Summary
- A policy forum can produce an institutionally valid rule without answering the separate contractual question of how that rule became binding on a member who accepted an earlier agreement.
- The five registry systems do not use one common mechanism. APNIC relies heavily on annual renewal and current APNIC Documents; RIPE NCC expressly provides for General Meeting amendment of its service agreement and incorporates current policies and procedures; ARIN distinguishes later RSA versions from changes to defined Service Terms.
- LACNIC's LAC-2019-9 record shows why chronology matters: proposal, staff analysis, ratification and implementation are observable, but those stages do not by themselves prove individual notice or contractual effect for every pre-existing recipient.
- Dynamic incorporation is not inherently defective. Its legitimacy depends on a bounded document set, a valid change route, timely notice, foreseeable consequences, review and an exit that does not turn operational dependence into nominal consent.
The changed rule on the old signature
Consider an operator that signed a registry agreement in 2014. The agreement referred to applicable number-resource policies. Years later, a policy forum adopted a new rule governing transfers, documentation or circumstances in which records could be revoked. The registry then told the operator that the new rule formed part of the relationship it had already accepted.
The operator's first response might be: I never signed that rule. The registry's reply might be equally direct: you signed an agreement that incorporated policies as they change, or you renewed membership under the current documents, or your members approved the revised agreement through the institution's constitutional process. Neither answer settles the dispute by slogan. The result depends on the words of the agreement, the identity of the incorporated documents, the amendment route, notice, chronology and governing law.
That distinction is easy to lose in Internet governance because policy development and service contracting sit close together. A proposal may be discussed on a public list, refined at a meeting, assessed by staff and declared to have reached consensus. The resulting policy may be entirely valid as an internal institutional product. Yet its production does not answer every question about the duties of an existing contracting party. A policy process asks what rule the institution should adopt. A contract inquiry asks whether, how and from when that rule changes the legal relationship between specified parties.
The difference is not an argument for freezing registry rules. Internet operations change. Fraud patterns change. Security practices change. Transfer markets change. Contact methods and documentation standards change. A registry cannot plausibly obtain a handwritten signature from every member each time a technical form or evaluation procedure changes. Commercial contracts commonly incorporate standards, schedules and operating rules that evolve.
The point is narrower and more demanding. A changing rule needs a bridge into an existing relationship. Sometimes the bridge is annual renewal. Sometimes it is an express power granted to a membership meeting. Sometimes it is a defined service-terms mechanism coupled with notice and a termination option. Sometimes a general compliance clause does the work, though the breadth of that clause may itself become disputed. The bridge should be identified rather than replaced with the claim that the community decided.
Policy legitimacy and contractual effect answer different questions
A policy can have strong procedural credentials. The proposal may be public, objections may be recorded, staff may publish an impact assessment, and a chair or membership body may ratify the result. Those features matter. They reduce surprise, expose technical defects and make institutional decisions easier to audit. They can also support an argument that a member had a fair opportunity to participate.
But participation is not the same as assent by the legal person bearing the obligation. A network engineer who comments on a proposal may not have authority to vary the employer's contracts. A database contact may be authorised to maintain records but not to accept new financial or termination terms. A member that did not attend a meeting may still be governed by a properly adopted association decision, but the reason lies in the association's rules and the service agreement, not in a fiction that silence at the microphone was consent.
The APNIC Standard Membership Agreement illustrates one bridge. The published APNIC-079 form makes annual renewal central and connects the member to APNIC Documents as amended. Renewal payment can therefore carry more contractual significance than participation in a policy discussion. It supplies a recurring moment at which the relationship is continued under the agreement then in force.
The RIPE NCC Standard Service Agreement uses another route. RIPE-812 states that the agreement may be amended by General Meeting resolution and refers to RIPE policies and RIPE NCC procedures in their current versions. Here the corporate vote and the contractual text are linked expressly. The vote is not merely atmospheric evidence of community approval; it is part of a named variation mechanism.
