Constitutions in Motion: How Collective-Action Systems Adapt Authority Without Losing the Capacity to Act

Central finding

Collective-action systems do not adapt successfully by choosing the correct point on a single scale from centralized to decentralized, democratic to hierarchical, consensual to majoritarian, or unitary to polycentric. The comparative evidence does not identify a generally superior position on any of those dimensions. Nor does it show that higher-order authority allocation, bounded amendment procedures, multi-level review, or reversible delegation are either necessary or sufficient for successful adaptation.

Formal safeguards are not necessary because authority has been constrained or reversed without pre-authorized constitutional channels: Bitcoin node operators enforced a date-bounded rule against a coalition of major firms and miners; Wikimedia communities forced the withdrawal of an infrastructure privilege for which no formal appeal existed; the Node.js fork ended in reunification under reconstructed governance; and OpenAI’s formally supreme nonprofit board was reversed by coordinated employee exit backed by an external employment sponsor.¹

They are not sufficient because Cyprus’s 1960 constitution combined an unamendable core with bicommunal voting requirements and nonetheless broke down within three years; Hungary concentrated emergency authority through valid constitutional amendment; and the United States’ National Emergencies Act combines annual expiration with a congressional review duty, yet 52 emergencies remained in effect as of September 2026 and the oldest had been renewed for nearly 47 years.²

The strongest recurring lesson lies below the level of these familiar mechanism families. A potentially consequential safeguard combines:

1. A named actor with standing to initiate or contest change;
2. A deadline that arrives without the power-holder’s cooperation;
3. A default adverse to continuation when the deadline arrives; and
4. A renewal act that is costly or requires actors other than the holder of the temporary power.

This structure appears in India’s post-Emergency constitutional reforms, Israel’s limits on emergency regulations, the IETF’s bounded variance process, and Bitcoin’s BIP 148 flag-day rule. But it remains a candidate design principle, not a demonstrated general solution: among these four, only BIP 148 has a retrieved record of activation and effect.

The more securely established conclusion is descriptive. Every system has both a formal constitution and a practical constitution. The latter consists of whoever controls agendas, implementation, enforcement, money, labor, infrastructure, credentials, client defaults, intellectual property, and viable exit. Adaptation succeeds or fails at the interface between the two.

---

1. The recursive constitutional problem

Constitutional adaptation is difficult because it is recursive: existing authority determines who may revise authority. A legislature votes on the limits of legislation; a court decides whether the amendment power is limited; a board controls the process for replacing the board; a software project’s maintainers decide which governance proposal enters the release distributed to users.

This recursion creates two distinct but interacting questions:

• Authorization: Who is entitled to initiate, approve, interpret, contest, and reverse a change?
• Control: Who possesses the resources needed to make the decision effective?

Formal amendment answers the first question through a recognized rule-changing procedure. De facto evolution changes practical decision rights without completing that procedure. It may occur through judicial doctrine, executive practice, convention, control of infrastructure, organizational dependence, or accumulated precedent.

The distinction is not simply written versus unwritten. Bitcoin’s BIP 148 was written and precisely time-bounded, but it was not enacted by an apex constitutional authority. A software permission can silently change the hierarchy of a community without amending any charter. Conversely, a judicial doctrine may become an enduring higher-order constraint even though no constitutional text assigned the court that role.

Cross-national research reinforces this distinction. Ginsburg and Melton found that measures of formal amendment difficulty are poorly correlated with one another and that prior amendment behavior—an “amendment culture”—explains observed patterns better than the institutional indices they tested. Their conclusion is not that institutions are irrelevant, but that institutional explanations are partial.³ Research on informal constitutional change likewise documents fundamental shifts outside formal amendment while leaving their legitimacy unresolved.⁴

The relevant constitutional architecture therefore includes at least four layers:

1. Declared authority: charters, constitutions, bylaws, statutes, and governance documents;
2. Procedural authority: agenda control, standing, voting thresholds, appeal routes, interpretation, and review;
3. Operational authority: enforcement, repository access, release channels, client defaults, employment, finance, and infrastructure;
4. Residual authority: emergency action, exit, fork, secession, litigation, protest, and refusal to comply.

A design can appear distributed at the first layer while remaining concentrated at the third. Conversely, actors excluded from the formal constitution may hold an effective veto because the organization cannot function without their labor, software adoption, capital, or cooperation.

---

2. What the evidence rules out

No mechanism family is necessary

Several systems changed direction without a pre-existing appeal or amendment path.

In 2017, Bitcoin’s BIP 148 specified a start date, end date, rejection rule, and automatic extinction once SegWit locked in. Its enforcement depended on economic nodes rather than a constitutional legislature. Later that year, SegWit2x’s sponsors—including major firms and miners—suspended their planned hard fork after acknowledging that they lacked sufficient consensus. The event demonstrates a distributed practical veto, although the evidence does not isolate the relative influence of maintainers, node operators, exchanges, miners, or businesses.⁵

Wikimedia’s Superprotect permission produced the opposite type of episode. Foundation staff created a technical privilege capable of overriding local administrators, without a recognized community amendment. Community protest and reputational pressure eventually secured its removal. The rollback shows that an informal check can reverse constitutional drift, but only after a legitimacy conflict, and without creating a binding rule against recurrence.⁶

OpenAI provides an organizational analogue. The nonprofit board validly exercised its broad formal discretion to remove the chief executive, but did so through a process later described in the company’s commissioned review as abridged and lacking notice, full inquiry, and an opportunity to respond. Hundreds of employees threatened to leave for Microsoft, providing the practical lever that reversed the board. The review also reported that the removal was not about safety, mission, product pace, finances, or investor communications. The episode therefore should not be described as capital defeating a mission-protective decision. It shows, more narrowly, that formal authority can be reversed through labor dependence and an externally supplied exit option.⁷

These cases do not prove that informal adaptation performs better than formal procedure. They show only that formal higher-order pathways are not indispensable to reversal.

