Sustaining Neutrality in Polarized Conflict - A Generative Relational Account of the Swiss Case

Transcript

Abstract

A neutral party in a polarized conflict withholds the alignment that each side requests, and each side reads that withholding as a favour to the other. Ordinary accounts answer this difficulty by describing neutrality as a position between the parties, as equidistance, or as the suspension of judgment. This paper proposes a different object. Neutrality is treated as the governance of a field of relational conditions: the set of arrangements under which the parties retain the capacity to form, revise, and resume relations with one another and with the neutral party. A viability formulation states the maintenance requirement, namely that the set of arrangements from which some continuation remains available stays non-empty for each party. A reduction makes that requirement decidable without enumerating the field, by replacing the field with the structure of possible contacts that the standing conditions support; the reduction is a lower bound, detecting total foreclosure rather than narrowing. Two results follow from that formulation. A modality constraint separates the influence a neutral party may exercise from the instruments through which it may exercise it: judgment, testimony, and public argument remain available, while control over access, hosting, procedure, and channel may not be used to secure a substantive outcome. An accumulation asymmetry separates a declared position, whose abandonment costs other parties nothing, from relational infrastructure, whose accumulation places a stake in the hands of others and thereby raises the cost of reversal. The asymmetry is stated as one source of durability among others rather than as a prediction of which neutralities survive, and the Austrian case is the reason for the weaker form. The Swiss case supplies the illustration. International recognition in 1815, the separation of the law of neutrality from the policy of neutrality, the concentration of institutional hosting in Geneva, armed neutrality, and the sanctions and re-export decisions of 2022 through 2024 are read as operations on a field of conditions rather than as adjustments of a position. Failure receives a two-part treatment. Mechanism failure describes the loss of the capacity itself through capture, unpredictability, or loss of relational utility. Normative defeat describes the separate case in which the capacity survives and its maintenance becomes the instrument through which a party loses the standing to be a party. The wartime record of Switzerland and the International Committee of the Red Cross tests both. The account is a revisable conceptual proposal. Its empirical use requires comparative cases, independent measures of relational infrastructure, and the separate evaluation of accuracy, distribution, contestability, and accountability.

Keywords: neutrality; mediation; relational conditions; viability; good offices; Switzerland; humanitarian principles.

Discussion Paper Note

This paper is a preliminary discussion paper intended to share an evolving idea and invite further dialogue, criticism, revision, and independent development. Its definitions, distinctions, and constructions remain provisional. Circulation across scholarly and practical communities is part of the purpose of releasing the manuscript at this stage.

The author treats the viewpoints, concepts, and lines of reasoning presented here as contributions to a shared field of inquiry. Similar or related ideas may have appeared in other intellectual, cultural, and disciplinary traditions. The manuscript therefore states its known antecedents, separates the researcher-origin proposal from later formal reconstruction, and leaves historical priority open pending a systematic originality review.

The arguments should be understood as provisional and historically situated. Readers are encouraged to question, test, revise, extend, reinterpret, or independently develop the ideas presented here. Where appropriate, acknowledgment of this paper as one point of encounter in the development of a related idea is appreciated. Such acknowledgment records an intellectual route; the ideas themselves remain available for criticism, revision, and independent development.

Responsible Use and Rights Reservation

This section separates requested scholarly conduct from the legal permissions stated on the following page. It records an ethical request for responsible use and then defines the narrower scope of retained legal rights.

The author encourages good-faith discussion, criticism, independent inquiry, and responsible use of the material in this work. Separately from the licence’s terms, the author asks users to consider foreseeable harms when adapting or applying the proposed framework. The account developed here concerns the conditions under which parties in conflict retain the capacity to relate, and its misapplication could support the defence of arrangements that entrench harm. This ethical request leaves the licence’s permissions and legally authorized uses unchanged.

The author retains the rights preserved under CC BY-NC 4.0 and may pursue remedies to which the author is legally entitled for breach of the licence or violation of the author’s independently applicable rights. Reuse remains independent from authorial endorsement. Third-party rights require authorization from their respective holders where applicable. Copyright exceptions and limitations, including applicable forms of fair use or fair dealing, remain fully available.

Notices

This page consolidates the manuscript’s publication status, licence, development disclosure, research-programme relation, and suggested citation.

Status.

This working draft records an evolving stage of the author’s position and is circulated for discussion. Definitions, section structure, statements, and numbering remain subject to revision. A systematic review of the political philosophy of neutrality, an expanded comparative case dossier, archival verification of the wartime record, specialist review of the viability formulation, and an originality audit remain future research stages.

Licence.

Except where otherwise indicated, copyright 2026 Wanhong Huang. This work is made available under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). Subject to its terms, the licence permits sharing and adaptation for noncommercial purposes with appropriate attribution, a link to the licence, an indication of changes, and attribution that preserves the licensor’s independence from the reuse. Reuse is governed solely by that licence; the responsible-use request on the preceding page remains separate from its terms. The licence deed and legal-code link are available at https://creativecommons.org/licenses/by-nc/4.0/. The licence governs in case of conflict with this summary. Third-party material remains subject to the rights held by its respective rights holders.

Statement on the use of language models.

The exploratory discussions and preparation of this paper involved Anthropic’s Claude. The model supported exploratory dialogue, source discovery followed by verification against publisher, journal, governmental, and institutional pages, argumentative criticism, and drafting in LaTeX. The author selected the research question, directed and approved the theoretical commitments and the epistemic status of the claims, and bears sole responsibility for the manuscript, including its definitions, constructions, taxonomy, arguments, conclusions, and errors. Authorship credit remains with the human author. The access level and claim limit for every cited source are recorded in the accompanying literature audit.

Related research programme.

This paper is project P001 and the first paper in a series on trust, neutrality, and the transmission of shared experience. Later projects in the series take responsibility for individual credible commitment and self-legislation, for the institutional analysis of recommendation, for the jurisprudence of co-experience, and for the governance of conversion between heterogeneous value systems. The present paper takes responsibility for the relational-conditions account of neutrality, the modality constraint, the accumulation asymmetry, and the two-part failure analysis.

Suggested citation.

Huang, Wanhong. “Sustaining Neutrality in Polarized Conflict: A Generative Relational Account of the Swiss Case.” Working discussion paper, 2026.

Introduction

A party that declines to align with either side of a conflict is rarely received as having declined anything. Each side reads the refusal as a concession to the other, and both extend the invitation again. The difficulty compounds under polarization, where the space of recognized positions narrows until abstention itself becomes legible as a position within the dispute. The neutral party then faces a demand that renews with every development, and each answer it gives is scored on a single axis it did not choose.

The difficulty is ordinarily addressed by giving neutrality a location. On this reading a neutral party stands between the sides, distributes its conduct evenly, and declines to judge the merits. The reading is intuitive and it fails under pressure in three ways. Even conduct across unequal parties reproduces the inequality, so that evenness delivers a substantive result while claiming to deliver none. A refusal to judge removes the neutral party from the moral community whose recognition it requires. And a location between two poles is defined by the poles, so that a neutral party becomes a function of the conflict it proposed to stand outside.

This paper proposes a different object of neutrality. What a neutral party maintains is neither a location nor an even distribution of conduct but a set of conditions: the arrangements under which the parties retain the capacity to form, revise, suspend, and resume relations with one another and with the neutral party itself. The object is a field of relational possibility rather than any relation within it. On this account a neutral party may hold a view, may treat unequally placed parties differently, and occupies no midpoint. What it must do is refrain from using the instruments that constitute the field in order to determine what happens inside it.

Two consequences organize the paper. The first is a constraint on the modality of influence. A neutral party may judge, testify, publish, argue, and persuade, and may do so asymmetrically, because these operate inside the field and are answerable within it. What the neutral party may not do is condition access to the field on alignment: the powers that constitute the field, including access, hosting, procedure, channel, and agenda, are withheld from the pursuit of substantive outcomes. The constraint separates the account from equidistance without collapsing it into partisanship, and it classifies rather than excuses the difficult passages of the historical record.

The second consequence concerns time. A declared position can be adopted and abandoned at low cost, and it accumulates nothing. Relational infrastructure – hosting capacity, standing mandates, institutional residency, an operating record that other parties have reason to preserve – accumulates, and its accumulation is what supplies the reinforcement that sustains neutral capacity across successive conflicts. The distinction predicts that non-alignment maintained as a stance is reversible under changed security conditions, while neutrality carried by infrastructure is costly for others as well as for its holder to dismantle.

Switzerland supplies the illustration. The choice is not a claim that Swiss neutrality is exemplary or that its record is defensible throughout. It is chosen because the case exhibits the mechanism at unusual length and because its most damaging passages are documented by the responsible institutions themselves. The paper reads the international recognition of 1815, the separation of the law of neutrality from the policy of neutrality, the concentration of institutional hosting in Geneva, armed neutrality, and the sanctions and re-export decisions of 2022 through 2024 as operations on a field of conditions. It then treats failure in two parts. Mechanism failure describes the loss of the capacity through capture, unpredictability, or loss of relational utility. Normative defeat describes the separate case in which the capacity survives and its maintenance becomes the instrument through which a party loses the standing to be a party at all. The wartime record of Switzerland and of the International Committee of the Red Cross tests both, and the account is written so that the test can fail.

Section 2 supplies the legal, historical, and conceptual material the argument presupposes. Section 3 locates the account among existing treatments of neutrality in mediation research, political philosophy, mechanism design, relational ontology, and the sociology of the third party, and states what each of them already establishes. Section 4 states the method, the grounds for case selection, and the conditions under which the account would be disconfirmed. Section 5 develops the account itself. Section 6 gives the mechanism by which neutral capacity is sustained. Section 7 applies both to the Swiss case. Section 8 develops the two-part failure analysis. Section 9 states what the case returns to the wider generative relational framework. Section 10 records the limits of the account and Section 11 consolidates the position.

Background and Preliminaries

This section supplies the material the argument presupposes. It states what the law of neutrality requires and where it stops, outlines the Swiss institution at the level of established fact, and defines the terms of the generative relational vocabulary used in the sections that follow. Analytic claims about these materials are reserved for Section 7 and Section 8.

Neutrality in International Law and State Practice

The law of neutrality in armed conflict rests principally on two instruments adopted at The Hague in 1907 and in force from 1910. Convention V governs the rights and duties of neutral powers and persons in war on land (“Convention (V) Respecting the Rights and Duties of Neutral Powers and Persons in Case of War on Land” 1907); Convention XIII governs the corresponding rights and duties in naval war (“Convention (XIII) Respecting the Rights and Duties of Neutral Powers in Naval War” 1907). The obligations they impose on a neutral state are narrow and largely negative in form: the territory of a neutral power is inviolable, belligerents may not move troops or convoys of war materiel across it, and a neutral government may not itself supply warships, munitions, or war materiel to a belligerent. Two features of this framework matter for what follows.

The first is the extent of what the law leaves unregulated. Convention V provides that a neutral power is not bound to prevent the export or transit, on behalf of belligerents, of arms or of anything that can be of use to an army or fleet (“Convention (V) Respecting the Rights and Duties of Neutral Powers and Persons in Case of War on Land” 1907). The prohibition falls on the neutral state as an actor, not on private commerce within its jurisdiction. Conduct that many readers would treat as the substance of neutrality therefore lies outside the law of neutrality and inside the discretion of the neutral state.

The second is the status of the instruments themselves. The treaty record indicates that several states did not ratify the two neutrality conventions, which are nonetheless understood to have been to a large extent declaratory of existing international law (“Convention (V) Respecting the Rights and Duties of Neutral Powers and Persons in Case of War on Land” 1907). The law of neutrality is thus better read as a floor of prohibitions with contested formal reach than as a complete specification of what a neutral state does.

