Summary

  • A related-party transaction is not evidence of corruption. It is a transfer made within a relationship capable of affecting terms, selection or accountability. That relationship makes ordinary market signals weaker, so the burden shifts to disclosure of the parties, decision authority, amount, pricing method, balances, commitments and measurable benefit.
  • APNIC Foundation's Australian financial statements for 2024 disclose AUD 7.008 million of purchases from APNIC Pty Ltd, AUD 10,000 of sales to it and an AUD 1.536 million year-end payable. The same report says connected dealings were on arm's-length terms. APNIC Pty Ltd separately recorded AUD 3.969 million of Foundation receipts and AUD 3.949 million of Foundation-funded project expense. The labels are useful, but readers still need a cross-entity reconciliation of services, timing, currencies, overhead and unperformed obligations.
  • RIPE NCC's 2024 financial report provides a different disclosure model: it named connected suppliers, reported EUR 56,000 of outgoing transactions and EUR 211,000 of incoming fees and sponsorship. Its 2025 consolidated report added a controlled Dubai subsidiary, eliminated intercompany balances and reported EUR 64,000 of outgoing transactions with suppliers related to Board members. This shows that specificity is possible without alleging that the dealings were improper.
  • Public-benefit vocabulary cannot establish an arm's-length result. A project may advance inclusion, training, routing security or registry continuity while its price, overhead allocation, procurement route or governance remains weak. Conversely, a connected service can be cheaper and better than an external purchase. Purpose and transaction quality must therefore be tested separately.
  • A credible Number Resource Society would publish a group map, a rolling related-party register, conflicts and recusals, contract value, pricing benchmark, service specification, overhead allocation, outstanding balances, assurance status and a renewal or exit decision. Member approval would authorise the arrangement but would not substitute for those facts.

Public interest is a purpose, not a transaction test

The vocabulary of Internet institutions is unusually favourable to connected spending. Money supports “the community”, “Internet development”, “capacity building”, “coordination”, “resilience” or “the public interest”. Each phrase points to a possible benefit. None identifies the supplier, the purchaser, the price, the alternative, the decision-maker or the obligation left at year end.

That distinction matters because number registries do not operate in an ordinary consumer market. A network cannot routinely move to another recognised regional registry because it dislikes a grant programme, shared-service charge or subsidiary. Core fee income is collected from organisations that remain dependent on authoritative records and associated services. The institution may then transfer part of that income to a connected foundation, coordination body, controlled company or supplier associated with a governor. The transfer may be entirely sensible. The payer's constrained exit nevertheless raises the standard of explanation.

Public purpose is often used as if it answers two questions at once. The first is whether an activity is worth doing. The second is whether the chosen legal entity and financial terms are the best way to do it. A cybersecurity training programme may be valuable while the administrative charge is excessive. A foundation may secure external grants while a service agreement leaves the registry carrying staff and systems without full reimbursement. A regional office may improve collection and representation while its governance duplicates senior roles and weakens member sight of commitments.

The reverse is also possible. A related organisation may provide expertise at cost, pool donor funds, isolate risk and deliver a service no external vendor could match. An insistence on open competition for every internal transfer can itself waste money. The correct discipline is not hostility to relationships. It is an account that allows a member to distinguish mission value from relationship privilege.

The first rule is therefore austere: never infer transaction quality from the noun attached to the programme. “Development” describes an aspiration. “Public interest” describes a claimed beneficiary. The financial question begins after those words, with control, price, performance and remedy.

Related party describes influence, not misconduct

Accounting uses “related party” more broadly and more neutrally than political debate usually does. Control, joint control, significant influence, key management status and close relationships can bring a person or entity inside the definition. A transfer of resources, services or obligations can be a related-party transaction even when no price is charged. The point is to alert the reader that the parties may transact on terms that an independent buyer and seller would not choose.