The ARIN Registration Services Agreement, version 14.0 draws a line that is especially useful for analysis. It distinguishes changes to defined Service Terms from substitution of an entirely later RSA version. Its published terms say an existing holder need not adopt a later RSA merely because ARIN has published one, while providing a route for certain Service Terms to change and a termination option for a materially adverse change. That is a more precise answer than treating every current document as one undifferentiated policy stack.
These examples show why the word policy is too broad to determine contractual effect. A policy could be a technical evaluation rule, a charging document, a security standard, a transfer condition, a closure procedure, a constitutional rule or an explanatory page. Each may enter the relationship through a different clause. Before asking whether a manual can amend a contract, one must identify which manual, which provision, which party and which bridge.
The four documents that are often collapsed into one
An effective amendment audit begins by separating four records. The first is the signed or accepted base agreement. That document establishes the initial relationship, governing law, service commitments, incorporated materials, liability terms, termination conditions and the mechanism for later changes.
The second is the policy instrument said to have changed the obligation. Its title, version, adoption date and scope matter. A proposal discussed in 2019 is not necessarily the rule implemented in 2021. A staff impact assessment may explain expected operation without itself being the binding text. An explanatory FAQ may accurately describe a contract while disclaiming any power to alter it.
The third is the institutional adoption record. Depending on the registry, that may be a consensus declaration, board ratification, member resolution, executive approval or publication after a defined review period. The adoption record answers whether the institution followed its own route. It does not necessarily answer whether the resulting text was within the category of documents dynamically incorporated by every existing agreement.
The fourth is the member-facing notice and acceptance record. This could be a renewal invoice, a portal acceptance screen, a direct email, a meeting notice, publication on an official site, a signed addendum or continued use after a contractually specified notice period. It may also include an objection, termination request or request for review. Public policy archives frequently preserve the first three records better than the fourth.
That missing fourth record is consequential. A registry may be able to show that a policy was discussed and validly adopted but not how an affected holder was told that a severe consequence had changed. The holder may be able to show an old signed agreement but not the renewal interface or notices received later. Without those records, confident claims about assent become difficult.
The problem grows when incorporated materials are described through broad labels. If an agreement says the member must comply with all current policies, does that include a board procedure, a fee schedule, a sanctions protocol, an external standard or only documents adopted through the number-resource process? A bounded list gives the member a better chance to foresee change. An open-ended reference to whatever the institution later calls policy gives management greater flexibility but creates a harder notice and scope problem.
APNIC: renewal as the recurring point of assent
APNIC's model deserves careful treatment because annual renewal changes the temporal question. The published membership agreement states that the agreement operates on an annual basis and treats payment of renewal fees as acceptance of the agreement then in force. It also makes APNIC Documents, as amended, central to the relationship and supplies Queensland governing law and dispute provisions.
This is stronger evidence of recurring assent than a theory based only on an open policy meeting. The member does not merely remain in a relationship created many years ago. It renews annually under a published structure. A registry defending a later rule can point to the renewal cycle, the agreement's dynamic reference and the member's payment after the rule became operative.
The strength has limits. Renewal payment may be operationally necessary if the member depends on APNIC services and registry recognition. That dependence does not automatically invalidate assent; many important services are renewed under standard terms. It does affect the practical evaluation of choice. If termination threatens records, routing-adjacent services, reverse DNS, RPKI or transfer capability, the member's formal option not to renew may carry a cost far beyond switching an ordinary supplier.
The public agreement also does not show the interface each member saw. Was a revised document highlighted? Did the invoice identify material changes? Was a comparison supplied? In which languages was notice available? Did a portal require an affirmative acknowledgment, or was payment alone treated as acceptance? How long did the member have to entity? Those facts can matter where the disputed change is substantial.
Scope remains important. The phrase APNIC Documents needs a document map. A member should be able to identify which policy families are incorporated, where authoritative versions live, and which change route applies to each. A technical allocation criterion adopted through the Policy Development Process is not necessarily equivalent to a constitutional amendment or a pricing change. The annual mechanism is clearer when the incorporated set is versioned and classified.