A third feature belongs here rather than in the analysis, because it is a statement of the law rather than a claim of this paper. Authoritative commentary on the law of neutrality holds that a state may be neutral, in the sense of not participating in hostilities, without being impartial in its attitude toward the belligerents; the essential condition of neutral status is non-participation, and whether a position of non-participation can be sustained in the absence of impartiality depends on the reaction of the aggrieved belligerent (“The Law of Neutrality,” n.d.). The same commentary records that a neutral’s use of force to resist attempts to violate its neutrality does not constitute participation in the hostilities (“The Law of Neutrality,” n.d.; “Convention (XIII) Respecting the Rights and Duties of Neutral Powers in Naval War” 1907). Both points are used in Section 7.

The Swiss administration draws the corresponding distinction explicitly. The law of neutrality (Neutralitätsrecht) comprises the obligations described above; the policy of neutrality (Neutralitätspolitik) comprises the discretionary measures a neutral state adopts to render its neutrality credible and effective, and it is adjustable as circumstances change (Federal Department of Foreign Affairs 2026). Sanctions, arms-export rules, good offices, and institutional hosting belong to the second category rather than the first. The distinction is doctrine rather than discovery, and Section 7 treats it as an object of analysis rather than as an authority.

Formation, Instruments, and Recent Decisions of Swiss Neutrality

Swiss neutrality is frequently dated to the battle of Marignano in 1515 and to the Peace of Westphalia, and Swiss official material reproduces this genealogy while noting that formal international recognition followed only at the Congress of Vienna in 1815 (Federal Department of Foreign Affairs 2026). The earlier dating is contested. Historians have noted that the cantons were not uniformly represented at Marignano and that neutrality was in significant part imposed by stronger neighbouring powers rather than adopted as a deliberate choice (swissinfo.ch, n.d.b); the convention of dating neutrality to 1515 is itself a product of twentieth-century historiography (E-International Relations 2024). This paper accordingly treats 1515 as a founding narrative and 1815 as the date from which permanent neutrality carried international recognition.

Four features of the institution recur in the argument. Armed neutrality couples the refusal of alignment to an independent defensive capacity, so that the refusal is enforceable rather than merely declared. Good offices and protecting-power mandates place Switzerland in the position of representing one state’s interests in another with which it has no diplomatic relations. The concentration of international institutions in Geneva supplies a standing hosting capacity that other states use and therefore have reason to preserve. And a federal structure that spans Catholic and Protestant cantons and German, French, Italian, and Romansh linguistic communities makes external alignment a question with internal consequences.

The contemporary record supplies the material for Section 7.5. In February 2022 Switzerland adopted the European Union’s sanctions package against Russia (Euronews 2022b). In the same period it declined requests from Germany and other states to permit the re-export of Swiss-manufactured war materiel to Ukraine, on the stated ground of equal treatment under the law of neutrality (Euronews 2022a). In August 2022 Russia declined Switzerland’s offer to represent Ukrainian interests in Moscow, stating that Switzerland had lost its status as a neutral state by joining Western sanctions (Al Jazeera 2022). Domestic opinion during this period supported neutrality in principle at a high level while a majority simultaneously favoured closer cooperation with NATO and regarded the sanctions as compatible with neutrality (Center for Security Studies, ETH Zurich and Military Academy at ETH Zurich 2023); a smaller majority favoured permitting other states to transfer Swiss-manufactured weapons to Ukraine (swissinfo.ch 2023). A Swiss expert commission subsequently recommended revisions to neutrality policy, including re-export permissions for a defined set of partner states and closer cooperation with NATO and the European Union (European Council on Foreign Relations 2024), and the government has continued to distinguish measures compatible with the law of neutrality from those it regards as excluded (swissinfo.ch 2024).

Vocabulary of Relations, Conditions, and Fields

The account uses a small vocabulary, stated here and applied thereafter.

A relation is treated as an ongoing process between parties rather than as a state that obtains or fails to obtain at a moment. A relation is described by what it produces: shared undertakings, revisable understandings, and the capacity of each party to raise a claim the other must answer.

The generativity of a relation is its capacity to continue producing such outcomes, including outcomes that no party can specify in advance. Generativity is a capacity of the relation rather than a property of either party, and it can be reduced by conditions external to both.

A relational condition is an arrangement whose presence or absence changes what relations can be formed or continued, without determining which ones are. Communication channels, meeting venues, procedural rules, standing mandates, and the availability of an intermediary are conditions in this sense.

The field of a set of parties is the set of arrangements consistent with those conditions. Condition governance is action on the field rather than on the arrangements within it.

Foreclosure is the removal of an arrangement from the field. Foreclosure is irreversible when no available sequence of actions restores it. Irreversibility is a property of the field under a stated horizon and not an absolute one, and the horizon is stated wherever the term is used.

The terms above belong to a wider framework of generative relational analysis developed by the author across a connected series of papers. The present paper uses only the definitions above and leaves the remainder of that framework aside. Its relation to social-constructionist relational ontology, and in particular to Gergen’s account of relational being (Gergen 2009), is addressed in Section 3.6.

Literature Review

This section locates the account among existing treatments of neutrality and states what each of them already establishes. Its purpose is to fix the boundary of the present contribution before the account is developed. Three of the traditions surveyed here anticipate parts of the argument closely enough that the paper proceeds by concession rather than by dispute.

Equidistance and the Arithmetic Conception of Neutrality

The most widely held conception treats a neutral party as one that divides its conduct evenly between the sides. On this reading neutrality is an arithmetic property: equal contact, equal concession, equal criticism, and a resulting position at the midpoint of the dispute. The conception is not the product of a particular scholarly tradition; it is the ordinary understanding against which specialist accounts define themselves, and this paper reconstructs it in that role rather than attributing it to a source.

Two difficulties are decisive. Even conduct toward unequally placed parties transmits the inequality rather than neutralising it, so that arithmetic evenness produces a substantive result while presenting itself as producing none. And a midpoint is defined by its endpoints, so that a party which locates itself between two poles has allowed the conflict to determine its position. Under polarization, where the recognized positions narrow, the second difficulty becomes acute: abstention is absorbed into the axis it declines to enter.

Indifference and Suspended Judgment in Humanitarian Doctrine

A second conception treats neutrality as the withholding of judgment about the merits. Humanitarian doctrine supplies the most developed rejection of this reading. The Fundamental Principles of the Red Cross and Red Crescent Movement separate impartiality, which requires that assistance be governed by need without discrimination, from neutrality, which requires abstention from taking sides in hostilities and from engaging in controversies of a political, racial, religious, or ideological character (Pictet 1979). The two principles are distinct, and the Movement’s own account treats neutrality and independence as practical instruments for making humanity and impartiality operative rather than as ends in themselves (International Committee of the Red Cross 2024).

That framing matters for what follows in two ways. It establishes that neutrality can be defended on operational rather than positional grounds: the justification is access, not equivalence. And the Movement itself has rejected what it describes as the binary of silence against denunciation (International Committee of the Red Cross 2024), which is the position the present account requires and which Section 5.5 states as a general constraint. The critical literature on humanitarian neutrality presses the opposite concern, that operational neutrality shades into acquiescence and that principled abstention can serve the party already advantaged (Slim 1997, 2015). Section 8.2 treats that concern as the substance of normative defeat rather than as an objection to be deflected.

Impartiality and Effectiveness in the Mediation Literature

Research on international mediation has established that impartiality is neither necessary nor sufficient for effective third-party intervention, and this paper concedes the point in full. Touval and Zartman treat the mediator as an actor with interests of its own rather than as a disinterested facilitator, and hold that mediators are seldom indifferent to the terms under negotiation (Touval and Zartman 1985; Zartman and Touval 1996). Kydd supplies the formal result: where the mediator’s role is to transmit credible information, a mediator aligned with the party being counselled is more believable than an unaligned one, because an unaligned mediator has an incentive to counsel restraint irrespective of the facts (Kydd 2003). Svensson’s empirical work extends the finding to outcomes, reporting that biased mediation is associated with stronger institutional guarantees in negotiated settlements (Svensson 2007, 2009, 2015). Zartman’s account of ripeness locates the decisive variable in the parties’ own situation rather than in the mediator’s characteristics (Zartman 1985).

These results are not an obstacle to the present account; they are the reason the account abandons equidistance. What they establish concerns the mediator’s function as a transmitter of information and a supplier of leverage within a negotiation. The present paper concerns a different function, the maintenance over time of the conditions under which a negotiation can be convened at all, and it accepts the burden of showing that the two functions are distinct. Adaptive mediation, which emphasises the facilitation of process and the emergence of outcomes over predetermined templates (Coning, Muto, and Saraiva 2022), stands closer to the present concern, and Section 9 states the division of labour between them.

Neutrality of Aim and Neutrality of Effect in Political Liberalism

The closest philosophical precedent lies in liberal political theory. Rawls distinguishes neutrality of aim, under which the state does not intend to favour any particular comprehensive doctrine, from neutrality of effect or influence, which he regards as impracticable and does not endorse; he also records that the term neutrality carries misleading connotations (Rawls 1993). The distinction anticipates the present account’s separation of what a custodian intends from what its conduct produces, and the paper concedes the priority.

Raz supplies the principal critique, arguing that comprehensive neutrality is unattainable, that neutrality and fairness are distinct ideals which can diverge, and that neutrality admits of degrees (Raz 1986). Larmore defends a procedural conception grounded in a norm of rational dialogue (Larmore 1987); Dworkin grounds neutrality in equal concern and respect (Dworkin 1985); Waldron examines the neutrality of legislation (Waldron 1993); and Wall develops a restricted state neutrality compatible with perfectionist commitments (Wall 2010).

The difference the present paper claims is in the object governed. Liberal neutrality restricts the justifications a state may offer for its arrangements. The account developed here restricts the instruments a custodian may employ, and it applies to actors that offer no public justification at all. Whether that difference survives Raz’s argument is addressed in Section 5.5.

Credible Neutrality in Mechanism Design

The nearest existing formulation of neutrality as a property of a field rather than of a position comes from mechanism design in distributed systems. Buterin proposes that a mechanism is credibly neutral when its design can be inspected and seen not to discriminate for or against specific people, and derives from this a set of construction rules of which the first excludes writing specific people or specific outcomes into the mechanism (Buterin 2020). The formulation is not developed within academic political theory, and its uptake has been concentrated in network governance rather than in the study of conflict; it nonetheless states the core intuition of the present account and is conceded as an antecedent.

Two differences are claimed. Credible neutrality is a property of a designed artefact, assessed by inspection at a point in time, whereas the present account concerns a capacity maintained by an actor across successive conflicts under changing conditions. And credible neutrality secures non-discrimination among participants already inside the mechanism, whereas the present account concerns whether the parties retain the capacity to enter into relation at all. The second difference is the substantive one, and Section 5.1 states it.

Relational Ontologies, Third-Party Position, and Condition Governance

Three further traditions supply the account’s conceptual furniture.

Relational ontology in the social sciences treats relationships as prior to the parties they connect. Gergen’s account describes the self as the intersecting point at which many relations meet, and develops the consequences of that priority for identity, morality, and practice (Gergen 2009). The present framework shares the priority claim and takes a different object: Gergen theorises selfhood and its social constitution, whereas this paper theorises the governance of the conditions under which relations among parties in conflict continue to be generated. The vocabularies are adjacent and the questions are distinct, and the paper states the difference rather than allowing the resemblance of terms to imply derivation.