This is different from a conflict of interest. A director can face a conflict when personal, professional or institutional loyalties may affect a decision. The conflict may be managed through disclosure, recusal and an independent vote. A controlled subsidiary is related to its parent even if no individual director has a personal conflict. A major supplier is not automatically a related party merely because the customer depends on it. The categories overlap, but they are not synonyms.

It is also different from abuse. A transaction can be connected, fully disclosed, fairly priced and beneficial. A bargain can even favour the registry or charity more than an arm's-length counterparty would. The presence of a relationship tells the reader where normal market inference is unreliable; it does not supply a verdict.

That neutrality is important for RIR governance. If every request for connected-party disclosure is treated as an accusation, boards become defensive and members hesitate to ask. If every public-purpose arrangement is presumed benign, the control disappears in the opposite direction. Mature institutions disclose the relationship precisely so that legitimacy does not depend on personal trust.

The relevant questions are factual. Who controls each legal entity? Which governors, executives or close organisational associates sit on both sides? What moved: cash, staff time, intellectual property, facilities, donor obligations, data access, insurance, guarantees or reputation? How was the amount set? Who approved it after recusals? What remained payable or receivable? What output was accepted? Could either party terminate, replace or reprice the service?

A clean audit opinion cannot answer all of these. It provides assurance on the financial statements under the applicable reporting basis. It does not certify that every connected programme was necessary, that an unobservable market price was optimal, or that members would have chosen the same institutional boundary. Accounting disclosure is the start of the governance examination, not its conclusion.

APNIC's institutional perimeter makes the issue visible

The APNIC family offers a useful exhibit because its public documents show several legal and operating surfaces. APNIC Pty Ltd is the regional registry company in Australia. Foundation activity has been reported through an Australian charitable entity and, historically, a Hong Kong company. The registry's people, systems and programmes can serve externally funded projects, while Foundation money can support activities delivered through the registry organisation.

This design has plausible advantages. A dedicated charity can seek grants that a membership-funded registry may not attract. Donors can restrict money to specified outcomes. Development work can be distinguished from core registration. Shared staff and systems can avoid creating an entirely separate administrative apparatus. Regional expertise can connect technical assistance to real operational needs.

The same design creates the need for a stronger perimeter account. A reader must know which entity employs the people, signs the donor agreement, owns the platform, accepts delivery, carries foreign-exchange exposure, bears an employment claim and owns an unfinished product. A programme can appear as revenue in one entity and expense in another. An amount payable at year end can represent a timing difference, an unreimbursed service or a disputed allocation. Consolidated rhetoric about one community does not eliminate separate fiduciary duties.

The 2024 reports also record institutional change. APNIC Pty Ltd, described as sole member of APNIC Foundation Limited in Hong Kong, resolved to deregister that company on 31 December 2024 after it had ceased business or held no investment for more than three months. Deregistration can be a routine simplification. It still raises the closure questions any controlled vehicle should answer: which contracts ended, which records moved, who assumed residual obligations, whether donor restrictions survived, and what governance became unnecessary.

None of this implies that the structure was created to evade oversight. The lesson is almost the opposite. Because the entities publish financial material, the public can see that “APNIC Foundation” is not a single self-explanatory budget label. Legal form affects reporting, control and risk. A public-interest institution should explain those seams rather than relying on a shared brand to make them disappear.

The 2024 numbers require a cross-entity bridge

The Australian Foundation statements in the 2024 annual report disclose AUD 7,007,549 of purchases from APNIC Pty Ltd, AUD 10,000 of sales to it and an AUD 1,535,983 payable to it at year end. Purchases from the Hong Kong Foundation entity were AUD 3,210, matched by an AUD 3,210 payable. Total related-party purchases were therefore AUD 7,010,759, while total connected payables were AUD 1,539,193.

APNIC Pty Ltd's own audited report presents another view of Foundation activity. It records AUD 3,969,214 as Foundation receipts and AUD 3,949,047 as Foundation-funded project expense in 2024. The difference is AUD 20,167. In 2023, the corresponding receipt and expense figures were both AUD 4,589,804. These lines do not, by themselves, reconcile to the Foundation's purchases from APNIC Pty Ltd.