No mechanism family is sufficient

Cyprus supplied an unusually strong form of constitutional commitment. Article 182 rendered the constitution’s Zurich-derived basic articles unamendable, while Article 78 required separate communal majorities for specified legislation; other amendments required two-thirds support within each community. The constitutional order broke down in 1963. Ethnic conflict and external intervention prevent a clean causal attribution, but the case decisively shows that entrenched cores and minority-protective amendment thresholds do not guarantee adaptation or stability.⁸

Hungary demonstrates a different failure: concentration through valid amendment rather than evasion of it. Its revised emergency architecture permits extensions bounded to 180 days per authorization but does not limit repetition. Peer-reviewed analysis reports that, with only a brief interruption, Hungary operated under emergency rule for roughly five years from 2020, issuing more than 1,000 emergency decrees.⁹

The United States supplies the strongest example of a formally sound-looking sunset that fails to force reconsideration. The National Emergencies Act directs each chamber of Congress to meet periodically to consider termination and provides that an emergency expires on its anniversary unless the president republishes a continuation notice. Yet, of 92 declarations since 1976, 52 remained in force as of September 16, 2026. The oldest, dating to November 1979, had been renewed annually for nearly 47 years.¹⁰ The problem is not the absence of a date or adverse default. Renewal is unilateral and cheap for the same office holding the power.

No general degree of centralization, participation, or veto power prevails

The best identified evidence often reveals tradeoffs rather than rankings. Denmark’s 2007 municipal restructuring created a quasi-experiment: 239 municipalities were amalgamated while 32 were left untouched. One study found considerable administrative scale economies per inhabitant; another found a causal and sizeable reduction in citizens’ internal political efficacy. A third documented a pre-merger common-pool spending surge. These are studies of one reform, not independent replications, but together they show that greater scale improved one outcome while worsening another.¹¹

A study of approximately 5,600 policy reforms in 13 Western European countries found no support for veto-player theory in its pure form, although the prediction survived for minimal winning cabinets. Crisis pressure and distance from the status quo performed better as general explanations.¹² A 101-country study similarly undercut earlier support for consociational arrangements, finding that proportional representation and parliamentarism were associated with more political violence at high levels of ethnic fractionalization, while the federalism result was less certain. Its outcome is conflict, however, not policy paralysis, so it does not directly settle debates about minority vetoes.¹³

The commons literature is equally conditional. Cox, Arnold, and Villamayor-Tomás coded 91 studies and 77 common-pool-resource cases. Ostrom’s design principles were supported overall, but the two cross-level components—external recognition of local rule-making rights and nested enterprises—were the weakest of eleven on reported effect size. Nested enterprises had a supportive-to-unsupportive ratio of 2.0 and φ=0.262; accountable monitoring, the strongest component, had a ratio of 11.7 and φ=0.792. The authors nevertheless considered the principles robust within their evidence base and explicitly declined to assume that they transfer across scales.¹⁴

The conclusion is not that federalism, polycentricity, subsidiarity, participation, or hierarchy fail. It is that their effects depend on jurisdictional boundaries, capacity, transfer of real discretion, accountability, and how the layers interact.

---

3. A comparative map of constitutional functions

The same constitutional functions appear in states, commons, standards bodies, software communities, firms, platforms, and decentralized protocols. Digitally mediated governance is not constitutionally thin: some digital institutions specify amendment, appeal, emergency, entrenchment, and unit-dissolution procedures more explicitly than many states. What is generally missing is evidence that these provisions work as intended.