The sociology of the third party supplies the structural analysis. Simmel distinguishes three positions available to a third element in a triad: the non-partisan who mediates, the tertius gaudens who profits from the separation of the other two, and the party that divides in order to rule (Simmel 1950). Burt’s account of structural holes gives the modern formalisation, in which brokerage across disconnected contacts confers informational and control advantages (Burt 1992); subsequent work in the same tradition separates brokerage that sustains open connection between the other parties from brokerage that maintains conflict between them (“Divide Et Impera: Conflictual Ties in Broker Relationships” 2016). That separation is precisely the distinction the present account requires, and Section 6.1 uses it to distinguish a custodian whose advantage depends on the parties’ capacity to connect from a broker whose advantage depends on their remaining apart.

Condition governance has its clearest precedent in Ostrom’s analysis of common-pool resources, which locates durable arrangements in institutions that sustain the conditions of continued use rather than in rules that prescribe particular outcomes (Ostrom 1990). The present account claims Ostrom as a forerunner and transposes the object from a physical resource to a field of relational conditions.

The literature on trust between adversaries is treated in Section 6, where it bears on the mechanism rather than on the concept.

Boundary of the Present Contribution

Table [tab:antecedents] records the role licensed by each tradition and the boundary of the present contribution.

@P0.24YY@ Tradition & Licensed role & P001 boundary
Mediation research & Mediator interest, bias, credibility, leverage, and ripeness (Touval and Zartman 1985; Kydd 2003; Svensson 2009; Zartman 1985) & Concerns intervention within a convened negotiation; the maintenance of the conditions of convening is treated separately here.
Liberal neutrality & Neutrality of aim against neutrality of effect, and its critique (Rawls 1993; Raz 1986; Larmore 1987) & Restricts justifications; the present constraint restricts instruments and applies to actors offering no public justification.
Humanitarian doctrine & The separation of neutrality from impartiality and the operational defence of both (Pictet 1979; International Committee of the Red Cross 2024; Slim 1997) & Supplies the operational framing; the general constraint and the failure analysis are present-paper constructions.
Credible neutrality & Neutrality as an inspectable property of a mechanism (Buterin 2020) & Static non-discrimination among participants; the present account concerns capacity maintained across successive conflicts.
Relational ontology & Priority of relationship over the parties related (Gergen 2009) & Theorises selfhood; the present object is the governance of relational conditions under conflict.
Third-party sociology & Positions available to a third element and the advantages of brokerage (Simmel 1950; Burt 1992; “Divide Et Impera: Conflictual Ties in Broker Relationships” 2016) & Supplies the connect-against-divide distinction; the custodial position is developed here.
Commons governance & Institutions that sustain conditions rather than prescribe outcomes (Ostrom 1990) & Claimed as forerunner; transposition to a relational field is the present contribution.

What the surveyed traditions leave unoccupied is a single position. No treatment located here specifies the object of neutrality as a field of relational conditions, states a constraint distinguishing the instruments a custodian may use from the influence it may exercise, or separates the resources that make neutral capacity self-reinforcing from those that do not accumulate. The claim is that this combination is unoccupied, not that any of its components is unprecedented; the components are conceded above.

Two terminological cautions follow from a search for the terms used here. The phrase relational infrastructure is already in use in adjacent literatures, where it denotes the arrangements through which heterogeneous resources and communities are produced, organised, and connected (“Infrastructuring Digital Humanities: On Relational Infrastructure and Global Reconfiguration of the Field” 2022). The present paper uses it in a narrower and different sense, for the accumulated stake that other parties hold in a custodian’s standing arrangements, and no continuity with that literature is claimed. And the search returned no scholarly use of generative neutrality, neutral custodianship, or a field of relational possibility in a sense related to the present one; that is a negative result of a bounded search rather than an established absence, and a systematic originality audit remains outstanding and is recorded in Section 10.

Method and Case Selection

This section states what kind of work the case is asked to do, why Switzerland was selected, and what observations would count against the account.

Theory-Building Case Study

The paper develops a conceptual account and uses a single case to exhibit it. The case is therefore illustrative and theory-building rather than confirmatory. A single case can show that a mechanism is instantiated, can supply the vocabulary in which its parts are described, and can expose passages where the account fits badly. It cannot establish frequency, cannot identify causal effect, and cannot rule out rival mechanisms that would produce the same observable record.

Three consequences follow for how the case is written. Established fact is separated from analysis: Section 2.2 records what happened and Section 7 records what the account makes of it. Contested matters are marked as contested rather than resolved in the account’s favour, and the dating of Swiss neutrality and the explanation of non-invasion during the Second World War are treated in this way. And the passages that fit the account badly are given more space than those that fit it well, which is the purpose of Section 8.

Selection of the Swiss Case

Switzerland was selected on four grounds, none of which is that its record is admirable.

The case runs long. International recognition dates from 1815 (Federal Department of Foreign Affairs 2026), which supplies a span across changing security environments rather than a single episode, and the accumulation argument of Section 6.3 requires such a span.

The case is documented by the institutions responsible for it. The wartime record was examined by a commission established by the Swiss parliament (Independent Commission of Experts Switzerland–Second World War 2002) and by a governmental study in another jurisdiction (Eizenstat 1998), and the humanitarian institution most closely associated with the case has published its own critical assessment of its wartime conduct (International Committee of the Red Cross 2024). Self-documentation of this kind is unusual and it permits the difficult passages to be treated from sources the case’s defenders cannot dismiss.

The case is live. The decisions of 2022 through 2024 concerning sanctions, the re-export of war materiel, and a protecting-power mandate (Euronews 2022b, 2022a; Al Jazeera 2022) occurred within a polarized conflict and are recent enough that their consequences remain visible.

The case contains its own counter-evidence. The wartime record supplies a documented instance in which the maintenance of the position coincided with conduct the responsible institutions have since repudiated, and Section 8.3 treats that instance as the account’s principal test.

The selection is nonetheless subject to survivorship. Switzerland is the neutrality that persisted, and an account built on it risks describing the characteristics of survival rather than the mechanism of neutrality. Section 8.4 accordingly examines cases in which neutrality was violated or abandoned, and the account is disconfirmed if the surviving and non-surviving cases do not differ along the dimension it identifies.

Conditions of Disconfirmation

The account should be narrowed or withdrawn under any of the following conditions.

First, if the field of relational conditions cannot be specified independently of the outcomes it is supposed to leave open, the central object collapses into a restatement of the outcomes and the account has no content.

Second, if the modality constraint of Section 5.5 cannot classify contested historical conduct without amendment introduced for each case, the constraint is a label rather than a criterion.

Third, if surviving and abandoned neutralities do not differ in accumulated relational infrastructure, the accumulation asymmetry of Section 6.3 is false.

Fourth, if the viability formulation of Section 5.3 supports no inference beyond a restatement of the maintenance requirement, it should be removed rather than retained for its formal appearance.

Fifth, if the record is better explained by a rival mechanism, and in particular by the material usefulness of a neutral party to the stronger belligerent, then that mechanism rather than the present one accounts for the case. Section 8.3 takes this rival seriously because the wartime evidence supports it in part.

A Generative Relational Account of Neutrality

This section develops the account. It specifies the object a neutral party maintains, states the maintenance requirement, distinguishes the symmetry the account requires from the symmetry it rejects, and derives the constraint that separates the influence a neutral party may exercise from the instruments through which it may exercise it.

The Field of Relational Possibility

Let a set of parties be given, together with the arrangements those parties might occupy with respect to one another: channels through which they communicate, venues in which they meet, procedures under which they raise and answer claims, undertakings they have made, and intermediaries through which they act when they do not act directly. Some of these arrangements obtain and others remain merely available. The set of those that could obtain, given the conditions currently in place, is the object with which this paper is concerned.

Definition 1 (Relational field). Fix a set of parties, a set of relational conditions, and a horizon. The relational field is the set of arrangements among those parties that remain available under those conditions within that horizon.

The field is neither a relation nor a set of relations that obtain. It is the set of relations that could be formed, revised, suspended, or resumed. Two parties in open hostility may occupy a field that is wide, if the channels, venues, and intermediaries through which they might later deal with one another remain in place; two parties at peace may occupy a field that is narrow, if those conditions have been dismantled. The distinction between the state of a relation and the field within which relations can be generated is the distinction on which the account rests.

Definition 2 (Neutral custodianship). A party is a neutral custodian with respect to a conflict when it maintains conditions constituting the relational field of the conflicting parties, and its maintenance of those conditions is not directed at securing any particular arrangement within the field.

Definition 2 makes neutrality a relation between a party and a field rather than a position between parties. It requires no claim about the custodian’s beliefs, no equal distribution of its conduct, and no midpoint. What it requires is a separation between the conditions the custodian maintains and the outcomes those conditions make possible.

Registers of Custodial Action on Relational Conditions

Acting on conditions differs from acting on arrangements in what it determines. An action on an arrangement selects among the members of the field: it brings one about, prevents another, or alters its terms. An action on a condition alters the field itself: it adds members, removes members, or changes the cost of reaching them, without selecting among those that remain.

Three registers of custodial action follow. The custodian may enable, adding conditions and thereby widening the field, as when it establishes a channel that did not exist. It may sustain, holding existing conditions in place against pressure to remove them, as when it maintains a venue that one party would prefer closed. And it may refrain, declining to remove a condition that it has the power to remove and that removing would serve its own interest. The third register is the one that carries the account’s normative weight, because a custodian’s power over the field is exercised as much by what it declines to do as by what it does.

Ostrom’s analysis of durable common-pool arrangements locates their success in institutions that sustain the conditions of continued use rather than in rules prescribing particular allocations (Ostrom 1990). The transposition here replaces a physical resource with a relational field and replaces users with parties in conflict; the structure of the claim is inherited rather than independently discovered, and the paper claims Ostrom as forerunner in Section 3.6.

A Viability Formulation of Field Maintenance

The maintenance requirement can be stated more precisely than Definition 2 states it. What a custodian owes is that the field be kept from emptying, rather than that it remain unchanged, which no actor can guarantee.

Viability theory studies systems constrained to remain within a specified set, and defines the viability kernel of that set as the collection of states from which at least one evolution of the system remains inside the set indefinitely (Aubin 1991; Aubin, Bayen, and Saint-Pierre 2011). The apparatus was developed for controlled dynamical systems and has been applied to problems in ecology, economics, and resource management (Aubin, Bayen, and Saint-Pierre 2011). Its transposition to relational conditions is the present paper’s construction and is not attributable to that literature.

Definition 3 (Field maintenance requirement). A relational field satisfies the maintenance requirement over a horizon when, for each party, the set of arrangements from which some continuation remains available within the horizon is non-empty.

The requirement is weaker than it may appear and this weakness is deliberate. It does not require that any particular relation be preserved, that the parties reconcile, or that the conflict end. It requires only that no party be placed in a position from which every route to future relation has been removed. The custodian’s obligation is correspondingly bounded: it maintains the conditions under which some continuation is available, and takes no responsibility for which continuation the parties choose.

Two consequences are worth stating. The requirement is indexed to a horizon, so that a field that empties within a decade and refills within a century satisfies it under one horizon and fails it under another; the horizon must therefore be stated wherever the requirement is invoked. And the requirement concerns each party separately, so that a field that remains wide for one party and empties for the other fails the requirement rather than satisfying it on average. The second consequence is what prevents the account from licensing arrangements that preserve access for the stronger party alone.

Specification without enumeration.

Definition 1 defines a field as a set of arrangements, and no procedure for enumerating such a set is available. The requirement proceeds without one. What Definition 3 asks is whether the set is non-empty for each party, and that question can be settled from the conditions rather than from the arrangements they permit.

Definition 4 (Contact structure). Let a set of parties and a finite inventory of relational conditions be given. Each condition supports a set of ordered pairs of parties, the first able to initiate through it and the second able to receive. The contact structure at a time is the union of these sets over the conditions then in place.