There may be entirely ordinary reasons. The entities may recognise different classes of projects, use different service boundaries, carry prior-period balances, separate donor funds from internal purchases, or record costs on different timing bases. One table may include staff and services that another presents in a different line. The Foundation report may contain accounts for more than one jurisdiction. A payable can lag the expense that created it. The public documents should not be forced into a false equivalence.

That is precisely why a bridge is needed. It should begin with each Foundation programme's gross expenditure, subtract grants paid directly to outside recipients, identify services bought from APNIC Pty Ltd, separate payroll or contractors employed by the Foundation, show pass-through costs, identify overhead and reconcile opening and closing balances. It should state the currency and recognition basis for each column. The APNIC Pty line for Foundation-funded project expense should then be mapped to the counterparty's purchases or explained where scopes differ.

Without that bridge, the member sees two true accounts but cannot easily answer a practical question: how much registry capacity did Foundation activity consume, what reimbursement did APNIC receive, and what net benefit or burden remained with fee payers? The issue is not whether the figures match visually. It is whether the economic exchange can be followed from donor or Foundation to registry input and completed public result.

An arm's-length sentence needs an evidential spine

The Foundation report states that sales to and purchases from related parties were made on terms equivalent to those prevailing in arm's-length transactions. It also says outstanding balances were unsecured and interest-free, settled in cash, with no guarantees and no expected-credit-loss provision for amounts owed by related parties. These are meaningful disclosures. They identify credit terms and make a testable assertion about pricing.

The assertion should not be mistaken for the test itself. International accounting guidance is careful on this point: an entity should describe related dealings as equivalent to arm's-length terms only when the claim can be substantiated. For a standard service, evidence may be competing bids or market rates. For a specialised bundle of registry engineers, training platforms, legal support, travel administration and donor reporting, a direct comparable may not exist.

Substantiation can then use a cost-based method. The service schedule should identify the staff grades, productive hours, contractor invoices, cloud consumption, travel and direct programme purchases. Shared costs should be allocated by a stable driver: headcount for human-resources support, invoice volume for finance, measured system use for infrastructure, occupied area for premises, or recorded legal hours for counsel. A mark-up, if any, should have a stated purpose and benchmark. Costs that APNIC would incur anyway should not automatically be charged again merely because a donor-funded project uses the same brand.

The method also needs an output side. A price can equal cost and still buy little. The purchaser should accept defined deliverables, reject deficient work and retain the right to compare internal provision with alternatives. Donor restrictions should be connected to the acceptance evidence. If APNIC receives a service fee for administering Foundation projects, members should see whether the fee covers the marginal cost and a fair share of capacity risk.

Publishing every timesheet is unnecessary. A compact assurance schedule could state the pricing method, allocation drivers, tested sample, independent reviewer, range of any mark-up, exceptions and year-end true-up. That would turn “arm's length” from a phrase of institutional confidence into a reproducible financial proposition.

Transfers without cash can be the most consequential

Cash attracts attention because it produces a visible line. Connected institutions also exchange value without an invoice. A registry may lend staff, host a platform, provide office space, carry insurance, permit use of a brand, supply procurement and finance functions, second an executive or absorb incident-response risk. A foundation may fund software that later becomes a registry asset. A coordination body may depend on one RIR's secretariat for administration while all five contribute in different proportions.

These transfers matter for two reasons. First, an omitted in-kind contribution understates the true cost of the receiving programme. A grant can appear efficient because the membership organisation quietly provides security, management and systems. Second, an unpriced transfer can shift decision power. The party supplying staff or infrastructure may shape programme priorities even where the formal board is separate.