APNIC therefore offers neither automatic invalidity nor automatic validity. It offers a recurring contractual route. In a specific dispute, the questions would be whether the relevant rule belonged to the incorporated set, was validly adopted, was in force at renewal, was adequately disclosed, and was applied within its stated scope. The policy forum may explain the rule's institutional history. The renewal record helps explain its contractual entry.
RIPE NCC: when the membership vote is written into the agreement
RIPE NCC's route is different because the Standard Service Agreement expressly recognises amendment by General Meeting resolution. The agreement also incorporates RIPE policies and RIPE NCC procedures in their current versions. That drafting connects corporate governance, the service contract and a changing document environment.
This matters because one-member-one-vote is not individual re-signature. It does not need to be if the existing agreement validly grants the General Meeting an amendment power. Association membership often entails collective decision-making. A dissenting or absent member may still be bound by a resolution within the body's competence and the terms already accepted. The legal explanation is collective contractual and association structure, not the claim that every affected operator personally agreed with the outcome.
The distinction also disciplines the institution. A General Meeting power must be used through the meeting requirements that make it legitimate: notice, agenda, quorum where applicable, voting rules, accurate text and a record of the resolution. If the agreement says the meeting may amend, staff publication alone should not be treated as an equivalent route for changing the agreement itself.
Current policies and procedures create a second path. The agreement need not be amended every time an incorporated operational document changes. Yet that path raises the same classification question seen at APNIC. Which documents are RIPE policies, which are RIPE NCC procedures, who may amend each, and what happens when a procedure carries a consequence more severe than its label suggests?
Severity matters because an amendment can affect more than meeting administration. If a later procedure changes evidence duties, suspension conditions, closure sequencing or record consequences, a member may argue that the institution used a dynamic operational reference to achieve what should have required a clearer agreement amendment. The registry may answer that the member accepted current procedures precisely so operations could evolve. The dispute then turns on drafting, purpose, foreseeability and governing Dutch law, not on whether policy change is generally desirable.
RIPE NCC's structure has a genuine accountability feature: members can act collectively through the General Meeting. But a vote should be described at its proper scale. It is an internal corporate mechanism for those entitled to vote. It does not prove that every network, end user or non-member affected by registry practice authorised the change. Nor does it show whether the amendment burden fell evenly across members.
The most useful public supplement would be a version ledger linking each material agreement or incorporated-document change to the resolution, notice date, effective date, affected clauses and member remedies. That would let an operator reconstruct the rule applicable on a historical date without treating the current website as timeless.
ARIN: protecting the signed version while changing defined terms
ARIN's current agreement provides the clearest warning against collapsing contract version and service rules. The version 14.0 RSA says, in section 14(l), that an existing holder does not have to adopt a later RSA simply because ARIN publishes one. Section 14(n) limits amendment of the agreement. At the same time, the RSA defines Service Terms and provides a mechanism for changes to them.
That architecture preserves something important: the historical signed version remains relevant. Two holders may have relationships governed by different RSA versions. An analyst cannot safely take the newest PDF and assume every holder is bound by every clause in it. Version control is part of the rights analysis.
The protection is not a freeze. Defined Service Terms can change. The agreement therefore preserves a stable base while allowing parts of the operating environment to evolve. The important question is whether the disputed rule is truly a Service Term within the contract's definition or an attempted amendment to the RSA itself. Labels chosen after a dispute should not decide that classification.
Section 13(c) adds a further control by providing a termination route for a materially adverse Service Terms change. ARIN's RSA version 14.0 FAQ explains both the version protection and the adverse-impact exit, while correctly stating that the FAQ does not alter the agreement. The contract, not the explanation, remains controlling.
The exit right is conceptually significant. It makes a changed term contestable in a practical way: the holder can say the change is materially adverse and seek to leave rather than accept it. It may also discourage ARIN from using the Service Terms mechanism for unexpectedly severe changes.