Proposition 5 (Reduction of the maintenance requirement). At a given time, the field is non-empty for a party if and only if the contact structure contains at least one ordered pair with that party as initiator and some other party as recipient. Given the condition inventory, the instantaneous maintenance requirement is therefore decidable.

Argument. An arrangement between two parties presupposes some condition through which one can address the other and the other can receive the address; where no condition supports such a pair, no arrangement between them is available, and the field is empty for that party. Where one exists, the arrangement consisting in the use of that condition is available, and the field is non-empty. The conditions are finite by hypothesis and each is an institutional fact, so the test is a finite inspection. ◻

The horizon-indexed form of the requirement follows by applying the same test to the evolution of the inventory: the states of the inventory from which some evolution keeps every party’s contact set non-empty throughout the horizon are the viability kernel of the constraint set, in the sense of the literature cited above (Aubin 1991; Aubin, Bayen, and Saint-Pierre 2011). The state space is finite and discrete, so the kernel is computable for inventories of the size a real case presents.

Three limits of the reduction should be stated. It replaces a field of arrangements with a graph of possible contacts, and it therefore registers only whether relation is possible, not on what terms; a field reduced to a single grudging channel and a field of many is indistinguishable under the test. It detects total foreclosure and is insensitive to narrowing, which is the more common harm. And the inventory must be drawn up by someone, so the test moves the discretion from the measurement to the specification rather than removing it. The reduction is accordingly offered as a lower bound on the field: what it reports as empty is empty, and what it reports as non-empty may be very nearly so.

Symmetry of Refusal and Asymmetry of Possibility

The account rejects equidistance and nonetheless requires a symmetry. The two are distinguished by the space in which the symmetry holds.

Claim 6 (Symmetric refusal). A neutral custodian refuses symmetrically in the space of field-constitutive action: no party is granted the power to determine, through the custodian, what the field contains. Judgment, responsibility, and outcome remain asymmetric.

The symmetry is a symmetry of denial rather than of provision. Neither party receives a veto over the conditions, neither receives privileged access, and neither receives the use of the custodian’s constitutive powers for its own ends. What the parties may receive unequally is everything else: the custodian’s assessment of the merits, its public statements, its testimony, its allocation of assistance according to need, and its judgment about responsibility.

Claim 6 identifies what is right in the ordinary conception and what is wrong with it. The ordinary conception observes correctly that neutrality involves an evenhandedness and then locates that evenhandedness in conduct and in judgment, where it produces the difficulties recorded in Section 3.1. Locating it instead in field-constitutive action preserves the intuition while removing its consequences.

The Modality Constraint on Field-Constitutive Powers

A custodian holds powers of a distinctive kind. Control over access to a venue, over the hosting of a process, over a procedure, over a communication channel, and over an agenda are powers whose exercise determines what the field contains. They are the powers by which the field exists, and they are held by the custodian precisely because the parties have accepted that it holds them.

Definition 7 (Field-constitutive power). A power is field-constitutive with respect to a set of parties when its exercise changes the set of arrangements available to them, rather than changing which available arrangement they select.

Proposition 8 (Derivation of the modality constraint). If a custodian uses a field-constitutive power in order to bring about a particular arrangement, then the field is restricted to those arrangements consistent with that outcome, and the determination of the outcome passes from the parties to the custodian. Under Definition 3 the maintenance requirement is then satisfied only as a by-product of the custodian’s aim rather than as a constraint on it.

Argument. By Definition 7 the exercise of a field-constitutive power changes the set of available arrangements. Directing that exercise at an outcome means selecting the change by whether the outcome follows, which is to restrict the available set to arrangements compatible with the outcome. The parties then choose within a set already selected for the result, so that the custodian rather than the parties determines what is chosen among. The maintenance requirement constrains the resulting set only where the custodian’s aim happens to leave it non-empty for each party, which is the condition Definition 3 was intended to impose rather than to inherit. ◻

The constraint follows.

Claim 9 (Modality constraint). Neutrality restricts the modality of influence rather than its content. A neutral custodian may judge, testify, publish, and persuade, and may do so asymmetrically. It may not condition access to the field on alignment, and it may not employ field-constitutive powers to secure a substantive outcome.

Claim 9 is a constraint on instruments rather than on aims or on effects, and this is what distinguishes it from the liberal conception recorded in Section 3.4. Raz’s argument that comprehensive neutrality is unattainable is directed at neutrality of effect and at the impossibility of arrangements whose consequences favour no conception (Raz 1986). That argument leaves the present constraint intact, because the constraint makes no claim about effects. A custodian whose conduct predictably advantages one party has not violated Claim 9 unless it employed a field-constitutive power to produce that advantage. The constraint is attainable in the way a rule of recusal is attainable, and it fails in the way such rules fail, through concealment and through disputes about what counts as an exercise of the power.

The constraint also supplies a test. Given a contested action by a custodian, the questions are whether the power exercised was field-constitutive under Definition 7, and whether its exercise was selected by the outcome it produced. Where both hold, the action violates the constraint regardless of the merits of the outcome. Where the first fails, the action is influence within the field and is answerable on its merits rather than under the constraint. Section 8.3 applies the test to the passages of the case that the account finds hardest, and the test earns its place only if it classifies them without amendment.

The Sustainment of Neutral Capacity

Section 5 specified what a neutral custodian maintains and what constrains its instruments. This section asks how the capacity to do so is sustained across time, and identifies the limits of the mechanism that sustains it.

Relational Utility and the Reinforcement Loop

A custodian is either used by the parties or left aside. Where it is used, its maintenance of the field produces something the parties value: a channel that functions when direct channels have closed, a venue whose availability neither side controls, a representative through which one state’s interests are handled in another with which it has no relations. Call this the custodian’s relational utility. Utility of this kind generates a reinforcement sequence. The custodian’s restraint under the modality constraint makes its conduct predictable; predictability makes it usable by parties that trust one another very little; use accumulates a record; the record makes the custodian more usable still; and parties that have come to rely on the arrangement acquire an interest in its continuation that is independent of the custodian’s own preferences.

Three features of the trust literature bear on this sequence. Luhmann’s distinction between trust, which involves a decision taken in awareness of alternatives, and confidence, which is a routine expectation held without considering them, identifies what a custodian’s record produces: parties come to rely on the arrangement without re-deciding it, and system trust of this kind attaches to the institution rather than to persons (Luhmann 1979, 1988). Kydd’s account of reassurance shows that adversaries can establish trust through signals costly enough to distinguish the trustworthy from the untrustworthy (Kydd 2005), which is what a record of restraint under pressure supplies. And Larson’s analysis of missed opportunities in Cold War relations identifies the situation a custodian addresses: agreements in both parties’ interest that fail because neither can credit the other (Larson 1997). Work on trust between leaders and on the institutional design of trust-building extends the same concern (Wheeler 2018; Hoffman 2006).

The sequence must be distinguished from a superficially similar one. Simmel’s analysis of the triad identifies a third position, the tertius gaudens, whose advantage arises precisely from the separation of the other two (Simmel 1950), and Burt’s account of brokerage across structural holes gives the modern formulation, in which the broker’s informational and control advantages depend on the disconnection it spans (Burt 1992). Work in the same tradition separates brokerage that sustains open connection among the other parties from brokerage that maintains their conflict (“Divide Et Impera: Conflictual Ties in Broker Relationships” 2016). The custodian described here occupies the first position. Its utility depends on the parties being able to reach one another through it, and it therefore has no interest in their continued separation. The distinction is not a moral one: it is a difference in where the advantage comes from, and it generates opposite incentives with respect to resolution.

Saturation, Non-Monotonicity, and Capture Risk

A reinforcement sequence with no limiting term describes unbounded growth and therefore describes nothing. Three limits operate on the sequence above, and stating them is necessary for the account to be a mechanism rather than a narrative.

The first is that relational utility is not monotone in detachment. A party with no relations has no channels to offer, no venue anyone recognises, and no record. Custodial capacity requires embeddedness: standing relations with both sides, participation in the institutions through which the parties act, and sufficient material and diplomatic weight for its offers to be worth accepting. Detachment beyond the point at which these are maintained reduces utility rather than increasing it, so the relation between distance and usefulness has an interior maximum rather than an increasing one.

The second is that the embeddedness which produces utility is also the condition of capture. A custodian with dense relations to both sides is a custodian with interests exposed to both, and the exposure is what a party seeking to convert the field to its own use will address. Capture does not require the custodian to abandon its position; it requires only that field-constitutive decisions begin to track one party’s preferences, which is a violation of Claim 9 that may be invisible from outside. Centrality therefore raises capacity and raises exposure together, and the two cannot be separated by choosing a level of one.

The third is that the value of the field to the parties can exceed the level at which they are content to leave it in the custodian’s hands. A custodian that becomes indispensable becomes worth controlling. This limit is the reason the account treats capacity as something to be maintained within bounds rather than maximised, and it is the mechanism-level counterpart of the normative limit developed in Section 8.2.

Positional Non-Alignment and Relational Infrastructure

Two things are commonly described as neutrality and they behave differently over time.

A position is a declared non-alignment. It can be adopted by announcement, it costs little to hold in the absence of pressure, and it can be abandoned by a further announcement. Nothing about holding it in one decade makes it easier to hold in the next, and its abandonment destroys no asset that anyone else values.

Relational infrastructure is the accumulated apparatus through which a custodian maintains a field: hosting capacity that other parties use, standing mandates that other parties have entrusted to it, institutional residency, procedural competence, and an operating record that would have to be rebuilt from the beginning if it were dismantled. Infrastructure of this kind is costly to build, is used by parties other than its holder, and cannot be abandoned without destroying value that those other parties hold.

Claim 10 (Accumulation asymmetry). The abandonment of a declared position costs other parties nothing. The dismantling of relational infrastructure destroys value that other parties hold, because its accumulation has placed a stake in their hands. Accumulated relational infrastructure therefore raises the cost of reversal, and supplies the reinforcement described in Section 6.1. It is one source of durability among others: it neither guarantees that a neutrality carrying it will persist, nor implies that a neutrality without it will be abandoned.

The claim was initially formulated in a stronger form, predicting that neutralities carried by position would be abandoned under changed security conditions and those carried by infrastructure would persist. Section 8.4 records the case that defeats the stronger form. Austrian neutrality is positional on the account’s own criteria, having been adopted in 1955 as the political precondition of restored sovereignty and remaining unilaterally terminable (“Constitutional Law on the Neutrality of Austria, Adopted 26 October 1955” 1955; Austrian Information, n.d.), and Austria has not abandoned it under the conditions in which Finland and Sweden abandoned theirs. A counterexample of this kind leaves untouched the mechanism the claim describes, which concerns the cost imposed on third parties; it defeats the categorical prediction, and the claim is stated accordingly.

The recent record nonetheless bears on the weaker form. Finland and Sweden ended long-standing non-alignment following the Russian invasion of Ukraine, acceding to NATO in 2023 and 2024 respectively (House of Commons Library 2024); analysis of those accessions traces how their non-alignment had been institutionally and politically constituted and how the reversal became available (Isaksson 2024). Switzerland in the same period revised elements of its neutrality policy under comparable pressure (European Council on Foreign Relations 2024) and retained the institution itself (Center for Security Studies, ETH Zurich and Military Academy at ETH Zurich 2023).

Three cautions apply and all are recorded rather than resolved. The comparison requires an independent measure of accumulated relational infrastructure, which this paper does not supply and which Section 10 records as an outstanding requirement. The Finnish and Swedish decisions have explanations in terms of proximity, threat perception, and alliance politics that do not depend on the present account at all. And durability plainly has sources besides third-party stake: domestic attachment to the status is an evident candidate, it would operate internally rather than relationally, and identifying it would require evidence this paper has not gathered.