The registry should therefore maintain a service catalogue. Each shared input needs an owner, user, quantity, cost driver and funding source. Small incidental support can be subject to a de minimis threshold. Recurrent or material use should be covered by a written agreement. The agreement should state data access, intellectual-property ownership, employment supervision, security obligations, insurance, incident allocation and termination assistance.

Guarantees and contingent support deserve particular attention. A parent may promise to keep a subsidiary solvent. One RIR may agree to support another during operational failure. A brand owner may carry reputational cost when a connected grantee fails. These commitments may never require cash, which is their desired outcome, but they consume risk capacity. The approving body should know the maximum exposure, trigger, decision authority and recovery priority.

Free services are not free to the institution that provides them. Nor are they necessarily suspect. Their omission simply prevents members from seeing which programme the fee base truly supports. A public-interest account should make the subsidy explicit, then ask whether it was deliberate, bounded and successful.

Directors and managers sit inside the economic map

Connected institutions are governed through people, not only company charts. Directors may represent member organisations that buy registry services, sponsor events, supply software, receive grants or employ other governors. Executives can hold roles in coordination bodies. Senior staff may influence both the specification of a service and the assessment of its delivery. These relationships are common in a small technical field.

The answer is not to exclude every knowledgeable person. That would strip boards of the expertise needed to understand registry security, routing and institutional continuity. The answer is to distinguish knowledge from decision entitlement. A person can inform a discussion while being excluded from selecting, pricing, voting, signing or evaluating a connected transaction.

The records should show the sequence. An interests register identifies standing relationships. The agenda asks for transaction-specific declarations. Minutes record who left, which quorum remained and who obtained independent advice. The contract is signed by authorised people who are not acting on both sides. Performance is accepted by someone whose budget or career does not depend on calling the project successful. Renewal returns to an unconflicted body.

Key management compensation belongs in the same map because management services can move between entities. The 2024 Australian Foundation statements reported AUD 751,209 of total compensation to key management personnel, up from AUD 728,068 in 2023. LACNIC's 2024 financial statements, using its own scope, reported USD 1,179,509 of fixed compensation for a group comprising Board members, the Deputy or Chief Executive Officer and managers, while stating that directors were unpaid. The amounts are not comparable as executive-pay rankings. They demonstrate how definitions determine what the reader can see.

A useful related-party report would identify the key-management population and any services supplied across entities, without exposing private personal data unrelated to governance. The central fact is not an individual's salary. It is whether authority, cost and accountability travel together when one management team serves several institutional purposes.

Shared missions still need separate acceptance decisions. An RIR and its foundation may agree on broad objectives: secure routing, capable operators, better connectivity and sustainable Internet infrastructure. Agreement on mission can weaken ordinary purchaser discipline. Staff on both sides want the project to succeed. A delayed deliverable can be reframed as learning. A platform built for one grant can be absorbed into a wider programme. The absence of an adversarial customer is pleasant but financially dangerous.

Each entity still needs its own decision. The Foundation asks whether the purchase advances its charitable purpose, respects donor restrictions and represents value. The registry asks whether supplying the service covers cost, protects core operations and fits the member-authorised mandate. One board's enthusiasm cannot answer for the other. Where directors overlap, independent members should carry the decision.

Acceptance should be specific enough to survive goodwill. A training programme might record entities who completed assessed modules, operator changes later verified, cost per successful completion and language coverage. A software project might define security review, uptime, adoption, maintenance ownership and portability. A research grant might require replicable data, documented limitations and a publication right that does not depend on a favourable conclusion.

This is where public-interest vocabulary often becomes least useful. “Capacity built” can conceal an attendance count with no evidence of operational change. “Community supported” can mean a conference sponsorship. “Infrastructure improved” can refer to equipment delivered but not maintained. The description need not be cynical to be too broad for accountability.

Separate acceptance also protects programme staff. If the deliverable and authority are clear, a manager is not forced to negotiate success informally with colleagues in the connected entity. A failed experiment can be reported as a failed experiment rather than rescued through ambiguous reallocations. Public institutions should be able to learn, but learning needs a budget and an endpoint.