Yet the value of exit depends on consequences. What happens to included number resources when the agreement ends? Which registry services continue? Can the holder maintain accurate public records, reverse DNS or security assertions? Does termination produce reversion, revocation, deregistration or another status? Are legacy resources treated differently? A right to exit that carries the loss of the operational position under dispute may be formally real but economically difficult to use.
The available public documents do not provide a denominator. They do not show how many holders have invoked the materially adverse change route, how ARIN evaluated materiality, how many terminations followed, or whether disputes reached a court or another forum. The clause is a meaningful design feature. Its practical force cannot be measured from text alone.
ARIN's separation nonetheless establishes a useful benchmark. An institution can allow policy evolution without declaring every new standard agreement compulsory for old signatories. It can define the changing terms, protect signed versions, give notice and create an objection or exit route. That is more precise than saying current policy always overrides an old signature.
LACNIC: a policy chronology is necessary but not sufficient
The LAC-2019-9 policy record provides a concrete test because it can be dated. The proposal was submitted in 2019, ratified in May 2020 and implemented in November 2021. The record includes the proposal's movement through interpretation, institutional decision and implementation, allowing a reviewer to compare the policy environment before and after an identifiable change.
That chronology is valuable. It prevents the common error of treating the latest manual as if it had always applied. If an alleged breach occurred in early 2020, a rule implemented in November 2021 cannot simply be projected backward. If a renewal or continuing service agreement occurred after implementation, the later date may support a different argument.
LAC-2019-9 also shows the role of staff analysis. Operational staff can identify effects, ambiguities, implementation work and interaction with existing rules. Such analysis is evidence that the institution considered how the proposal would function. It may help an operator foresee what a ratified rule will mean.
But staff interpretation is not necessarily the contract. Consensus is not necessarily the contract. Ratification is not necessarily individual notice. Implementation is not necessarily assent. Each is a stage with distinct evidentiary value. To establish binding effect against a pre-existing recipient, the institution would still need to identify the contractual clause that incorporates the policy, the recipient's applicable agreement, the notice route and any renewal or continued-service mechanism.
The missing denominator is again substantial. How many affected holders received direct notice? How many objected? How many changed conduct? How many renewed after implementation? How many faced a revocation-relevant consequence under the revised language? How many sought review? A public policy page cannot answer those member-level questions.
This does not make the policy illegitimate. It means the policy page proves what it proves: institutional chronology and public text. It should not be made to prove private notice or contract formation facts it does not record.
LACNIC's example is therefore a model for any historical amendment analysis. Fix the dates. Preserve each text. Separate proposal from implementation. Then retrieve the agreement and notices that connect the implemented rule to the affected party. Without that sequence, a reviewer risks reasoning backward from current policy to past obligation.
AFRINIC: a broad compliance duty under Mauritius law
The AFRINIC Registration Service Agreement, dated 27 November 2017 in the available record, requires compliance with the agreement and applicable resource policies. It connects breach to service and revocation consequences and identifies Mauritius law as governing the relationship.
That drafting creates a dynamic compliance environment. A member cannot plausibly read the agreement as containing every operational rule in its own pages. Resource policies do substantive work. The agreement is the bridge that gives applicable policy a contractual role.
The word applicable is doing important work. It implies limits. A policy must apply to the resource, party, conduct and date in question. A policy for new allocations may not govern an old transfer. A rule for a specified resource class may not govern every registration. A proposal not yet ratified or implemented may not be applicable. A staff opinion may not have the same status as an adopted policy.
Chronology is equally important. If the claimed right or duty arose under an earlier agreement or earlier policy, a later text cannot be assumed to alter it without identifying the dynamic clause and its lawful reach. The fact that a contract requires compliance with applicable policy does not answer every question about vested expectations, severe consequences, notice or mandatory rules under Mauritius law.