Constitutional function	Recurring mechanisms	Representative implementations and failures	Evidentiary position
Revision of governing rules	Supermajorities, double majorities, two-key approval, entrenched cores, judicial limits	Australia’s double-majority referendum; India’s Article 368 and judicial basic-structure doctrine; Cyprus’s unamendable core; Debian’s 3:1 constitutional vote; Python’s two-thirds amendment vote; ICANN’s three-quarters board vote plus Empowered Community approval	Strong evidence that rules exist; formal difficulty is a weak predictor of actual amendment behavior; little comparative effectiveness evidence
Jurisdiction and subsidiarity	Local rule-making, nested units, boundary rules, higher-level review	Commons governance; Maine lobster zones; EU subsidiarity review; Kubernetes SIGs and repository-scoped authority	Effects are conditional; Maine shows that delegation can move conflict to borders
Delegation and revocation	Scoped delegation, retained residual authority, no-confidence rules, principal review	Maine zone councils; Debian delegates; Kubernetes Steering Committee; OpenTofu TSC; Meta Oversight Board	Framework evidence is rich; only Maine supplies a substantial longitudinal exercise record
Vetoes and checks	Communal vetoes, bicameralism, distributed technical non-adoption, board recall	Cyprus and Bosnia; Bitcoin node non-adoption; ICANN board recall	Veto paralysis is documented but causally entangled; large-N veto theory is conditional rather than general
Appeal and dispute resolution	Multi-tier appeals, annulment, judicial review, independent panels	IETF appeal ladder; ICANN Independent Review Process; Debian Technical Committee; Meta Oversight Board; litigation over Maine zone decisions	Mostly implemented frameworks; exercise and user-level outcome evidence is sparse
Membership and standing	Reserved seats, contributor ladders, employer caps, restricted electorates, minority initiation	India’s reserved village-council leadership; Kubernetes scoped membership; Python employer cap; EU parliamentary-minority standing	Strong operational evidence in one randomized membership case; little evidence about constitutional revision
Sunset and periodic review	Automatic lapse, recurring review, continuing-purpose tests	U.S. National Emergencies Act; Israel’s three-month emergency-regulation limit; ICANN’s five-year reviews; statutory sunset regimes	Mixed: some lapses occur, but renewal by inertia is common; most documented mechanisms lack exercise records
Emergency authority and reversion	Declaration limits, supermajority renewal, minority requisition, temporary variance, kill switches	India’s 44th Amendment; Israel’s Basic Law; RFC 2026; MakerDAO Emergency Shutdown Module; Hungary’s repeatable emergency regime	Strong framework evidence and strong evidence of emergency persistence; weak comparative evidence on which safeguards prevent it
Exit, fork, and secession	Negotiated secession, textual exit, software fork, competing chain	Quebec Reference; Ethiopia Article 39; Node.js/io.js; OpenTofu; Ethereum Classic	Exit can discipline, restructure, or fragment; successful-fork evidence is heavily selected
Creation, dissolution, and restructuring of units	Amalgamation, retirement, divestiture, operator replacement	Danish municipalities; Kubernetes repository lifecycle; ICANN’s IANA-operator separation process	Denmark shows a measured efficiency–efficacy tradeoff; digital provisions largely lack exercise records

The map also reveals a recurrent interface failure. Jurisdictional decentralization can be formally generous while the center retains funding, permissions, personnel, or enforcement. Appellate authority can be binding in selected cases while the delegator controls the information and technical implementation. Fork rights may exist in a license but remain unusable without maintainers, identity, distribution, and infrastructure.

---

4. Standing matters—but standing on paper is not enough

A constitutional architecture does more than specify voting thresholds. It distributes distinct roles in the revision process:

System	Who may initiate?	Who approves?	Who contests or interprets?	Who can reverse?
India—emergency powers	Cabinet by written advice; 10% of the Lok Sabha may force a disapproval meeting	Both houses at the constitutional-amendment majority, six months at a time	Parliament and courts	Parliament by resolution; courts through constitutional doctrine
Canada—secession	Each participant in the federation may initiate constitutional change	Constitutional amendment following negotiation	Political actors determine clarity; courts define the framework	No unilateral route; negotiated amendment
EU—subsidiarity	National parliamentary chambers; in one case, 60 of 577 French deputies	Commission may maintain, amend, or withdraw	National parliaments and the Court of Justice	Proposal withdrawal or judicial remedy
Maine lobster governance	Zone councils and legislature	Two-thirds of voting license holders, subject to commissioner review	Licensees, commissioner, courts	Litigation, legislation, or inter-zone negotiation
Debian	Any developer may propose or sponsor a General Resolution	Developers, with elevated thresholds for constitutional change and overrides	Secretary and Technical Committee	Collective override, leader recall, or withdrawal of ongoing delegations
Python	Core team	Steering council for policy; two-thirds of core team for constitutional amendment	Core team and council	Individual removal or whole-council dissolution by no-confidence vote
ICANN	Empowered Community participants and eligible claimants	Board plus Empowered Community for fundamental changes	Independent Review Process and community escalation	Director removal, board recall, rejection of budgets or plans
IETF	Any participant may raise objections or appeal	Rough consensus and responsible standards bodies	Working-group chair, Area Director, IESG, IAB, and ultimately ISOC trustees on procedural adequacy	IAB annulment can restore the prior position
MakerDAO	Anyone may call the shutdown function once the token threshold is reached	The threshold is met through irrecoverably burned MKR	Token holders and governance	Governance may disarm the module; fork remains the residual remedy
OpenAI in 2023	Board	Board	Employees lacked a formal contest channel	Coordinated labor exit with an external sponsor produced practical reversal

Minority standing to initiate review recurs across otherwise dissimilar systems: 10% of one parliamentary chamber, 60 deputies, any Debian developer, any IETF participant, or any token holder willing to incur a cost. Yet the evidence rarely follows these rights into practice. The presence of standing makes the system more inspectable; it does not establish that invoking the right is affordable, safe, timely, or consequential.

When formal standing is absent, contestation often follows the resource on which the system depends. Employees use exit, communities use protest, node operators refuse software, and local actors litigate boundaries. Practical authority therefore tracks the control stack as well as the charter.

---

5. The most promising recurring structure: standing, clock, default, and costly renewal

A sunset is commonly treated as one mechanism. The evidence instead suggests that it should be decomposed into four properties.

1. A named actor has standing

Someone must be able to place continuation or termination on the agenda. India’s 44th Amendment gives 10% of Lok Sabha members the right to requisition a special meeting to consider disapproval of an emergency. The EU’s subsidiarity process can be activated by parliamentary minorities. The IETF allows affected participants to appeal. MakerDAO allows anyone to call its emergency shutdown once the required amount of MKR has been irreversibly committed.