The related-party schedule should therefore attach the same performance status used for an external supplier: contracted, delivered, accepted, delayed, varied, terminated or recovered. Shared purpose should reduce coordination cost, not the quality of evidence.

The NRO illustrates coordination without a simple corporate perimeter

The Number Resource Organization was formed through a memorandum among the RIRs rather than as a conventional parent company owning five subsidiaries. Its Executive Council is composed through the RIR leadership, its chair and secretariat rotate, and authorised activity can be funded through contributions among the registries. The arrangement supports collective representation and coordination while preserving the legal identity of each RIR.

That form creates a different connected-transaction problem. An NRO expense may be paid directly by one registry, reimbursed by others, counted as an in-kind contribution or allocated according to an agreed formula. The registry holding the secretariat can supply staff and systems. Travel may serve both regional and collective purposes. A legal claim arising from authorised NRO activity can trigger contribution questions under the memorandum.

Accounting standards in each jurisdiction may not classify every RIR-to-NRO dealing in the same way as a parent-subsidiary transaction. The governance need remains. Members should see the annual NRO plan, gross commitment, allocation formula, cash paid, in-kind contribution, host-registry cost and material variance. The report should identify which RIR contracted with each supplier and who accepted the work.

The same discipline applies to the Joint RIR Stability Fund. The published description says commitments are voluntary, funds need not be transferred until required and in-kind assistance can count with all RIRs' agreement. This is not a pooled bank account in the ordinary sense. A ledger that merely records a pledged headline could obscure the liquidity, conditions and opportunity cost behind each registry's commitment.

Coordination can be a public good while its incidence remains uneven. A smaller RIR may contribute expertise that is not visible in a revenue-based formula. A larger RIR may carry more cash or secretariat cost. The answer is not to force equality. It is to show the basis of difference and revisit it.

The NRO example widens the central principle: legal related-party status is a minimum accounting question. Institutional connectedness can also require disclosure where common governors, shared obligations and non-market allocation weaken the member's ability to infer who paid for what.

Vendors and professional advisers require a wider lens

Not every important connection sits inside a formal group. RIR boards and management operate in a specialised market. The same technical organisations can supply DNS software, sponsor meetings, employ directors, receive sponsorship and participate in policy discussions. Law firms, auditors, recruitment advisers and consultants may accumulate knowledge that makes replacement expensive. Dependence can exist without meeting the accounting definition of a related party.

RIPE NCC's disclosures illustrate how this can be reported narrowly and concretely. Its 2024 financial report identified companies associated with Executive Board members. It recorded EUR 56,000 of outgoing supplier transactions: EUR 10,000 for CZ.NIC's KNOT DNS software support, EUR 1,000 for CSNOG sponsorship and EUR 45,000 for Internet Society organisational sponsorship. It also reported EUR 211,000 of incoming dealings, comprising LIR service fees and meeting sponsorship, including a large RIPE dinner sponsorship from CZ.NIC.

Naming the parties does not imply an unfavourable price. KNOT may be the appropriate technical choice; sponsorship can have a defensible audience and purpose. The value of the disclosure is that members can connect Board relationships, revenue and spending rather than seeing only broad expense lines.

Procurement rules then determine how much assurance is available. ARIN's published external-contracting process generally applies from USD 50,000 a year and seeks at least two competitive bids, but it excludes categories including meeting venues, operations and maintenance, insurance, benefits, travel, legal services and opportunities whose information should not be public. The exclusions may be practical. Collectively, they cover many recurring areas in which relationships and incumbent advantage can matter.

A threshold or exception should therefore change the evidence, not end it. Below-threshold connected awards can still record a market check. Sole-source work can state why competition was impracticable, how price was tested and when the market will be retested. Legal confidentiality can protect advice without hiding total spend, selection authority or relationship status. The connected-party question is whether influence affected the bargain; the answer cannot depend on whether procurement happened to fit a public tender category.