AFRINIC's enforcement consequences make precision especially important. The more a policy can affect service continuity or resource-record status, the stronger the case for a transparent version history, direct notice and independent review. A minor documentation update and a rule enabling revocation should not be treated as institutionally equivalent merely because both appear in a manual.
The agreement also cannot prove how courts would resolve a contested later change. The available evidence includes no reported judgment that establishes a universal rule for dynamic incorporation in an AFRINIC dispute. A responsible analysis should therefore stop short of declaring all later policies binding or invalid. The answer would depend on the agreement, the policy, chronology, notice, conduct, remedy and applicable law.
Dynamic incorporation is not a legal defect by itself
The strongest case for registries is operational. Number-resource administration cannot function with every member on an entirely private policy universe. Registry accuracy, transfer compatibility, fraud controls, security practices and technical evaluation require common rules. If every change demanded unanimous re-signature, a single holdout could preserve an obsolete process indefinitely.
Dynamic incorporation is a familiar answer to that problem. Contracts refer to technical standards, exchange rules, scheme manuals, tariffs and operating procedures that can be changed through a defined process. The parties agree at the outset not only to today's text but to a bounded method of future revision.
In the registry setting, annual renewal strengthens that answer. Member voting can strengthen it. Public proposals and archived objections strengthen it. Advance notice strengthens it. A right to challenge, cure, terminate or seek review strengthens it. None of these mechanisms has to replicate legislative procedure to be meaningful.
The critique should therefore avoid a false rule that no policy can ever change an existing obligation. That would confuse a concern about mandate with a prohibition on ordinary contracting. It would also make stable registry service harder, not safer.
The proper concern is scope. A dynamic clause should not become an unlimited blank cheque. It should identify the family of documents that may change, the actor authorised to change them, the process, the notice route and the consequences. A technical manual should not quietly become a vehicle for rewriting liability, ownership representations or termination rights if the base contract reserves those subjects to formal amendment.
Foreseeability is the practical measure. At the time of signing or renewal, could a reasonable member understand the kinds of changes that might later arrive through the incorporated process? Could it find the authoritative documents? Could it tell when they became effective? Could it entity or obtain review? Could it distinguish a policy duty from non-binding guidance? The clearer those answers, the stronger the institutional case.
Notice is part of the authorising chain
Notice is sometimes treated as administrative courtesy, as though publication somewhere on a registry website completes the matter. For a dynamic relationship, notice is more central. It tells the member that the agreed change machinery has produced a new result and when conduct must change.
The adequacy of notice can vary with severity. A small update to a form may reasonably be published through routine channels. A change that alters transfer eligibility, audit duties, suspension conditions or resource-record consequences calls for a clearer method. Direct notice, a change summary, the complete text, effective date and remedy information reduce uncertainty.
Notice also has an identity problem. Registries maintain technical contacts, administrative contacts, abuse contacts, billing contacts and corporate member representatives. Sending a message to one address does not prove that a person with contractual authority received it. Institutions need a defined notice address and a process for members to keep it current. Members, in turn, should not be able to defeat notice by neglecting the designated contact.
Language and access matter without converting every change into a global plebiscite. An institution operating across many legal and linguistic environments should identify the controlling text and provide reliable translations where promised. A member cannot assess a material change it cannot reasonably read. If one language controls, the status of translations should be explicit.
Evidence of delivery should be preserved. Publication logs, email records, portal acknowledgments and renewal screens can settle disputes years later. Without them, both parties rely on general practice. The registry says members were informed; the member says the change was buried. A versioned notice archive is cheaper than reconstructing institutional memory during litigation.
The exit test: formal termination or usable choice
An exit right is one of the strongest controls on a change mechanism. If a member can refuse a materially adverse revision and leave, the institution cannot simply impose every new burden while keeping the relationship captive. ARIN's adverse-impact route makes this control visible. APNIC's annual renewal makes non-renewal conceptually possible. Other agreements provide termination routes of varying scope.