2. A clock runs independently of the power-holder

The relevant date should arrive without a discretionary decision to schedule review. Examples include an anniversary, a three-month expiration, a flag day, and a five-year continuing-purpose review.

3. The default is adverse to continuation

If no valid action occurs, the extraordinary authority should terminate, revert to the ordinary procedure, or extinguish itself. RFC 2026 treats variance as one-time only and requires permanent change to use the normal standards process. It also prevents the exception procedure from being used against the appeal clause or against itself—an entrenched protection of revisability.¹⁵

4. Renewal requires cost or breadth beyond the power-holder

This is the feature most visibly missing from the U.S. National Emergencies Act. The presidential republication needed for continuation is cheap and unilateral. India’s 44th Amendment instead requires the same parliamentary majority used for constitutional amendment. Israel separates the layers: the Knesset may repeatedly renew an emergency declaration, but emergency regulations expire after three months unless extended by legislation and lapse no later than 60 days after the emergency ends.¹⁶

This distinction between emergency status and emergency measures is important. A polity may tolerate a continuing legal status while forcing individual measures to expire; or it may sunset status while allowing durable policies enacted under it to survive. Measuring only declarations or only decrees obscures where the ratchet operates.

The four-part structure is a plausible design hypothesis because it recurs in an Indian constitutional amendment, an Israeli basic law, an Internet standards procedure, and a Bitcoin coordination rule. It is not yet a proven principle. No exercise record was retrieved for the Indian renewal and minority-requisition provisions, Israel’s three-month rule, or the IETF variance process. The Bitcoin case is the only one in this group shown to have fired and affected an outcome.

The broader sunset literature reinforces the caution. A 2025 study reports that earlier empirical work had not shown sunset legislation to improve efficiency or reduce waste and that agencies were often renewed through the inertia the mechanism was designed to defeat. Its own positive GDP estimates were fragile under synthetic-control analysis, with robust findings only for Tennessee.¹⁷

---

6. Adaptation pathways and their characteristic risks

Formal amendment: legitimacy and commitment, with risks of rigidity or abuse

Formal amendment provides notice, recognized standing, and an authoritative record. It can preserve continuity by channeling conflict into an agreed procedure. Python’s transition after Guido van Rossum’s resignation is an illustrative implementation: 62 of 94 authorized core developers selected a governance model, replacing personal authority with an elected steering council subject to constitutional amendment and no-confidence rules.¹⁸

But formal pathways can be too rigid, as Cyprus illustrates, or usable by incumbents to weaken constraints, as Hungary illustrates. India’s basic-structure doctrine adds a further recursion: the higher-order constraint on amendment was judicially constructed rather than explicitly granted. The Supreme Court held that Parliament could not destroy the constitution’s identity or delegate its constituent amendment power to another legislature.¹⁹ Such doctrines may protect the constitutional order, but this evidence does not establish their comparative effects on legitimacy or stability.

Multi-level review: sometimes political signaling rather than legal veto

The EU’s subsidiarity “yellow card” has been triggered three times in roughly 15 years. In the Monti II episode, the Commission concluded that the proposal did not violate subsidiarity but withdrew it after determining that sufficient political support was unlikely. The other two proposals were maintained with reasons; the stronger “orange card” has never been used. As of a 2021 European Parliament working paper, the Court of Justice had not annulled a measure on subsidiarity grounds.²⁰

The mechanism therefore was rarely triggered but decisive once in three cases—not through legal compulsion, but by altering the Commission’s political forecast. This is a reminder that constitutional mechanisms can work indirectly. A formally nonbinding review can matter if it makes future defeat, noncompliance, or legitimacy loss credible.

Bounded delegation: workable, but conflict moves to the boundary

Maine’s lobster co-management system delegated rule-making to elected zone councils, required a two-thirds referendum among specified license holders, retained commissioner review, and preserved legislative authority. It broke a legislative deadlock that had persisted through 17 legislatures. Yet limited entry made zone boundaries consequential, producing disputes in all seven zones, litigation reversing one zone’s decision, negotiated buffer zones, and renewed legislative intervention.²¹

This is one of the few cases with a substantial exercise record. Its lesson is not that reversible delegation resolves conflict. It can move conflict from a central legislature to questions about constituency, territory, boundaries, and appellate authority.

Kubernetes, Debian, OpenTofu, and ICANN contain similar formal compositions of scoped delegation and residual authority. Kubernetes delegates technical responsibility to specialized groups while the Steering Committee retains undelegated authority over repositories and project assets. ICANN combines community powers, board recall, an independent-review framework, five-year continuing-purpose reviews, and a process for replacing the operator of the IANA naming function.²² These are sophisticated constitutions, but no systematic exercise record was retrieved to show whether the mechanisms produce legitimate or timely outcomes.

Informal adaptation: flexibility or drift depending on the control point

Informality is valuable when the formal route cannot address the actual conflict or has no authority over the relevant resource. The Monti II signal, BIP 148, the Node.js fork and reunification, and the rollback of Superprotect all illustrate adaptation outside an apex amendment process.

The same flexibility becomes constitutional drift when a hidden or unreviewable control point acquires a superior decision right. Examples include software permissions that override community administrators, a client default translating a nonrepresentative signal into implementation, a unilateral license change enabled by ownership, or nominal decentralization to bodies that lack real discretion.

