 2017 Queensland and Western Australian elections, as a constraint on converting its first-choice support into lower-house seats.

The Limits of Party Bans

Germany’s power to prohibit parties under Article 21 of the Basic Law is an extraordinary instrument of democratic self-defence. The question whether its legal conditions are met is distinct from whether prohibition is the wisest response, or whether the goals of militant democracy can be realised indirectly through the appropriate electoral system.

Tom Daly and Brian Jones explain why contemporary threats strain the familiar repertoire of party regulation. An organisation’s democratic credentials may be ambiguous; an established party may be captured by an authoritarian leadership. Prohibition leaves the sources of electoral support intact, while a movement may return under another name. Judicial intervention can also be counterproductive because it could strengthen a party’s claim to be the victim of entrenched elites.

Electoral reform does not dissolve the social conflicts on which such parties thrive either. Its comparative attraction is that it can change the rewards for exploiting those conflicts, while preserving the freedom to organise and contest elections. A ban may remain justified against an acute threat. But constitutional design should also make these emergencies less likely to arise.

Regulating Competition

In my account of moderated parliamentarism, I argue for a preference, where effective, for second-order regulation of political parties. First-order regulation directly commands or prohibits conduct. Second-order regulation arranges the institutions within which parties compete so that they have reasons to behave differently. The latter is generally more respectful of party autonomy and less dependent on judicial enforcement. Every electoral system already shapes political competition. Retaining the existing rules is as much a regulatory choice as changing them. The concern is not radical disagreement with existing policies or institutions, but parties that deny some people an equal claim to political concern.

Under RCV, voters rank candidates. If nobody has a majority of first preferences, the lowest-polling candidate is eliminated and their ballots transfer to each voter’s next-ranked remaining candidate. The process continues until someone has a majority. Where voters may leave candidates unranked, this is a majority of continuing ballots.

The incentive follows. Candidates who cannot win outright have an incentive to attract later preferences. A polarising party that attracts relatively few second or third choice preferences must compensate with stronger first-choice support. Thus, parties have a reason to ask voters to tolerate them even when they cannot persuade those voters to put them first. For an exclusionary faction, cultivating that tolerance may require changing more than its campaign language. Voters outside its core become people whose support it has reason to seek.

The relevant willingness already exists. Sven Hillen and Jessica Kuhlmann find that voters strategically crossed party lines to support viable non-AfD constituency candidates in competitive races in 2025. RCV would allow a voter to express that fallback choice without first abandoning her favourite and correctly guessing the strongest challenger. Parties could remain distinct while their voters combined behind an acceptable alternative. Smaller parties could also seek policy concessions in exchange for recommending later preferences. More of the work of building consent would occur before the election, and voters could accept or reject the proposed agreements themselves.

Isn’t PR Fairer?

The objection from proportional representation (PR) is serious. Why should a party’s seats depend on anything other than its share of first-choice votes? Because a single-choice ballot records only one part of what voters think. A party may be the first choice of a substantial minority and unacceptable to most others. Another may have a similar first-choice share but be widely tolerable. List PR translates those first choices into seats without directly registering the difference. Single-member RCV gives later preferences a role in deciding who wins. It does not measure the intensity of aversion, but it allows voters to act on it.

Furthermore, where thresholds are low, PR can make an exclusionary appeal electorally sustainable without requiring its advocates to broaden their support. In a fragmented chamber, a party with 15 per cent of the vote and a roughly corresponding share of seats may become pivotal to the formation or survival of a government. These incentives can encourage political entrepreneurs to distinguish themselves by adopting more exclusionary positions.

One may object that list PR is not the only proportional system: the single transferable vote (STV) is also based on rankings. It is true that STV gives candidates incentives to attract transfers. But its proportional character still permits a candidate to secure a seat without support beyond a relatively small bloc. In a six-member constituency, the usual Droop quota is approximately one-seventh of valid votes (roughly 14%). A candidate who reaches that quota on first preferences need not attract a single transfer from outside that bloc. Ranking therefore does not impose the breadth-of-support requirement associated with single-member RCV.

Proportional systems can bring these interests together through post-electoral coalition bargaining. But the capacity to form a coalition among party leaders is different from the ability to attract broader support from voters. Refusal by other parties to cooperate can contain an exclusionary faction, but a firewall depends on their continuing willingness to maintain it. A parliamentary foothold can also amplify a faction’s influence over public debate even when it remains outside government. Other parties may respond by adopting some of its positions or by sharpening their opposition to them. Where PR makes that foothold easier to secure, it can exert a centrifugal pressure on the polity.

RCV, on the other hand, would give voters a more direct part in political coalition-making. Parties have an incentive to negotiate pre-electoral preference agreements. Voters can then accept or reject those recommendations through their own rankings, rather than leaving the terms of inter-party cooperation entirely to post-election bargaining. RCV allows voters’ later preferences to determine which candidate secures a majority of continuing ballots. List PR instead allocates seats principally by first-choice support. These arrangements give different weight to breadth and intensity of support; neither is self-evidently fairer.

For Germany, adopting RCV would require reconsidering how seats in the governing chamber are allocated, not merely adding rankings to constituency ballots. Retaining proportional entitlements would largely preserve the existing parliamentary arithmetic. The proposal therefore concerns the constitutional embrace of proportionality in the Bundestag, not simply the design of the ballot. My objection, however, only concerns proportionality in the chamber that makes and unmakes governments. I accept that proportional representation can perform a different and valuable role in a checking chamber. That said, the extent to which the Bundesrat could become a proportional checking chamber, while retaining its federal functions, is a question best left for another day.

General Rules and Partisan Interests

The case for RCV should turn on its particular preference incentives, not on an indiscriminate defence of majoritarianism. First-past-the-post plurality system (FPTP), under which a divided opposition can hand a faction victory on a minority of the vote, is not the alternative proposed here. FPTP can limit an exclusionary party’s representation for a time, but it can also hand such a party a parliamentary majority on little more than 30 per cent of the vote against a divided opposition, as happened in India in 2014.

The rules must be defensible independently of the AfD’s fortunes. Fair constituency boundaries, protection against incumbent manipulation and broad participation in designing reform are essential. A change made to entrench today’s governing parties would reproduce the problem it purported to address. An electoral system serves many functions: selecting representatives fairly, facilitating effective governments, and stabilising democratic politics are all important factors that should determine the choice.

Indirect militant democracy seeks to make exclusionary politics electorally costly while preserving the freedom to organise and compete. Its rules must be justified by the value of broader support, not by the identity of the party they currently disadvantage. Sonneberg shows why the distinction matters: the AfD won, but its difficulty expanding beyond its initial supporters made the contest closer. Electoral design should make that difficulty matter before democratic defence has to depend on a court excluding a party altogether.