RIPE NCC shows how the perimeter changes over time

Related-party reporting must evolve with institutional structure. RIPE NCC's 2024 financial report described the association as having no parent or group company and no entity exercising control or joint control. It nevertheless recognised indirect relationships through Board members who were senior managers of LIR accounts or connected to vendors, and disclosed the transactions.

The 2025 consolidated report had a wider perimeter. It included RIPE NCC Middle East FZ-LLC as a subsidiary, eliminated intercompany balances and transactions, and treated Executive Board and Management Team members as key management personnel. Incoming transactions with related parties were EUR 160,000: EUR 94,000 of LIR service fees and EUR 66,000 of event sponsorship. Outgoing transactions with Board-related suppliers were EUR 64,000, including EUR 14,000 to CZ.NIC and EUR 50,000 to the Internet Society.

The change does not mean the association suddenly became more conflicted. It means a controlled legal entity now existed within the reporting group and the accounts adapted. Consolidation prevents internal transfers from inflating group revenue or expense. It can also make the subsidiary's individual economics less visible unless a segment or intercompany note supplies them.

Board minutes provide the missing formation context. RIPE NCC Middle East FZ-LLC was established in February 2024, had no bank account or operating transactions during that year, and relied on the Dutch association to advance establishment costs and commit financial support during the start-up phase. The RIPE NCC acted as sole member and shareholder to adopt the subsidiary's financial statements. These are ordinary facts of a new controlled entity, but they are also the terms members need to assess risk.

Future reporting should keep the bridge after consolidation: services supplied in each direction, pricing method, staff, outstanding commitments, local revenue, start-up subsidy, decision rights and closure conditions. A subsidiary can solve banking or regional-presence problems. Its success should be measured against those problems, not inferred from the mere existence of a consolidated clean audit.

Member approval cannot manufacture an independent bargain. Member voting is a powerful source of authority in an association. It can approve accounts, charging schemes, constitutional changes or specified transactions. It cannot turn incomplete information into an arm's-length test. A majority may authorise a connected arrangement without knowing the alternatives, allocation method or long-term exit cost.

This is not a reason to displace members with experts. It is a reason to separate authorisation from verification. Members decide the mandate and risk appetite. Unconflicted directors choose within that authority. Finance records the transfer. Programme owners accept delivery. Internal or external assurance tests the controls. The next member meeting receives enough information to challenge renewal.

The electorate also has limits. One organisation may hold several voting rights under the relevant rules; many downstream networks and Internet users do not vote at all. A programme can serve a wider regional public than the legal members funding it. Member approval therefore demonstrates associational consent, not universal public benefit.

Conversely, low turnout does not make every connected transaction invalid. Members delegate ordinary administration because they cannot vote on each invoice. The governance question is whether materiality, novelty, conflict or duration should bring a decision back to them. A multiyear commitment to a controlled entity deserves more visibility than routine monthly shared-service billing under an already approved agreement.

A sensible approval matrix uses both value and risk. It raises authority when a transaction involves a governor, creates a new entity, transfers data or intellectual property, guarantees another body's obligations, lasts beyond the approving Board's term, uses a non-competitive award, or is difficult to reverse. A small payment can be high-risk if it sets a precedent or grants control. A large pass-through grant can be lower-risk if the donor, recipient and deliverable are independently fixed.

The member resolution should state what it actually does: authorises a purpose, maximum exposure and duration. It should not certify price fairness unless members were given the method and evidence needed to make that finding.

Cross-RIR comparison needs a common disclosure grammar

RIR reports are shaped by different law, accounting standards and corporate forms. APNIC publishes a related-party table through Foundation accounts. RIPE NCC names Board-connected suppliers and now consolidates a subsidiary. LACNIC discloses a defined key-management group and a fixed compensation total. ARIN's United States Form 990 supplies individual compensation and governance questions, while its public contracting policy supplies a separate view of vendor selection. AFRINIC's available financial material has its own scope and interruptions.