But exit must be examined as an operational event. A member does not consume registry service like office stationery. Registry records, transfer recognition, reverse DNS, routing security services and counterparties may depend on the relationship. Leaving can affect customers and networks that never participated in the amendment debate.
The relevant question is not whether termination exists in a clause. It is what survives termination. Does the registry preserve an accurate historical record? Does it deregister the resource? Can the holder move registration service elsewhere? Do security entities remain valid during a dispute? Is there a cure or transition period? Are different resource classes treated differently? Can a member terminate the changed service without surrendering every other relationship?
If exit means losing the practical position whose terms are disputed, it is weaker evidence of voluntary acceptance. That does not necessarily make the change invalid, but it changes the institutional legitimacy analysis. A formal choice between accepting a new term and risking network continuity is not equivalent to choosing among substitutable vendors.
Portability would improve the amendment system. If registration service could move without renumbering and without destroying reliable records, an adverse change would face a market and governance constraint. Members could reject an institution's expanding rule while preserving uniqueness and continuity. Exit would become a real check rather than a threat to the operator itself.
Until such portability exists, registries should be cautious about using termination as proof of consent. The more severe the exit consequence, the greater the need for notice, narrow drafting, review and preservation of the last verified operational state while a dispute is resolved.
Enforcement must follow the same chronology as amendment
Even a validly incorporated policy does not prove a breach. Amendment authority and enforcement authority are separate. An institution may have validly adopted a new documentation rule, yet still fail to show that the member's conduct fell within it, that the rule was in force on the relevant date, that proper notice was sent or that the prescribed cure period expired.
An enforcement file should therefore identify six dates: the applicable agreement date, policy adoption date, notice date, effective date, alleged conduct date and decision date. If renewal is the asserted assent mechanism, the renewal date belongs in the sequence as well.
The file should also identify the consequence. A warning is not a suspension. Suspension is not contract termination. Termination is not necessarily membership loss. Membership loss is not automatically resource deregistration. Each step may have its own clause, actor and remedy.
This separation protects both sides. The institution can demonstrate a disciplined chain rather than relying on broad policy language. The member can challenge the exact break rather than denying every registry rule. A dispute may concern notice rather than validity, classification rather than institutional competence, or proportionality rather than the existence of the duty.
Independent review is especially valuable where the registry is both author of the policy and decision-maker in the member's case. A reviewer should be able to ask whether the institution used the correct version, interpreted the incorporated category reasonably, proved notice, applied the stated facts and selected an authorised consequence. Review need not redesign the policy to correct an application error.
Public aggregate reporting would reveal whether this machinery works. Registries could publish counts of material policy changes, direct notices, objections, renewals, adverse-impact claims, terminations, suspensions, reversals and decided appeals. No sensitive member file is needed to show whether amendment disputes are rare, concentrated or routinely resolved before enforcement.
A practical amendment test
A member, registry, court or arbitrator assessing a later policy can use a compact sequence.
First, identify the base contract and the version accepted by the affected party. Do not substitute the newest agreement merely because it is easiest to find.
Second, identify the exact later rule, its version and its effective date. Separate proposal text, staff analysis, ratified text and implementation guidance.
Third, locate the bridge. Is the rule incorporated as a current policy, accepted at annual renewal, approved under an express General Meeting power, changed as a defined Service Term or introduced through another agreed mechanism?
Fourth, test institutional competence. Did the authorised actor use the required process? A staff explanation cannot replace a member resolution where the agreement requires one. A policy consensus cannot replace a formal agreement amendment where the subject is reserved.
Fifth, test scope and foreseeability. Does the rule belong to the incorporated document family? Does it regulate a subject the dynamic clause could reasonably reach? Is it a technical implementation change or a rewrite of a protected base term?
Sixth, test notice and assent. What was sent, to whom, in what language, on what date, through which designated channel? Was renewal or continued use specified as acceptance? Was there a meaningful objection period?