The evidence does not yield a validated rule for distinguishing flexibility from drift. It does support a practical diagnostic:

• Was the change visible and reasoned?
• Were affected parties able to contest it?
• Was the practical decision-maker identifiable?
• Could the change be reversed without destroying the shared capability?
• Did implementation resources follow the formally authorized decision?
• Was the informal action later regularized through a legitimate process?

These are design implications, not measured predictors of success.

Exit, fork, and secession: a check that may also destroy the common good

The Supreme Court of Canada’s Quebec Reference deliberately separated democratic legitimacy from legal effect. A clear majority on a clear question would confer democratic legitimacy and trigger a reciprocal duty to negotiate, but would not itself produce secession; legal exit would require constitutional amendment. The Court left “clarity” to political actors. This is a carefully constructed procedural pathway, but it remains untested under the condition that matters: an actual clear-majority referendum for departure.²³

Open-source forks supply more observable cases. The Node.js/io.js split ended in reunification under a neutral foundation and revised technical governance. Following HashiCorp’s license change, OpenTofu combined portable code with maintainers, corporate participation, Linux Foundation infrastructure, and replacement governance.²⁴ These cases suggest that exit is credible only when a coalition can carry not merely source code but also labor, identity, infrastructure, distribution, and legitimacy.

The selection problem is severe. The evidence contains successful forks but no comparable set of failed forks or communities that acquiesced after relicensing. Forking can discipline an incumbent, reconstruct governance, or preserve dissent; it can also duplicate maintenance and permanently divide the network, as Ethereum Classic demonstrates.

Membership: formal inclusion can alter decisions, but does not eliminate substitution

India’s reservation of village-council leadership positions for women provides the strongest identified evidence that a constitutional membership rule changed operational outputs. Because the reserved councils were randomly selected, Chattopadhyay and Duflo could attribute changes in public-goods investment to the rule: investment shifted toward goods for which women expressed greater demand, including drinking water.²⁵

The study concerns service outputs in two districts, not constitutional revision, legitimacy, or long-run institutional stability. It also found evidence of informal substitution—some reserved officeholders were spouses of former male council heads. Formal inclusion can alter outcomes without fully displacing pre-existing networks of influence.

Appeals: authority can be delegated without transferring sovereignty

The IETF’s written process permits appeals from a working-group chair through the Area Director, the Internet Engineering Steering Group, and the Internet Architecture Board, which may annul a decision and restore the prior position. A final route to the Internet Society trustees concerns claims that the procedures themselves are inadequate.²⁶

Meta’s Oversight Board is a narrower organizational example. Its decisions in selected cases are binding, but policy recommendations are advisory; Meta continues to control implementation, information supply, funding, and part of the charter-amendment process. In 2023 the Board reported 398,597 appeals and only 53 decisions. Its high overturn rate cannot establish platform-wide error or redress because the docket is small and nonrandom, and the implementation figures are the Board’s own account.²⁷

Appeal design must therefore be evaluated by its jurisdiction, access, remedy, implementation dependence, and capacity—not merely by the existence of an ostensibly independent body.

---

7. The recurring failure modes

Formal participation with de facto capture

The most consistently observed failure is not an absence of participatory rules. It is the coexistence of those rules with concentrated practical control. The pattern appears in commons whose collective-choice institutions remain formally intact while powerful local or bureaucratic actors control them; administrative decentralization that transfers obligations without discretion; cooperatives whose elected bodies lack the expertise to challenge management; and digital systems controlled through credentials, licenses, trademarks, infrastructure, or employment.

This is a recurrent pattern across unlike cases, not an estimate of frequency. Capture is also relative: local governance may be less captured than national governance in some settings. More participation is not always the remedy. Following some cooperative failures, proposed reforms shifted authority toward skills-based boards or delegate assemblies rather than direct member voting. The cooperative evidence is especially weak, however: no bylaws, amendment histories, board minutes, or congress resolutions were retrieved, and the Fagor material is secondary.

Emergency permanence

Emergency powers exhibit a ratchet particularly at the level of legal status. The United States combines automatic expiration with cheap unilateral renewal. Israel’s text permits repeated declarations while placing shorter limits on regulations. Hungary demonstrates repeated or prolonged authorization. Historical Weimar experience shows decrees substituting for parliamentary government.

Counterevidence matters. Ginsburg and Versteeg’s survey of more than 100 national responses to COVID-19 concluded that courts, legislatures, and subnational governments often constrained national executives and that pandemic governance was closer to a checks-and-balances model than theories of an unbound executive would predict. Their own qualification is that conventional emergency theory is built largely on national-security crises and may not fit pandemics.²⁸ Emergency permanence is therefore well attested, but not every crisis produces unconstrained centralization.

Rigidity and veto paralysis

Cyprus is the strongest rigidity case because the disputed constitutional structure included provisions that could not lawfully be changed. Bosnia supplies a further case of minority-protective vetoes coinciding with prolonged inability to implement a human-rights judgment. Neither supports a simple causal claim: ethnic conflict, external intervention, and substantive disagreement are inseparable from the constitutional machinery.

The evidence is thinner on the opposite failure—destabilization through excessive formal amendability. This may be because mutability manifests as capture or abusive constitutionalism rather than as a separately labeled failure, or because the evidence base searched more directly for rigidity and emergency permanence.

Fragmented decentralization

Decentralization fails when authority, capacity, finance, and accountability do not move together. A health-system synthesis covering 51 studies in 25 countries found mixed results, conditional on institutional capacity, equalization transfers, explicit accountability, and combinations of centralization and decentralization.²⁹ Its health-sector scope should not be generalized to governance as a whole, but its mechanism is consistent with the Maine case: local authority without workable interfaces merely relocates conflict.