The Temporal Condition of Neutral Capacity

A consequence of Claim 10 concerns when custodial capacity can be built. Infrastructure accumulates over spans longer than a crisis, is built through use rather than through declaration, and requires that other parties have had occasion to rely on it. A party that announces custodial availability at the onset of a polarized conflict therefore offers a position and not an infrastructure, and its offer carries the reliability of a position. Capacity is built in the interval between conflicts, and it must therefore be in place before the occasion for its use arises.

The economic literature on decision under irreversibility supplies the shape of the corresponding value. Where a decision forecloses options and where information is expected to arrive, there is a value attaching to preserving the capacity to act later rather than committing now; this quasi-option value was introduced by Arrow and Fisher and by Henry (Arrow and Fisher 1974; Henry 1974) and given its formal statement by Hanemann (Hanemann 1989). It should be distinguished from the option value developed in the real-options literature on irreversible investment (Dixit and Pindyck 1994): the two are not equivalent, and the difference concerns the value of learning under preservation as against the value of postponement (Mensink and Requate 2005).

The quantity relevant here is the first. What a custodian preserves is the capacity of the parties to act on information they do not yet have, and the cost of that preservation is the return foregone by declining alignment. The transposition of the concept from environmental preservation to relational conditions is the present paper’s construction and is not attributable to the cited literature. Its use in this paper is to state a structure and not to compute a value: no measurement of the relevant quantities is attempted, and Section 10 records that as a limitation rather than an oversight.

Two-Scale Structure of Internal Heterogeneity and External Neutrality

A rival explanation of the Swiss case does not treat neutrality as an achievement at all. On that explanation, a confederation spanning Catholic and Protestant cantons and German, French, Italian, and Romansh communities cannot align externally without reproducing the alignment internally, so that neutrality is a requirement of domestic cohesion rather than a governance capacity directed outward.

The literature on consociational and concordance democracy supplies the mechanism such an explanation needs. Lijphart’s account identifies deliberate elite accommodation as the means by which fragmented societies remain stable, and treats Switzerland as a principal instance (Lijphart 1969, 1977); Lehmbruch developed the parallel account of concordance democracy for Switzerland and Austria (Lehmbruch 1967); Steiner documented the practice of amicable agreement in Swiss decision-making (Steiner 1974); and Vatter examines whether the pattern is being displaced (Vatter 2016). None of these works advances the claim that Swiss neutrality is explained by internal heterogeneity. The mechanism is theirs; the application is the present paper’s, and it is offered as a candidate explanation rather than as an established finding.

The account absorbs this rival rather than disputing it. If a polity maintains its own cohesion by sustaining conditions under which its constituent communities can continue to deal with one another without any of them determining the terms, then it is already performing custodianship in the sense of Definition 2, at an internal scale. External neutrality is then the same operation performed on a wider field, and the two are instances of one principle rather than competing explanations of one outcome.

The absorption carries a risk that must be stated. An account that can accommodate its own principal rival at both scales is at risk of accommodating anything, and Section 4.3 accordingly requires that the field be specifiable independently of the outcomes it leaves open. The two-scale reading is testable to the extent that it predicts a relationship between the internal and external practices of a polity: a state that governs its internal heterogeneity by majority determination rather than by accommodation should not sustain external custodianship well, and a state that abandons internal accommodation should be expected to lose external custodial capacity thereafter. Neither prediction is examined here.

Case Study: Switzerland

This section reads the Swiss record through the account. The material was set out in Section 2.2; what follows is the analysis, and each subsection states what the account makes of the material and where the material resists it.

Formation and International Recognition, 1815

The account predicts that a custodial neutrality is constituted before the conflicts in which it is used, because infrastructure accumulates and positions do not. The Swiss record is consistent with the prediction and falls short of establishing it.

Permanent neutrality received international recognition at the Congress of Vienna in 1815 (Federal Department of Foreign Affairs 2026), in a settlement rather than in a crisis, and the conflicts in which Swiss custodial capacity was later exercised followed that settlement by decades. The earlier dating to 1515 is treated here as a founding narrative rather than as a date of constitution, for the reasons given in Section 2.2. Two features of the 1815 settlement bear on the account. Recognition was extended by the powers surrounding Switzerland rather than claimed unilaterally, so that from the outset the arrangement was one in which other parties held an interest. And the historical assessment that neutrality was in substantial part imposed by stronger neighbours rather than chosen (swissinfo.ch, n.d.b) cuts against any reading of the case as an achievement of policy.

The second point is worth retaining rather than minimising. If neutrality was imposed, then its origin lies outside the mechanism this paper describes, and the account applies to its maintenance rather than to its creation. That is a narrower claim and it is the claim the evidence supports.

The Law–Policy Distinction and the Maintenance of the Field

The Swiss separation of the law of neutrality from the policy of neutrality (Federal Department of Foreign Affairs 2026) is read here as an instrument of field maintenance rather than as a taxonomy of permissions.

The law fixes a small number of prohibitions on the state as an actor (“Convention (V) Respecting the Rights and Duties of Neutral Powers and Persons in Case of War on Land” 1907; “Convention (XIII) Respecting the Rights and Duties of Neutral Powers in Naval War” 1907). What it fixes is precisely the class of conduct by which a neutral state could place its own weight behind a belligerent, and it is therefore the legal residue of Claim 9: the constraint on instruments, given a narrow codification, for the single instrument of military support. The policy layer holds everything else, and its discretionary character is what allows the custodian to adjust the conditions it maintains as circumstances change.

The division does work that a single undifferentiated commitment could not. A neutrality specified entirely in law would be unable to respond to conditions its drafters did not anticipate, and a neutrality specified entirely in policy would offer no fixed point against which a party could check the custodian’s conduct. The pairing supplies a stable floor together with an adjustable apparatus above it, which is the structure a custodian requires if predictability and responsiveness are both conditions of its capacity.

The account should not present the distinction as its own discovery. It is Swiss governmental doctrine, developed for the purpose it serves, and the present contribution is the reading of that doctrine as field maintenance rather than the doctrine itself.

Geneva and the Accumulation of Relational Infrastructure

Geneva is the clearest instance in the case of what Section 6.3 calls relational infrastructure.

The concentration is a matter of record. The League of Nations was headquartered in Geneva from 1920, initially in the Palais Wilson and from 1936 in the purpose-built Palais des Nations, whose assets passed to the United Nations on the League’s dissolution in 1946 (United Nations Office at Geneva, n.d.a). The city now houses approximately forty international organisations, one hundred and eighty permanent missions, and more than four hundred non-governmental organisations; the United Nations Office at Geneva is one of only three major administrative offices of the organisation outside New York; and entities headquartered there include the World Health Organization, the International Labour Organization, the Office of the High Commissioner for Human Rights, the United Nations High Commissioner for Refugees, the World Intellectual Property Organization, the World Meteorological Organization, and the World Trade Organization (United Nations Office at Geneva, n.d.b). The same source records that International Geneva grew through the presence of the League and also through the Red Cross and through organisations such as the International Telecommunication Union, founded in Berne in 1865 and moved to Geneva in 1948 (United Nations Office at Geneva, n.d.b).

Three features of this concentration correspond to the properties Section 6.3 attributes to infrastructure. It was built over more than a century rather than adopted. It is used by parties other than its holder, and those parties bear the cost of its loss. And it cannot be abandoned by declaration, because what would be abandoned is a set of arrangements that other states have built their own conduct around.

The protecting-power practice exhibits the same properties in a more direct form. Swiss good offices comprise protecting-power mandates, host-state provision, and mediation (Federal Department of Foreign Affairs, n.d.a). Under a protecting-power mandate Switzerland assumes consular or diplomatic functions for one state in another with which it has broken relations, which permits the two states to retain a minimum of communication (Federal Department of Foreign Affairs, n.d.b). The volume fluctuates with the state of the world’s conflicts: Switzerland undertook more than two hundred such mandates during the Second World War (Swiss Federal Council, n.d.) and currently holds a small number, reported as six in a Federal Council report (Swiss Federal Council, n.d.) and as eight in the department’s current statement, the latter including representation of Iranian interests in Egypt since 1979, United States interests in Iran since 1980, Russian interests in Georgia and Georgian interests in Russia since 2009, and Iranian interests in Canada since 2019 (Federal Department of Foreign Affairs, n.d.b).

A protecting-power mandate is a relational condition in the sense of Definition 1 and is close to the paradigm case. It creates no relation between the parties and settles nothing between them. What it supplies is the arrangement under which a relation remains formable after direct relations have been severed. That the durations are measured in decades is the accumulation claim in its most concrete form.

Armed Neutrality and the Boundary Condition of Openness

Armed neutrality appears paradoxical on a positional reading, since a party that maintains substantial military capacity appears to do rather more than stand aside. On the account developed here the pairing is expected rather than anomalous.

A custodian maintains conditions it does not own. Those conditions are useful to the parties precisely because no party controls them, and a condition that can be seized by whichever party finds it inconvenient supplies nothing. The defensive capacity is therefore not a departure from custodianship but the requirement that the field be beyond unilateral capture. The boundary is what makes the openness inside it dependable.

The same reasoning identifies where the pairing becomes unstable. A capacity sufficient to prevent seizure is a condition of custodianship; a capacity sufficient to make the custodian a factor in the parties’ own calculations begins to place its weight inside the field, which Claim 9 excludes. The account therefore predicts a bounded rather than a maximised capacity, and the recent Swiss debate about defence expenditure and cooperation with neighbouring security structures (European Council on Foreign Relations 2024) is a debate about where that bound lies. This paper takes no position on the answer and records only that the account makes the question intelligible.

Sanctions, Re-Export, and the Protecting-Power Mandate, 2022–2024

The recent record supplies the account’s most informative passage, because the decisions ran in opposite directions and the consequences are visible.

In February 2022 Switzerland adopted the European Union’s sanctions package against Russia (Euronews 2022b). Under Claim 9 the decision is permissible. Sanctions are an exercise of economic and political weight directed at a party’s conduct; they are not an exercise of a field-constitutive power, since they neither close a channel that the parties use to reach one another nor condition access to Swiss hosting on alignment. On the account, a custodian may judge and may act on its judgment in this register.

In the same period Switzerland declined requests to permit the re-export of Swiss-manufactured war materiel to Ukraine, citing equal treatment under the law of neutrality (Euronews 2022a). This decision is also permissible under the account, and for a different reason: it is the observance of the legal floor described in Section 7.2. It is nonetheless costly, and the cost fell where the account predicts. Domestic majorities favoured relaxation (swissinfo.ch 2023), an expert commission subsequently recommended re-export permissions for defined partner states (European Council on Foreign Relations 2024), and the government continued to distinguish measures compatible with the law of neutrality from those it regards as excluded (swissinfo.ch 2024).

The consequence appeared in the custodial register. In August 2022 Russia declined Switzerland’s offer to represent Ukrainian interests in Moscow, stating that Switzerland had lost its status as a neutral state by joining Western sanctions (Al Jazeera 2022). The same period saw other Swiss offers declined: an attempt to convene Russian–Ukrainian talks in March 2022 did not proceed, and negotiations in which Switzerland held standing mandates were conducted elsewhere (swissinfo.ch, n.d.a).

Three readings of this sequence are available and the account takes the third. On the first, the sanctions decision violated neutrality and the loss of the mandate is its correct consequence. On the second, the sanctions decision was compatible with the law of neutrality and the Russian response was a pretext. On the third, which the account supports, both parts are true at once and the case exhibits a structural feature rather than an error: a decision permissible under the modality constraint can nonetheless reduce relational utility, because utility depends on the other parties’ willingness to use the custodian, and that willingness is not governed by the constraint. Permissibility is a property of the custodian’s conduct; usability is a property of the relation, and the parties determine it.