These differences make a league table irresponsible. The absence of a line called “related parties” does not prove the absence of connected dealings. A larger disclosed amount may reflect a broader definition or a group that deliberately routes shared services through one entity. A smaller amount may reflect fewer connections, a higher materiality threshold, netting, consolidation or thin disclosure.

The right comparison is completeness. Does the institution identify control and significant influence? Does it include key management and Board-linked organisations? Are incoming and outgoing dealings both shown? Are in-kind support, guarantees and commitments included? Are balances and terms reported? Is arm's-length equivalence substantiated? Can the member reconcile group entities and understand which costs disappear on consolidation?

A common grammar would preserve jurisdictional accounts while adding a governance schedule. Categories should include controlled entities, foundations and trusts, joint arrangements, key management, Board-linked members, Board-linked suppliers, common coordination bodies and other relationships requiring explanation. For each, show gross inflow, gross outflow, balance, commitment, non-cash contribution and approval route.

Materiality should include qualitative risk. A transaction below the financial-statement threshold can still matter if it involves election services, independent review, legal representation, critical DNS or RPKI software, a Board member, or a transfer of member data. Aggregation should not combine unlike parties so completely that the relationship disappears.

Standardisation would not declare one RIR virtuous. It would let members ask the same questions despite different legal wrappers. That is the useful form of comparability: not a rank, but a shared ability to locate influence and follow value.

A rolling register should connect decision, money and delivery

Annual accounts arrive too late and at too high a level to carry the whole control. A Number Resource Society should maintain a rolling register for material related and connected dealings. The public version can protect personal and security-sensitive details while preserving the facts needed for accountability.

The register begins with identity: legal name, registration jurisdiction, relationship category, ultimate controller and relevant governor or manager. It then records the transaction: purpose, service or asset, contract date, term, maximum value, pricing method, allocation driver, procurement route and reason for any exception. It identifies the approving body, declarations, recusals, independent advice and vote.

Performance fields connect the bargain to reality. They include milestones, acceptance owner, amount invoiced, amount paid, outstanding balance, variation, unpriced contribution, incidents, remedies and status. At renewal, the record adds a benchmark, alternative suppliers or internal options, termination cost, data return, intellectual-property position and Board conclusion.

The register should not become a document dump. A member needs a concise row with links to the agreement summary, decision minute and assurance result. Commercially sensitive unit rates can be shown as a tested range. Personal conflicts can be described by relationship type without publishing home addresses or family details. Security architecture can remain protected while the existence and value of the contract are public.

Three reconciliations make the register trustworthy. Its annual totals must reconcile to the audited related-party note or explain reporting differences. Its outstanding balances must reconcile to receivables and payables. Its listed contracts must reconcile to procurement and Board records. Unrecorded in-kind support should be sampled through staff time, system access and shared facilities.

The result would answer the question that broad reports leave open: not merely how much was spent on a mission, but through whom, under whose authority, at what tested price, for what accepted result and with what remaining obligation.

Public-benefit claims need a beneficiary and incidence account

Even a well-priced transaction can distribute benefits unfairly. A programme funded by every member may chiefly serve a small group of operators, conference entities, vendors or economies. A foundation grant may attract external money but require registry matching funds. A training platform may be open to the region while its design and language make actual use narrow. “Public” is not an incidence analysis.

Each material connected programme should identify who pays, who receives the immediate service, who gains the durable asset and who carries failure risk. Fee-paying members, grant donors, staff, direct trainees, downstream networks and the wider routing system are different groups. Benefits may legitimately extend beyond members, but the transfer should be explicit rather than hidden behind one collective noun.

Measurement should follow the mechanism. For routing-security work, count operational adoption and validated routing change, not only attendees. For a registry platform, measure accurate records, secure authentication, availability, correction time and operator use. For coordination, identify the decision or common standard achieved and the duplicated cost avoided. For representation, record the concrete policy issue, intervention and result without pretending attribution is certain.