Exit without an adequate successor

A formal right of exit does not itself produce an orderly transition. The ability to leave may be obstructed in practice, may intensify conflict, or may divide assets and infrastructure that cannot be duplicated. Conversely, a costly but credible possibility of exit can force incumbents to negotiate. The outcome depends less on the textual right than on procedural clarity, portability, successor capacity, and whether reunification or federation remains possible.

---

8. The tradeoffs are real, but there is no general exchange rate

Adaptability versus commitment

Entrenchment protects expectations but can render a system incapable of lawful response. Easy revision increases responsiveness but may facilitate incumbent self-dealing. The constitutional-endurance literature indicates that design affects mortality, but the present evidence does not support a reliable general rate at which flexibility should be exchanged for commitment.

Local autonomy versus coherence

Maine’s delegation enabled decisions that the legislature could not make, then created new boundary and distributional disputes. Digital systems make the same trade: specialized groups can govern technical domains, while residual authority over shared repositories, brands, or infrastructure is retained to preserve coherence.

Decisiveness versus constraint

Ethereum’s DAO response depended on a small group’s ability to package and implement an option under time pressure; the residue was a permanent chain split and an unresolved question about the legitimate electorate. OpenAI’s rapid reversal preserved organizational operations while bypassing a transparent process for contesting the board. The evidence suggests—not proves—that concentration during a crisis may be compatible with legitimacy only when accompanied by disclosure, contestability, and reversion.

Continuity versus renewal

Node.js preserved the project by forking and later reunifying under revised governance. Python preserved technical continuity while replacing personal authority with an amendable constitution. Hungary preserved legal continuity while changing the substantive character of emergency government. Institutional survival and preservation of constitutional character are therefore different outcomes.

Formal legitimacy versus practical effectiveness

OpenAI’s board action was formally authorized but procedurally defective and practically unsustainable. The employee response was practically effective but constitutionally unstructured. The Quebec Reference offers a rare deliberate treatment of the gap: democratic legitimacy can obligate negotiation without creating immediate legal effect.

Scale economies versus democratic experience

Denmark’s restructuring produced considerable administrative economies alongside a sizeable decline in citizens’ internal political efficacy. There is no evidence-based aggregate verdict that converts these outcomes into a single measure of institutional success.

---

9. Design implications and testable possibilities

The evidence does not justify a universal constitutional template. It does support several design possibilities that can be stated as hypotheses for implementation and evaluation.

Map the real control stack

A constitutional review should inventory not only voting rules but also:

• Agenda and interpretation rights;
• Enforcement and implementation authority;
• Repository, account, and credential ownership;
• Funding and employment dependencies;
• Intellectual property, trademarks, and distribution channels;
• Data and platform access;
• Practical vetoes and viable exit destinations.

This reveals where formal authority may fail to control implementation and where formally excluded actors possess an unacknowledged veto.

Specify standing, clock, default, and renewal separately

A sunset clause is incomplete unless it answers:

• Who can force the question?
• When does the deadline arrive?
• What happens without action?
• Who must approve continuation?
• What costs do renewal proponents bear?

The U.S. emergency record suggests that a date and adverse default can be neutralized by cheap unilateral renewal. India’s post-Emergency design suggests a stronger composition, but its operation still needs to be studied.

Separate emergency status from emergency measures

Different clocks may be appropriate for the declaration, each implementing measure, and the residual effects after termination. Israel’s Basic Law demonstrates the feasibility of this separation. Whether it produces better outcomes remains an empirical question.

Couple subsidiarity to explicit residual authority and boundary procedures

Delegation should specify:

• The scope of the local unit;
• Who resolves overlapping jurisdiction;
• The conditions for central intervention;
• The appeal route;
• How the delegation is revoked;
• How boundaries and constituencies are revised;
• How obsolete units are dissolved.

Maine shows why boundary and re-entry rules are not secondary details. Kubernetes and ICANN provide detailed written models whose actual use should be evaluated.

Preserve restoration, not merely appeal

An appeal body that can issue reasons but cannot restore the prior state may offer transparency without an effective remedy. The IETF’s explicit annulment and status-quo restoration power is a notable model, though its performance record was not retrieved.

Treat exit as an institutional ecosystem

Fork or secession rights should be assessed against the portability of assets, labor, identity, infrastructure, and data. Where exit is intended as a constitutional check rather than a destructive last resort, designs might also support federation, interoperability, negotiation, and reunification. This is a plausible experimental direction, not an established best practice.

Convert crisis improvisation into reviewable precedent

Informal adaptations may be unavoidable. A system can nevertheless require subsequent disclosure, review, expiration, and a decision on whether to codify the precedent. This would preserve flexibility while reducing the risk that emergency improvisation silently becomes permanent authority.

Measure mechanisms by exercise, not drafting quality

The highest-value next step is not another inventory of carefully written provisions. It is a comparative record of:

• Invocation frequency;
• Who invokes and who is excluded;
• Time and cost to remedy;
• Reversal and compliance rates;
• Effects on participation and legitimacy;
• Distribution of gains and losses;
• Whether practical control changes after formal reform;
• Whether temporary authority actually reverts.

Without those observations, constitutional craftsmanship can be mistaken for constitutional effectiveness.