The consequence for the account is a qualification worth stating plainly. Claim 9 constrains what a custodian may do; it does not guarantee that a custodian which observes it will remain usable. The two come apart under polarization, when the parties themselves cease to recognise distinctions between conduct inside the field and conduct upon it. The account therefore predicts that the maintenance of custodial capacity becomes harder exactly where it is most needed, which is a cost of the position rather than a defect of the analysis, and Section 8 takes it up.

Mechanism Failure and Normative Defeat of Neutrality

Neutrality fails in two ways that require separate treatment. In the first the capacity itself is lost, and the custodian ceases to be able to maintain the field. In the second the capacity survives, functions, and should not be exercised. The two have different diagnostics, different remedies, and different relations to the account, and running them together is what allows a defence of neutrality to answer a moral objection with an operational reply.

Capture, Unpredictability, and the Loss of Relational Utility

Mechanism failure has three forms, corresponding to the three limits identified in Section 6.2.

Capture is the conversion of field-constitutive decisions into instruments of one party’s purposes. It does not require the custodian to announce an alignment, and it may be invisible from outside, since the observable conduct of a captured custodian and an uncaptured one can coincide for long periods. Capture is a violation of Claim 9 and it is also the failure the constraint is least able to detect, because the constraint governs the selection of an exercise by its outcome and selection is a matter of reasons rather than of conduct.

Unpredictability is the loss of the property that makes a custodian usable by parties that trust one another very little. A custodian whose responses cannot be anticipated supplies no reduction in the parties’ uncertainty, which is the service they were purchasing. This failure can arise from good motives: a custodian that adjusts its conduct case by case according to the merits will be more responsive and less usable.

Loss of relational utility is the condition in which the parties cease to use the custodian, whatever the custodian does. Section 7.5 recorded an instance and drew the consequence that matters here: utility is not within the custodian’s control, because it depends on the parties’ willingness, and that willingness is not governed by the constraint. A custodian can therefore observe every requirement of the account and lose the capacity notwithstanding. The account should be understood as stating what a custodian must do, not as promising what it will thereby retain.

Normative Defeat and Irreversible Foreclosure

Normative defeat is a different matter. Here the custodian retains its independence, is predictable, and is used. The question is whether it should continue.

The obvious answer weighs the good of the preserved field against the harm its preservation permits. This paper declines that answer, because a weighing of this kind licenses whatever produces the larger sum and supplies no principled stopping point; it is also unavailable to an account whose central claim is that the value at stake is a capacity rather than an outcome. The criterion offered instead operates inside the account’s own vocabulary.

Claim 11 (Normative defeat). Custodianship is normatively defeated when the maintenance of the field is itself the mechanism through which the relational possibility of some persons is irreversibly foreclosed.

Claim 11 does not ask how much harm the arrangement permits. It asks whether the arrangement is the instrument of the harm. A custodian that maintains conditions under which one party proceeds to destroy another has not, by that fact alone, been defeated; a custodian whose maintenance of those conditions is what makes the destruction possible has been. The criterion is therefore causal in form and it is demanding to apply, which is a cost recorded rather than concealed.

Two features require comment. First, the claim is stated in terms of persons rather than parties. An earlier formulation in terms of the parties to the conflict was too narrow, and Section 8.3 explains why the case requires the wider form: the gravest failures associated with neutrality have fallen on persons who were not parties to the conflict and who had no standing in the field at all. Second, irreversibility is indexed to a horizon, as Section 2.3 requires. The annihilation of a person is the limiting case, in which the horizon does not matter because no horizon restores the possibility.

Wartime Switzerland and the International Committee of the Red Cross

This section is the account’s principal test. It is written to be capable of failing, and it records two respects in which it does.

The documented record.

The Independent Commission of Experts established by the Swiss parliament found that the Swiss National Bank received substantial quantities of gold from German sources during the war, a large part of it looted and a portion taken from individual victims, and documented the depth of Swiss economic relations with the Reich and the restrictive character of Swiss refugee policy (Independent Commission of Experts Switzerland–Second World War 2002); a governmental study in another jurisdiction reached corroborating conclusions on the gold transactions (Eizenstat 1998). On the question why Switzerland was not invaded, the archival evidence establishes that German operational planning against Switzerland was serious preparatory work rather than staff speculation (Urner 2001), and the scholarly literature weighs armed deterrence against economic and structural considerations without resolving to either alone (Steinberg 1996; Weinberg 1999; Vagts 1997). The deterrence account has its principal advocates outside the professional historiography of the period (Halbrook 1998; Codevilla 2000), and this paper accordingly treats non-invasion as jointly explained rather than attributing it to deterrence.

The humanitarian record is documented by the institution itself. Scholarship establishes that the International Committee of the Red Cross possessed knowledge of the extermination programme and issued no public condemnation (Steinacher 2017). The Committee’s president stated in 1995 that the organisation had suffered a moral failure in relation to the Holocaust, having failed to move beyond the limited legal framework the states had established (“Press Conference Given by the President of the ICRC (Geneva, 30 May 1995)” 1995). At a commemoration in Geneva in 2015 a later president stated that the organisation had failed to protect civilians and in particular the Jews persecuted and murdered by the Nazi regime, and that it had failed as a humanitarian organisation because it had lost its moral compass (Euronews 2023). The institution has since stated that it rejects the binary of silence against denunciation (International Committee of the Red Cross 2024).

Conduct classified by the modality constraint.

Applying the test of Section 5.5: hosting, good offices, and the protecting-power mandates undertaken in large numbers during the war (Swiss Federal Council, n.d.) are exercises of field-constitutive power that were not selected by the outcomes they produced, and they pass. Had Switzerland closed its hosting or its mandates to one belligerent while keeping them open to the other, the exercise would have been selected by the outcome and would have failed. That is the classification the account promised, and on this material it delivers it.

Conduct beyond the reach of the constraint.

It does not follow that the constraint classifies the record. Two of the gravest elements fall outside it.

The gold transactions were commercial and financial dealings. They are an exercise of economic weight rather than of a power constituting the belligerents’ field, and Definition 7 therefore does not capture them. Whatever their character under the law of neutrality or under ordinary morality, Claim 9 is silent about them.

The refugee policy is the harder case and the more important one. Control of a border is unmistakably a power over access. But the access it governed was access by persons who were not parties to the conflict and who held no place in the belligerents’ field. The constraint, as stated, governs a custodian’s powers over the field of the parties, and on its own terms it does not reach conduct toward those who are not parties to it.

This is a limit of the constraint and the paper states it rather than repairing it by amendment. Section 4.3 required that the constraint classify contested conduct without case-by-case amendment, and the requirement is met in the sense that the constraint gives a determinate answer here: it does not apply. What the case establishes is that the constraint is a constraint on custodianship and not a complete ethics of a custodian’s conduct. A custodian can satisfy it entirely and act monstrously in a register it does not govern.

Conduct classified by normative defeat.

Claim 11 was stated in terms of persons for this reason. The ICRC’s position during the war is the case the criterion was built to reach: the maintenance of access to belligerent authorities, and the silence that maintenance was understood to require, coincided with the irreversible foreclosure of relational possibility for the persons about whom the institution remained silent. Whether the maintenance of access was the mechanism of that foreclosure, in the causal sense Claim 11 requires, is a historical question this paper does not settle; the institution’s own successive statements (“Press Conference Given by the President of the ICRC (Geneva, 30 May 1995)” 1995; Euronews 2023) indicate a judgment closer to the affirmative than the operational defence of neutrality permits.

Two conclusions follow, and the second is the more uncomfortable. The Committee’s rejection of the binary of silence against denunciation (International Committee of the Red Cross 2024) is the correct institutional lesson and it is the lesson Claim 9 states in general form: public judgment and the maintenance of access are different registers, and a custodian that treats speech as a field-constitutive act has conceded to the parties a power the account denies them. But the account did not need to be developed for the Committee to have seen this, and it was not seen. An account that identifies, after the fact, a distinction whose observance would have required a courage the institution lacked has explained less than it appears to have explained. The failure was not principally conceptual, and the paper should not present a conceptual remedy for it.

Violated Guarantees and Abandoned Non-Alignment

Section 4.2 recorded that an account built on the neutrality that survived risks describing survival. Two classes of non-surviving case bear on the account, and the evidence available here is partial.

Guarantee without capacity.

Belgian neutrality was not merely declared: it was guaranteed by the European powers under the Treaty of London of 1839, which required Belgium to remain perpetually neutral and forbade it alliances with the surrounding powers. The consequence was a state exposed on three borders between two long-standing opponents, whose political parties resisted defence expenditure, whose few fortifications were outdated, and which adopted conscription only in 1909 (Shelby 2014). The guarantee did not prevent the invasion of 1914.

The case bears on Section 7.4 rather than contradicting it. The account holds that a field must be beyond unilateral capture if it is to be useful, and that the defensive boundary is what secures this. Belgium had the guarantee and lacked the boundary, and the guarantee proved to be a commitment by others rather than a capacity of its own. This supports the reading of armed neutrality offered above and it should be stated with the caution its evidential basis requires: one case, drawn from a source concerned with a different subject, does not establish a general relation between defensive capacity and neutral survival.

Neutrality invoked by the party violating it.

The German operation against Denmark and Norway on 9 April 1940 was presented to both governments as a measure for the protection of their neutrality against Allied aggression, and was described in German terms as the military measures for the protection of the neutrality of Denmark and Norway (Imperial War Museums 2021). The claim was made in the act of violating the status it invoked.

This is a distinct failure mode and the account should name it. A custodian’s position depends on other parties recognising that a set of conditions is not theirs to determine. A belligerent that appropriates the vocabulary of protection does not merely violate the conditions; it asserts standing to determine what protecting them requires. The account’s constraint governs the custodian and supplies no defence against this, because the party making the claim is not the custodian and is not bound by a constraint on custodial conduct. Where the language of neutrality can be seized by a belligerent, the custodian’s own observance of the constraint becomes invisible, which is a further respect in which permissibility and usability come apart.

Position without infrastructure.

Finland and Sweden ended long-standing non-alignment and acceded to NATO in 2023 and 2024 (House of Commons Library 2024), under conditions in which Switzerland revised its neutrality policy (European Council on Foreign Relations 2024) and retained the institution (Center for Security Studies, ETH Zurich and Military Academy at ETH Zurich 2023). Claim 10 predicts the divergence: a position can be exchanged, and an infrastructure that other parties use cannot be exchanged without destroying value those parties hold.

The prediction is not established by these cases and Section 6.3 recorded why. The comparison requires an independent measure of accumulated relational infrastructure, and the Finnish and Swedish decisions admit explanations in terms of proximity and threat that do not involve the account. Analysis of how their non-alignment had been constituted (Isaksson 2024) is the material from which such a comparison could be built, and building it is left to further work.

Neutrality enacted at the restoration of sovereignty.

Austria supplies a third form and the one that bears most directly on Claim 10. On 26 October 1955 the Austrian parliament enacted a constitutional law by which Austria, of her own free will, declared her permanent neutrality, undertook to maintain and defend it with all the means at her disposal, and provided that she would never accede to military alliances nor permit foreign military bases on her territory (“Constitutional Law on the Neutrality of Austria, Adopted 26 October 1955” 1955; Centre virtuel de la connaissance sur l’Europe, n.d.). The declaration followed the Austrian State Treaty of 15 May 1955 and the withdrawal of the occupying forces, and it had been promised in the Moscow Memorandum of April 1955, in which Austria undertook a neutrality of the same type as that maintained by Switzerland (Austrian Information, n.d.).