Uncertainty belongs in the account. Public-interest effects are often diffuse and delayed. A programme may reduce a low-probability risk that never materialises. The institution can state the theory, baseline, evidence and confidence rather than fabricate a precise monetary benefit. Independent evaluation should focus on whether the causal sequence is plausible and whether less connected alternatives were considered.

This protects worthwhile non-commercial activity from crude cost cutting. The answer to an uncertain benefit is not automatically to stop funding it. The answer is to cap experiments, define review dates and preserve the ability to redirect money when evidence remains weak. A permanent relationship should not rest forever on an untested first-year aspiration.

Public benefit becomes a defensible conclusion only after incidence, evidence and alternatives. Before that, it is a proposal.

The exit test reveals whether the relationship serves the institution

Related arrangements often look efficient because they remove interfaces. Shared people know one another. Systems are integrated. Contracts are light. Problems can be resolved by a call. The hidden price appears when the institution tries to change direction.

An exit test asks what happens if funding ends, a donor changes terms, a governor's affiliation creates a new conflict, service quality falls, a subsidiary loses banking access, one RIR enters institutional distress or members reject the programme. Can data be returned in usable form? Can staff be reassigned? Who owns code and curriculum? Which licences survive? Are balances settled? Does a guarantee activate? Can an external supplier operate the service without the related party's informal knowledge?

The test is especially important for critical registry functions. A connected entity should not hold the only credentials, recovery knowledge or contractual rights needed to maintain authoritative records, RPKI publication, reverse DNS or member authentication. Development programmes can tolerate a measured pause; core registry custody cannot.

Exit does not require annual retendering or artificial separation. It requires credible optionality. Contracts can have data standards, documentation duties, transition assistance, escrow for essential code, step-in rights and a cap on termination cost. Shared-service pricing can be reviewed against a periodic external benchmark. A subsidiary can publish the conditions under which its regional function will be absorbed, sold, transferred or closed.

The same test applies to the wider RIR system. Mutual assistance and shared institutions should strengthen continuity, not make every registry dependent on an undocumented web of favours. The cost and authority of emergency support should be agreed before a crisis.

If a relationship cannot be exited, its governance begins to resemble control regardless of legal label. Members should then see it as a long-term institutional commitment, not an ordinary annual programme expense.

The credible vocabulary is transaction-level evidence

The language of public interest need not be abandoned. Number registries do protect shared technical goods: unique registration, reliable public data, secure delegation and coordinated continuity. Foundations can extend skills and infrastructure beyond the organisations able to pay. Collective bodies can solve problems no region can solve alone.

Those claims become stronger, not weaker, when separated from the connected bargain. A defensible transaction can pass six tests. The relationship is identified. The decision is made by people with conflicts declared and managed. The price or allocation method is substantiated. The service and risk transfer are written down. Delivery is accepted against evidence. Renewal or exit remains possible on known terms.

APNIC's disclosures show material services moving between connected institutions and provide amounts, balances and an arm's-length statement. RIPE NCC's reports show that Board-linked suppliers can be named and that a new subsidiary can be brought into consolidation. ARIN's contracting policy shows how competition can be made the default while also revealing broad exception categories. LACNIC's key-management note shows why the population behind a compensation or influence figure must be defined.

No single exhibit proves a system-wide failure. Together they show that current disclosure is fragmented across accounts, minutes, corporate pages and programme reports. A diligent member can reconstruct parts of the picture. The institution should do that work once and publish the bridge.

The reform is modest in concept. Put the legal group map beside the money. Put the interests register beside the decision. Put the service agreement beside the output. Put the outstanding balance beside the exit obligation. Then use public-interest language for the conclusion supported by those records, not as a substitute for them.

Related-party transactions are normal. In a territorial coordination system funded by dependent users, unexplained related-party transactions should not be.

Sources