---

10. Important uncertainties and limits

The dominant limitation is the gap between framework evidence and effectiveness evidence. The evidence base contains detailed primary texts for India, Israel, Cyprus, Ethiopia, ICANN, the IETF, MakerDAO, Debian, Python, Kubernetes, Meta, and others. For many of their most interesting mechanisms, no exercise record was retrieved. A provision should not be called effective merely because a serious institution drafted it carefully.

No comparative longitudinal study in the evidence links amendment rules, appeal structures, sunset provisions, veto arrangements, or revocable delegation to common measures of legitimacy, autonomy, accountability, stability, or coherent action. The strongest identified causal studies instead concern operational outcomes: village public goods, municipal administrative costs, and citizens’ political efficacy. Those findings should not be silently converted into claims about constitutional revision.

Other major limitations are:

• Construct validity: a sovereign legislature’s amending power, a court’s constitutional jurisdiction, a software project’s repository authority, and a token contract’s shutdown function are not identical constructs. They can be compared by function, standing, threshold, clock, default, and remedy, but not assumed equivalent because they share a label.
• Purposive sampling: the cases establish possibilities and recurring patterns, not their frequency.
• Fork survivorship: successful forks are visible; failed forks and acquiescent communities are missing.
• Cooperative evidence: this is the weakest domain, with no primary bylaws or revision histories.
• Platform outcomes: appeal volumes and institutional activity do not measure legitimacy or distributional effects.
• Emergency heterogeneity: national-security, pandemic, migration, and economic emergencies may operate differently.
• Informal adaptation: the evidence documents both necessary flexibility and constitutional drift but does not identify a validated threshold separating them.
• Excessive mutability: evidence of rigidity, capture, and emergency permanence is much richer than evidence of systems destabilized by too much amendability.

---

Conclusion

Constitutional adaptation is not primarily a choice between centralization and decentralization, hierarchy and democracy, or formal rules and informal practice. It is the continuing problem of aligning recognized authority with the resources that make authority effective, while preserving a credible way to contest and revise both.

Higher-order authority, bounded amendment, nested review, reversible delegation, vetoes, appeals, sunsets, and exit rights can all contribute to that task. None is universally necessary, none is sufficient, and each can be converted into its opposite: entrenchment into rigidity, veto into paralysis, delegation into fragmentation, participation into capture, sunset into ritual renewal, and exit into dissolution.

The most promising recurring architecture is not a particular institutional form but a forcing sequence: someone identifiable can act; a clock runs independently; inaction ends the extraordinary arrangement; and continuation requires approval beyond the power-holder. Even that conclusion remains provisional because three of its four clearest implementations lack retrieved exercise records.

The more durable practical lesson is methodological and institutional at once: evaluate the constitution that is written, the constitution that is practiced, and the interface between them. Adaptation is legitimate and coherent only when actors can see who changed authority, contest the change, obtain an effective remedy, and keep the system revisable without making every commitment arbitrarily mutable.

---

Footnotes

1. Mike Belshe et al., “Segwit2x Final Steps,” Bitcoin-segwit2x mailing list, November 8, 2017; Bitcoin Improvement Proposal 148, “Mandatory Activation of SegWit Deployment,” canonical bitcoin/bips repository; Wikimedia contributors, “Superprotect”; Linux Foundation, “Node.js Foundation Advances Community Collaboration, Announces New Members and Ratified Technical Governance”; OpenAI employees, “Letter to the Board of Directors at OpenAI,” November 20, 2023. ↩

2. Constitution of the Republic of Cyprus, 1960, arts. 78 and 182, text from the Comparative Constitutions Project repository; Attila Horváth, “Government by Decree: A Never-Ending Story? Analysis of Hungary’s Emergency Legislation (2020–2024),” Review of Central and East European Law (2026), DOI 10.1163/15730352-bja10125; Office of the Law Revision Counsel, U.S. House of Representatives, 50 U.S.C. §1622; Brennan Center for Justice, “Declared National Emergencies Under the National Emergencies Act”. ↩

3. Tom Ginsburg and James Melton, “Does the Constitutional Amendment Rule Matter at All? Amendment Cultures and the Challenges of Measuring Amendment Difficulty,” International Journal of Constitutional Law 13, no. 3 (2015): 686–713, https://academic.oup.com/icon/article/13/3/686/2450807. ↩

4. Stephen M. Griffin, “Understanding Informal Constitutional Change,” 2015, https://www.estudosinstitucionais.com/REI/article/download/24/16/78. ↩

5. BIP 148, “Mandatory Activation of SegWit Deployment”; Belshe et al., “Segwit2x Final Steps”; Adem Efe Gencer et al., “Decentralization in Bitcoin and Ethereum Networks,” 2018, https://arxiv.org/abs/1801.03998. ↩

6. Wikimedia contributors, “Superprotect”. The retained account is a community-maintained retrospective rather than an independent outcome evaluation. ↩

7. OpenAI, “OpenAI Announces Leadership Transition,” November 17, 2023; OpenAI employees, “Letter to the Board of Directors at OpenAI”; OpenAI, “Review Completed & Altman, Brockman to Continue to Lead OpenAI,” March 8, 2024. The last source is OpenAI’s summary of its commissioned review; the underlying WilmerHale report was not published. ↩

8. Constitution of the Republic of Cyprus, 1960, arts. 78 and 182, Comparative Constitutions Project repository. The constitutional text is primary; attribution of the 1963 breakdown to the rules rather than ethnic conflict and intervention remains contested. ↩