Three features distinguish the Austrian case from the Swiss one on the account’s own terms. The status was adopted as the political precondition of restored sovereignty rather than accumulated through use. It was adopted voluntarily as a matter of law, with the consequence that Austria may, if it chooses, terminate the status unilaterally (Austrian Information, n.d.) – which is the defining property of a position in the sense of Section 6.3. And Austria’s own interpretation diverged from the Swiss almost immediately, treating permanent neutrality as compatible with membership of the United Nations, which Switzerland did not (Austrian Information, n.d.).

The comparison is instructive precisely because the two states adopted the same legal status. A promise to maintain neutrality of the Swiss type produced, in the Austrian case, a constitutional commitment; it did not produce Geneva.

The case also constrains the account. Austria is positional on the criteria just given, and Austria did not abandon its neutrality under the conditions in which Finland and Sweden abandoned theirs. An earlier formulation of Claim 10 predicted that it would, and Section 6.3 accordingly states the claim in the weaker form: accumulated infrastructure raises the cost of reversal and is one source of durability among several. Austrian durability appears to rest on something other than third-party stake, and identifying what would require evidence this paper has not gathered. The comparison remains the most tractable available test of the weaker claim, because legal status is held constant across the two cases, and the present paper does not carry it out.

Cases left outside this comparison.

The invasions of the Netherlands and Luxembourg in May 1940 are not examined here, because the sources required were not returned by the verification passes underlying this draft. Their absence narrows the class of violated neutralities considered and is recorded in Section 10.

Implications for the Generative Relational Framework

The preceding sections used a small vocabulary drawn from a wider programme of generative relational analysis. This section states what the case returns to that programme. It is a discussion section rather than a second exposition, and it records four results, two of which are corrections.

Neutrality within Generative Condition Governance

The account treated neutrality as a species of a wider genus: governance that acts on the conditions under which relations can be generated rather than on the relations themselves. Section 5.2 distinguished three registers of such action, enabling, sustaining, and refraining, and the case indicates that the third carries most of the weight. A custodian’s characteristic act is declining to remove a condition that it has the power to remove and that removing would serve its interest.

The consequence for the wider programme is that condition governance requires a theory of restraint and not only a theory of provision. What distinguishes a custodian from a provider is not what it supplies but what it declines to withdraw, and an analysis that describes only the enabling register will misdescribe the position.

Generality of the Modality Constraint beyond Neutrality

Claim 9 was derived from the maintenance requirement and not from anything specific to conflict. Its form is general: an actor holding powers that constitute the field within which others act is constrained in the use of those powers, and is not constrained in the influence it exercises inside the field. The constraint therefore states a condition on any actor occupying a constitutive position with respect to others, whether that position is diplomatic, institutional, professional, or editorial.

Two instances outside diplomacy indicate that the form is recognised independently. In telecommunications, a neutral host owns and maintains network infrastructure that it leases to competing operators, and its governing rule is parity of service among tenants combined with the absence of control by any one of them (Telecom Infra Project 2024): an actor holding constitutive powers over a field, constrained in their use and disinterested in which tenant prospers. In psychoanalysis, the concept of neutrality has been placed under pressure by the turn toward relational models of the analytic situation, which have unsettled belief in the analyst’s neutrality and in its desirability as an aim (“Three Papers on the Concept of Neutrality: Editor’s Introduction” 2022). The first is an independent instance of the constraint’s structure; the second is the same difficulty this paper addresses, arising in a discipline with a different vocabulary, and neither is developed here.

Section 8.3 established the corresponding limit and it generalises with the claim. The constraint governs an actor’s conduct toward those whose field it constitutes, and it is silent about that actor’s conduct in every other register. Any transposition of the constraint to another domain inherits this silence, and the transposition should carry the limit with it rather than presenting the constraint as a sufficient ethic of the position.

Foreclosure, Revisability, and the Generative Criterion

Claim 11 identified the point at which the maintenance of a field becomes the mechanism of foreclosure. Its statement in terms of persons rather than parties was required by the case, and the requirement is worth carrying back into the wider framework: a field is populated by those who could enter into relation, and an analysis restricted to those already recognised as parties will exclude precisely the persons whose exclusion is the gravest outcome available.

A second result concerns revisability. The account treated irreversibility as indexed to a horizon and identified the limiting case in which no horizon restores what was foreclosed. That limiting case is what makes foreclosure a distinct category rather than a severe instance of loss, and it supplies the generative criterion with a lower bound: an arrangement that removes the capacity for any future relation is not evaluated by its consequences within a horizon, because the horizon has been eliminated for those it affects.

Neutral Capacity at Organizational and Individual Scale

Section 6.5 treated internal and external custodianship as one principle at two scales. A further scale lies below both. An individual occupying a position between parties in conflict maintains conditions in the same sense, is subject to the same constraint, and is exposed to the same mechanism failures. Two differences are visible from the present case and neither is developed here.

The first is that the resources identified in Section 6.3 are unequally available across scales. A state accumulates hosting capacity, standing mandates, and institutional residency; an individual can accumulate a record and very little else, and a record is more easily destroyed than an institution.

The second is that a state’s custodianship is exercised through an office, whereas an individual’s is exercised through a person. Demands for alignment made on an office are answered by the office’s rules; demands made on a person are answered by the person, one demand at a time. The cost structure differs accordingly, and the consequence for the sustainability of individual custodianship is the subject of the next paper in this series.

Limits of the Account

This section records where the account stops. The list is longer than the customary limitations paragraph because several of the limits were discovered in the course of drafting rather than anticipated.

Conditions of Falsification

Section 4.3 stated five conditions under which the account should be narrowed or withdrawn. Their status after the analysis is as follows.

The requirement that the field be specifiable independently of outcomes is partly met. Proposition 5 supplies a decidable test that does not require enumerating a field and does not refer to outcomes, and Definition 4 states the observables it needs. The test is a lower bound: it detects total foreclosure and not narrowing, it reports possibility and not terms, and the drawing up of the inventory remains a discretionary act. An account whose central object can be measured only at its extreme is measurable in a thin sense, and the reader should weigh the construction accordingly.

The requirement that the modality constraint classify contested conduct without amendment is met, in the strict sense that Section 8.3 obtained determinate answers without amending the constraint. It is met at a price: two of the gravest elements of the record fall outside the constraint’s scope, and the constraint is correspondingly less powerful than its statement suggests.

The requirement that surviving and abandoned neutralities differ in accumulated infrastructure defeated the original form of Claim 10, which is why the claim now states a cost-raising mechanism rather than a prediction of survival. In its weaker form the claim is untested. Section 8.4 identifies four classes of case, including the Austrian comparison in which legal status is held constant, and performs none of them; no measure of accumulated relational infrastructure has been proposed. The weaker claim is also weaker in what it forbids, and a reader may reasonably ask what would now count against it. The answer is evidence that dismantling accumulated infrastructure imposes no cost on the parties that used it, which is the mechanism the claim asserts and the only thing it asserts.

The requirement that the viability formulation support an inference is met more fully than the first draft of this account managed. Definition 3 supplies the per-party and horizon-indexed structure on which Proposition 8 depends, and Proposition 5 gives the constraint set a finite discrete state space over which a kernel can in principle be computed rather than invoked. The formulation nonetheless remains modest beside the use it receives in the literature from which it is drawn (Aubin 1991; Aubin, Bayen, and Saint-Pierre 2011): no kernel is computed here, and the reduction that makes computation possible is the lower bound just described.

The rival explanation in terms of material usefulness to the stronger belligerent remains open.

A sixth limit belongs here although Section 4.3 did not anticipate it. The originality of the combined construction rests on a bounded search, reported in Section 3.7, which returned no prior use of the paper’s central terms in a related sense. A bounded search establishes that a term was not found and not that it does not exist, and the components of the construction are in any case conceded to the traditions recorded in Table [tab:antecedents]. Readers who know an established framework carrying this combination under other terminology are invited to say so. Section 8.3 treats non-invasion as jointly explained and the evidence does not permit more.

Extensions Left to Further Work

Three extensions would materially strengthen the account and each is identified rather than attempted.

A comparative study of violated and abandoned neutralities, including the invasions of the Netherlands and Luxembourg absent from Section 8.4, would test Claim 10 against the class of cases the present selection excludes. The Austrian comparison identified in Section 8.4 is the most tractable of these, because the two states adopted the same legal status and differ in the dimension the claim identifies.

An operationalisation of relational infrastructure would convert Claim 10 into a measurable proposition. The construct to be measured is the aggregate cost that other parties would bear in rebuilding elsewhere the arrangements they currently hold through the custodian; hosting volume, mandate duration, and institutional residency are proxies for that quantity rather than components of it, and mandate duration is the more informative of them, since a mandate held for decades represents an accumulated stake that a recently opened one does not. Two design requirements follow. The outcome variable must separate adjustment of a neutrality policy from abandonment of the status, under a coding rule fixed before the cases are examined. And infrastructure must be measured before the outcome it is used to explain, since durable neutrality generates institutions and the reverse inference would be circular. With four or five cases available, the result would be a structured comparison rather than a statistical test.

A measure of field width, as against the emptiness test that Proposition 5 supplies, would reach the harm that matters most often. Narrowing rather than total foreclosure is the ordinary form of damage to a relational field, and the reduction offered here is blind to it. A candidate direction is to weight contacts by their cost of use and by the number of parties able to use them, which would replace a binary test with a graded one at the price of the discretion that weighting introduces.

An extension of the constraint to conduct toward persons outside the field would address the gap Section 8.3 exposed. This paper does not attempt it, because an extension constructed to cover a case the original failed to reach is the kind of amendment Section 4.3 was written to forbid. The gap should be closed by an argument with its own grounds or left open.

Conclusion

Neutrality is difficult in polarized conflict because a refusal to align is read as an alignment, and the reading has force so long as the refusal is understood as a position between the parties. This paper proposed that what a neutral party maintains is not a position but a field: the set of arrangements under which the parties retain the capacity to form, revise, suspend, and resume relations. The maintenance requirement is that this set not be allowed to empty for any party within a stated horizon.

Two constructions followed. The modality constraint separates the influence a custodian may exercise from the instruments through which it may exercise it: judgment, testimony, and public argument remain available, and the powers that constitute the field are withheld from the pursuit of substantive outcomes. The constraint was derived from the maintenance requirement rather than stipulated, and it answers the objection that neutrality of effect is unattainable, because it makes no claim about effects. The accumulation asymmetry separates a declared position, whose abandonment costs other parties nothing, from relational infrastructure, whose dismantling destroys value that other parties hold and whose accumulation therefore raises the cost of reversal. The asymmetry was initially stated as a prediction about which neutralities survive; the Austrian case defeated that form, and the claim is now confined to the mechanism it can support.

The Swiss case exhibits both. The separation of the law of neutrality from the policy of neutrality supplies a fixed floor beneath an adjustable apparatus; Geneva and the protecting-power mandates are relational infrastructure in the strict sense, built over decades and used by parties other than their holder; and armed neutrality is the boundary that keeps the field beyond unilateral capture. The decisions of 2022 to 2024 exhibit something the account did not anticipate and now states: a decision permissible under the constraint can still cost a custodian its usability, because usability depends on the parties’ willingness and the constraint does not govern that.

The failure analysis divides mechanism failure from normative defeat, and the wartime record tests both. The constraint classifies the custodial conduct and does not reach the gold transactions or the refugee policy, which establishes that it constrains custodianship without constituting an ethics of a custodian’s conduct. Normative defeat was accordingly stated in terms of persons rather than parties, because the gravest failures associated with neutrality have fallen on those who were never parties at all. And the International Committee of the Red Cross had the distinction available to it without this account, and did not act on it, which indicates that the failure was not principally one of concepts.