Seventh, test consequence and remedy. What happens on non-compliance? Is there cure, internal review, arbitration, court access or an adverse-change exit? What happens to records and operational services while the dispute is unresolved?
Eighth, apply the governing law. APNIC's Queensland clause, RIPE NCC's Dutch setting, ARIN's governing provisions and AFRINIC's Mauritius clause are not interchangeable. A universal answer would ignore the legal systems the contracts themselves select.
This test does not predetermine the result. It makes the result explainable. A later policy may bind because the member accepted a clear dynamic mechanism and renewed with notice. It may fail against a particular holder because the wrong version was used, the rule fell outside the incorporated set or the required amendment route was not followed. It may be valid generally but inapplicable to conduct before its effective date.
What the public record still cannot tell us
The contracts and policy pages establish real institutional design choices. They do not show comparable practice across all five registries. The missing evidence prevents a confident ranking of which system gives members the most effective control.
There is no common table of major changes, affected-holder counts, direct notices, objections, renewals, exits, appeals and successful challenges. The documents do not show how often a member was told that a change was material, how often a registry accepted an objection, or whether any court held an incorporated policy invalid or inapplicable.
Renewal interfaces are also missing from the public comparative record. APNIC's agreement makes renewal legally significant, but the agreement is not a screen capture of each year's member experience. RIPE-812 identifies a General Meeting mechanism, but the agreement does not show attendance, proxy use or how non-attending members understood each change. ARIN's adverse-impact clause is visible, but its use denominator is not. LAC-2019-9 gives a policy chronology, but not individual notices. AFRINIC's RSA gives a compliance clause, but not a decided catalogue of later-policy disputes.
The operational effect of termination is the largest unresolved issue. A nominal exit is difficult to value without knowing what happens to included resources and critical services. Public documents should distinguish contract end, membership end, service end, record status and security-service transition. Without that map, claims of voluntary choice remain incomplete.
The absence of these records should not be converted into an allegation. It is an uncertainty. Good institutional analysis marks the boundary: the text shows a mechanism; the practice denominator is not available. That is more reliable than assuming either perfect consent or systematic coercion.
The policy manual can amend only through the bridge
So, can a policy manual amend a membership contract? Sometimes, but not by its own force and not merely because a participatory process produced it.
APNIC's annual structure can make renewal under current documents a recurring assent event. RIPE NCC's agreement can make a General Meeting resolution part of an express amendment route and can dynamically incorporate current policies and procedures. ARIN can preserve an old RSA version while allowing defined Service Terms to change, with a materially adverse change route that invites scrutiny of exit consequences. LACNIC's policy archive can prove a dated institutional lifecycle without proving every holder's notice.
AFRINIC's RSA can make applicable resource policy a continuing duty while leaving jurisdiction-specific questions about scope, chronology and severe consequences.
The common principle is not that policy is weak. It is that policy and contract have different jobs. Policy defines the institution's operational rule. Contract identifies why a specified party is bound. Corporate procedure identifies who may change the document. Notice tells the party that change occurred. Renewal or another agreed mechanism may evidence assent. Review and exit constrain abuse. Governing law decides disputes the documents cannot settle themselves.
Registries strengthen their legitimacy when they show that entire chain. They weaken it when they replace the chain with a single word such as consensus. Consensus can be excellent evidence of technical deliberation. It is not a universal solvent for contract formation.
The best future design is therefore neither frozen agreements nor infinitely elastic manuals. It is a versioned, bounded and reviewable system: stable base terms; clearly identified dynamic documents; authorised change routes; direct notice for material changes; date-specific archives; independent review; aggregate use reporting; and an exit that preserves network continuity.
A member should be able to answer a simple question without reconstructing decades of institutional custom: which text bound me on the date that mattered, and by what agreed route did it become binding? If the registry can answer that question with the contract, policy, notice, chronology and remedy in one chain, the later rule has a defensible foundation. If it can answer only that the community decided, the most important part of the amendment remains unproved.