9. Horváth, “Government by Decree,” DOI 10.1163/15730352-bja10125. ↩

10. Office of the Law Revision Counsel, 50 U.S.C. §1622; Brennan Center, “Declared National Emergencies”. ↩

11. Jens Blom-Hansen, Kurt Houlberg, and Søren Serritzlew, “Size, Democracy, and the Economic Costs of Running the Political System,” American Journal of Political Science 58, no. 4 (2014): 790–803, DOI 10.1111/ajps.12096; David Dreyer Lassen and Søren Serritzlew, “Jurisdiction Size and Local Democracy,” American Political Science Review 105, no. 2 (2011): 238–258, DOI 10.1017/S000305541100013X; Jens Blom-Hansen, “Municipal Amalgamations and Common Pool Problems,” Scandinavian Political Studies 33 (2010), https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1467-9477.2009.00239.x. ↩

12. Mariyana Angelova, Hanna Bäck, Wolfgang C. Müller, and Daniel Strobl, “Veto Player Theory and Reform Making in Western Europe,” European Journal of Political Research (2017), DOI 10.1111/1475-6765.12226. ↩

13. Joel Selway and Kharis Templeman, “The Myth of Consociationalism? Conflict Reduction in Divided Societies,” Comparative Political Studies 45, no. 12 (2012): 1542–1571, DOI 10.1177/0010414011425341. ↩

14. Michael Cox, Gwen Arnold, and Sergio Villamayor-Tomás, “A Review of Design Principles for Community-Based Natural Resource Management,” Ecology and Society 15, no. 4 (2010), https://www.ecologyandsociety.org/vol15/iss4/art38/ES-2010-3704.pdf; Elinor Ostrom, “Beyond Markets and States: Polycentric Governance of Complex Economic Systems,” 2010, https://web.pdx.edu/~nwallace/EHP/OstromPolyGov.pdf. ↩

15. Scott Bradner, “The Internet Standards Process—Revision 3,” RFC 2026 / BCP 9, October 1996, especially §§6 and 9. ↩

16. Constitution (Forty-fourth Amendment) Act, 1978, Legislative Department, Government of India; Israel, Basic Law: The Government, §§38–39, Knesset text. ↩

17. Tanner Jones and Ryan Quandt, “An Iridescent Sunset: An Empirical Analysis of Sunset Legislation,” Journal of Regulatory Economics 68 (2025): 85–123, DOI 10.1007/s11149-025-09498-5. ↩

18. Python core development community, “PEP 13—Python Language Governance”; Ernest W. Durbin III, “Python Governance Vote: December 2018 Results”. ↩

19. Kesavananda Bharati v. State of Kerala (Supreme Court of India, 1973), full judgment; Constitution (Forty-fourth Amendment) Act, 1978. The widely reported 7–6 division is a secondary count rather than a fact stated in the judgment. ↩

20. European Commission, “Subsidiarity Control Mechanism”; European Parliament, “The Principle of Subsidiarity”. ↩

21. James M. Acheson, “The Development of the Maine Lobster Co-management Law,” c. 2001, https://dlc.dlib.indiana.edu/dlcrest/api/core/bitstreams/93b52c41-e0ae-4813-baf1-3ab4dfc64ff1/content; Maine Legislature, Title 12 §6447, “Lobster Management Policy Councils”. ↩

22. Kubernetes Steering Committee, “Steering Committee Charter”; Kubernetes community, “Community Membership” and “GitHub Organization Owners Guide”; ICANN, Bylaws for Internet Corporation for Assigned Names and Numbers, arts. 4, 6, 18, and 19 and annex D. ↩

23. Reference re Secession of Quebec, [1998] 2 SCR 217, Supreme Court of Canada, https://www.canlii.org/en/ca/scc/doc/1998/1998canlii793/1998canlii793.html. ↩

24. HashiCorp, “HashiCorp Updates Licensing FAQ Based on Community Questions,” August 21, 2023; OpenTofu, “Governance”; Dawn Foster and Matt Germonprez, “Freeriding and Rebellion: An Investigation of Open Source Vendor Relicensing and Member Hard Forking Events,” Information Systems Journal (2026), DOI 10.1111/isj.70060. ↩

25. Raghabendra Chattopadhyay and Esther Duflo, “Women as Policy Makers: Evidence from a Randomized Policy Experiment in India,” Econometrica 72, no. 5 (2004), https://economics.mit.edu/sites/default/files/2022-08/Women%20as%20Policy%20Makers%20Evidence%20from%20a%20Randomized.pdf. ↩

26. Bradner, RFC 2026 / BCP 9, §§6 and 9. ↩

27. Oversight Board, Oversight Board Charter; Oversight Board, 2023 Annual Report. The activity and implementation figures are institutional self-report. ↩

28. Tom Ginsburg and Mila Versteeg, “The Bound Executive: Emergency Powers During the Pandemic,” International Journal of Constitutional Law 19, no. 5 (2021), https://academic.oup.com/icon/article-abstract/19/5/1498/6308959. The evidence available here was the published abstract rather than the full article. ↩

29. Seye Abimbola, Leonard Baatiema, and Maryam Bigdeli, “The Impacts of Decentralization on Health System Equity, Efficiency and Resilience,” Health Policy and Planning 34, no. 8 (2019): 605–617, https://academic.oup.com/heapol/article/34/8/605/5543691. ↩