The account is a revisable proposal. It supplies a definition, a decidable test of the maintenance requirement, a derived constraint, a cost-raising mechanism, and a two-part failure analysis; it supplies no measure of how wide a field is rather than whether it is empty, no comparative test of the mechanism, and no treatment of conduct toward those outside the field. Those are the next requirements, and the account should be judged by whether they can be met.

Acknowledgments

The present definitions, constructions, arguments, conclusions, and errors remain the author’s responsibility.

Al Jazeera. 2022. “Russia Rejects Swiss Plan to Act as a Go-Between with Ukraine.” Al Jazeera.

Arrow, Kenneth J., and Anthony C. Fisher. 1974. “Environmental Preservation, Uncertainty, and Irreversibility.” Quarterly Journal of Economics 88 (2): 312–19.

Aubin, Jean-Pierre. 1991. Viability Theory. Boston: Birkhäuser.

Aubin, Jean-Pierre, Alexandre M. Bayen, and Patrick Saint-Pierre. 2011. Viability Theory: New Directions. 2nd ed. Springer.

Austrian Information. n.d. “Austria’s Permanent Neutrality.” Austrian Press and Information Service.

Burt, Ronald S. 1992. Structural Holes: The Social Structure of Competition. Harvard University Press.

Buterin, Vitalik. 2020. “Credible Neutrality as a Guiding Principle.” Nakamoto.com.

Center for Security Studies, ETH Zurich and Military Academy at ETH Zurich. 2023. “Survey Security 2023: Slight Majority for NATO Rapprochement and a More Critical Assessment of Neutrality.” ETH Zurich.

Centre virtuel de la connaissance sur l’Europe. n.d. “Constitutional Law on the Neutrality of Austria (26 October 1955).” CVCE.

Codevilla, Angelo M. 2000. Between the Alps and a Hard Place: Switzerland in World War II and the Rewriting of History. Regnery.

Coning, Cedric de, Ako Muto, and Rui Saraiva, eds. 2022. Adaptive Mediation and Conflict Resolution: Peace-Making in Colombia, Mozambique, the Philippines, and Syria. Palgrave Macmillan.

“Constitutional Law on the Neutrality of Austria, Adopted 26 October 1955.” 1955. International Constitutional Law, University of Bern.

“Convention (V) Respecting the Rights and Duties of Neutral Powers and Persons in Case of War on Land.” 1907. The Hague, 18 October 1907; entered into force 26 January 1910.

“Convention (XIII) Respecting the Rights and Duties of Neutral Powers in Naval War.” 1907. The Hague, 18 October 1907; entered into force 26 January 1910.

“Divide Et Impera: Conflictual Ties in Broker Relationships.” 2016. In Academy of Management Proceedings.

Dixit, Avinash K., and Robert S. Pindyck. 1994. Investment Under Uncertainty. Princeton University Press.

Dworkin, Ronald. 1985. “Liberalism.” In A Matter of Principle. Harvard University Press.

E-International Relations. 2024. “The Neutralities of Austria and Switzerland: Akin but Not Alike.” E-International Relations.

Eizenstat, Stuart E. 1998. “U.S. And Allied Efforts to Recover and Restore Gold and Other Assets Stolen or Hidden by Germany During World War II.” United States Department of State.

Euronews. 2022a. “Switzerland Blocks German Request to Re-Export Swiss Ammunition to Ukraine.” Euronews.

———. 2022b. “Ukraine War: Switzerland Adopts Entire EU Sanctions Package Against Russia.” Euronews.

———. 2023. “The Red Cross Is Failing Jews Again.” Euronews.

European Council on Foreign Relations. 2024. “Breaking Tradition: Why Russia’s War Is Making Switzerland Question Its Neutrality.” ECFR.

Federal Department of Foreign Affairs. 2026. “Neutrality.” About Switzerland, Swiss Confederation.

———. n.d.a. “Good Offices.” Swiss Confederation.

———. n.d.b. “Protecting Power Mandates.” Swiss Confederation.

Gergen, Kenneth J. 2009. Relational Being: Beyond Self and Community. Oxford University Press.

Halbrook, Stephen P. 1998. Target Switzerland: Swiss Armed Neutrality in World War II. Sarpedon.

Hanemann, W. Michael. 1989. “Information and the Concept of Option Value.” Journal of Environmental Economics and Management 16 (1): 23–37.

Henry, Claude. 1974. “Investment Decisions Under Uncertainty: The Irreversibility Effect.” American Economic Review 64 (6): 1006–12.

Hoffman, Aaron M. 2006. Building Trust: Overcoming Suspicion in International Conflict. State University of New York Press.

House of Commons Library. 2024. “NATO Enlargement: Sweden and Finland.” CBP-9574. UK Parliament.

Imperial War Museums. 2021. “One Story, Different Voices: The Norwegian Campaign, 1940.” Imperial War Museums.

Independent Commission of Experts Switzerland–Second World War. 2002. Switzerland, National Socialism and the Second World War: Final Report. Zurich: Pendo Verlag.

“Infrastructuring Digital Humanities: On Relational Infrastructure and Global Reconfiguration of the Field.” 2022. Digital Scholarship in the Humanities 37 (2): 534–34.

International Committee of the Red Cross. 2024. “Principles Under Pressure: Have Humanitarian Principles Really Stood the Test of Time?” ICRC Humanitarian Law and Policy Blog.

Isaksson, Erik. 2024. “Identity Continuities, Far-Right Acquiescence, and the ‘New’ and the ‘Old’: Finnish and Swedish NATO Accession and Neutrality.” Austrian Institute for International Affairs (oiip).

Kydd, Andrew H. 2003. “Which Side Are You on? Bias, Credibility, and Mediation.” American Journal of Political Science 47 (4): 597–611. https://doi.org/10.1111/1540-5907.00042.

———. 2005. Trust and Mistrust in International Relations. Princeton University Press.

Larmore, Charles. 1987. Patterns of Moral Complexity. Cambridge University Press.

Larson, Deborah Welch. 1997. Anatomy of Mistrust: U.S.–Soviet Relations During the Cold War. Cornell University Press.

Lehmbruch, Gerhard. 1967. Proporzdemokratie: Politisches System Und Politische Kultur in Der Schweiz Und in Österreich. Tübingen: J.C.B. Mohr (Paul Siebeck).

Lijphart, Arend. 1969. “Consociational Democracy.” World Politics 21 (2): 207–25.

———. 1977. Democracy in Plural Societies: A Comparative Exploration. Yale University Press.

Luhmann, Niklas. 1979. Trust and Power. John Wiley.

———. 1988. “Familiarity, Confidence, Trust: Problems and Alternatives.” In Trust: Making and Breaking Cooperative Relations, edited by Diego Gambetta, 94–107. Blackwell.

Mensink, Paul, and Till Requate. 2005. “The Dixit–Pindyck and the Arrow–Fisher–Hanemann–Henry Option Values Are Not Equivalent: A Note on Fisher (2000).” Resource and Energy Economics 27 (1): 83–88.

Ostrom, Elinor. 1990. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

Pictet, Jean. 1979. The Fundamental Principles of the Red Cross: Commentary. Geneva: Henry Dunant Institute.

“Press Conference Given by the President of the ICRC (Geneva, 30 May 1995).” 1995. International Review of the Red Cross.

Rawls, John. 1993. Political Liberalism. Columbia University Press.

Raz, Joseph. 1986. The Morality of Freedom. Oxford University Press.

Shelby, Karen D. 2014. “The Battle of the Golden Spurs: Enlistment Propaganda and the Front Movement.” In Flemish Nationalism and the Great War, 65–82. London: Palgrave Macmillan. https://doi.org/10.1057/9781137391735_4.

Simmel, Georg. 1950. The Sociology of Georg Simmel. Edited by Kurt H. Wolff. Glencoe, IL: The Free Press.

Slim, Hugo. 1997. “Relief Agencies and Moral Standing in War: Principles of Humanity, Neutrality, Impartiality and Solidarity.” Development in Practice 7 (4): 342–52.

———. 2015. Humanitarian Ethics: A Guide to the Morality of Aid in War and Disaster. Hurst; Oxford University Press.

Steinacher, Gerald. 2017. Humanitarians at War: The Red Cross in the Shadow of the Holocaust. Oxford University Press.

Steinberg, Jonathan. 1996. Why Switzerland? 2nd ed. Cambridge University Press.

Steiner, Jürg. 1974. Amicable Agreement Versus Majority Rule: Conflict Resolution in Switzerland. University of North Carolina Press.

Svensson, Isak. 2007. “Bargaining, Bias and Peace Brokers: How Rebels Commit to Peace.” Journal of Peace Research 44 (2): 177–94.

———. 2009. “Who Brings Which Peace? Neutral Versus Biased Mediation and Institutional Peace Arrangements in Civil Wars.” Journal of Conflict Resolution 53 (3): 446–69.

———. 2015. International Mediation Bias and Peacemaking: Taking Sides in Civil Wars. Routledge.

Swiss Federal Council. n.d. “The Federal Council Approves a Report on Switzerland’s Good Offices.” Media release, Swiss Confederation.

swissinfo.ch. 2023. “Small Majority of Swiss Back Re-Exporting Arms to Ukraine.” SWI swissinfo.ch.

———. 2024. “Why Switzerland Stopped Short of Formally Sanctioning Ukraine.” SWI swissinfo.ch.

———. n.d.a. “Switzerland Loses Two Mandates as Protecting Power.” SWI swissinfo.ch.

———. n.d.b. “The Day Switzerland Became Neutral.” SWI swissinfo.ch.

Telecom Infra Project. 2024. “Neutral Host Infrastructure Sharing Framework.” Telecom Infra Project.

“The Law of Neutrality.” n.d. In International Law Studies. United States Naval War College.

“Three Papers on the Concept of Neutrality: Editor’s Introduction.” 2022. The Psychoanalytic Quarterly. https://doi.org/10.1080/00332828.2022.2097559.

Touval, Saadia, and I. William Zartman, eds. 1985. International Mediation in Theory and Practice. Westview Press.

United Nations Office at Geneva. n.d.a. “Palais Des Nations.” ungeneva.org.

———. n.d.b. “The UN in Geneva.” ungeneva.org.

Urner, Klaus. 2001. “Let’s Swallow Switzerland”: Hitler’s Plans Against the Swiss Confederation. Lexington Books.

Vagts, Detlev F. 1997. “Switzerland, International Law and World War II.” American Journal of International Law 91 (3): 466–75.

Vatter, Adrian. 2016. “Switzerland on the Road from a Consociational to a Centrifugal Democracy?” Swiss Political Science Review 22 (1). https://doi.org/10.1111/spsr.12203.

Waldron, Jeremy. 1993. “Legislation and Moral Neutrality.” In Liberal Rights: Collected Papers 1981–1991. Cambridge University Press.

Wall, Steven. 2010. “Neutralism for Perfectionists: The Case of Restricted State Neutrality.” Ethics 120 (2): 232–56.

Weinberg, Gerhard L. 1999. “German Plans and Policies Regarding Neutral Nations in World War II with Special Reference to Switzerland.” German Studies Review 22 (1): 99–103.

Wheeler, Nicholas J. 2018. Trusting Enemies: Interpersonal Relationships in International Conflict. Oxford University Press.

Zartman, I. William. 1985. Ripe for Resolution: Conflict and Intervention in Africa. Oxford University Press.

Zartman, I. William, and Saadia Touval. 1996. “International Mediation in the Post-Cold War Era.” In Managing Global Chaos: Sources of and Responses to International Conflict, edited by Chester A. Crocker, Fen Osler Hampson, and Pamela Aall, 445–61. United States Institute of Peace Press.