RIPE NCC and the economics of policy-proposal transaction costs

Summary

• The central question is not whether RIPE policy is formally open. It is. The economic question is whether a small LIR, access provider or time-poor network can convert a real operational burden into a viable policy proposal before larger repeat players shape the agenda.
• RIPE's documented policy process is open, transparent, consensus-based and built around working-group discussion. Its own documentation says anyone interested in the Internet may propose policy and take part in discussion, and that proposals move through creating, discussion, review and concluding phases.
• That formal openness still leaves a long fixed-cost chain: noticing the issue, proving it is policy rather than operations, assembling evidence, learning the dialect of RIPE documents, drafting precise text, answering mailing-list questions, joining meetings, surviving revision, responding to impact analysis, monitoring implementation and measuring downstream costs.
• These costs are not distributed evenly. Large operators, cloud platforms, telecom groups, brokers, vendors, consultants, lawyers and policy veterans can reuse knowledge across many debates. A small member may pay the full cost for one narrow issue while also running the network.
• The risk is not that bottom-up governance is false. The risk is that openness without cost equalisation can allow a formally open process to produce concentrated influence, especially where IPv4 scarcity, transfers, RPKI, registry records, abuse contacts, legal review and regional diversity make policy text economically valuable.
• Better design should preserve bottom-up control while lowering the fixed cost of serious contribution: plain-language problem statements, pre-proposal clinics, evidence templates, multilingual summaries, implementation-cost explainers, remote-session records, small-member impact notes, version histories, chair reason memos, post-implementation review windows and published cost-of-participation metrics.
• RIPE NCC's role is partly administrative and partly economic. It publishes proposals, tracks deadlines, supports drafting, provides facts and statistics and produces impact analysis. Those supports are valuable, but the remaining cost of influence should be treated as a governance variable, not as a private inconvenience.

The network that sees the burden first

A small LIR usually discovers policy friction at the worst possible time. A transfer is waiting on documentation from a seller whose corporate trail has become untidy. An RPKI change has to be coordinated with a customer migration. A database record needs correction before a cloud platform will accept an address range. An abuse-contact duty is consuming staff time without reducing the complaints that actually belong to other networks. A regional-use rule, a waiting-list rule or a transfer-lock provision is not an academic phrase; it is the difference between closing a customer order this month and explaining to management why an address plan has stalled.

The engineer who sees the problem is not necessarily the person with time to write policy. In a small operator, the same person may handle BGP sessions, customer tickets, reverse DNS, supplier escalations, peering requests, security notices, registry portal updates and emergency changes after a cable failure. A sentence in a RIPE document may be causing repeat pain, but the pain arrives as scattered episodes: a due-diligence question here, a ticket delay there, a legal memo after a transfer, a lost afternoon decoding whether an old mailing-list position still controls a new case. The problem is real. It is not yet a proposal.

To turn the burden into policy, the operator must do several jobs that are outside ordinary network operation. It must decide whether the problem belongs in RIPE policy, RIPE NCC service practice, contract terms, database guidance, RPKI tooling, transfer documentation or a working-group best-practice note. It must find the current policy text. It must read prior proposals. It must identify which working group is relevant. It must frame the issue in a way that does not sound like one company's grievance. It must anticipate objections from larger holders, brokers, cloud networks, security specialists and registry staff. It must draft alternative text rather than only complain.

That is the transaction-cost problem. Policy influence is not bought only with votes, money or titles. It is bought with attention, language, timing, evidence and stamina. Open access removes one barrier: nobody needs a formal licence to speak. It does not remove the fixed cost of speaking well enough, early enough and long enough to alter the rule.

This matters especially for RIPE NCC because its region is broad and its policy surface is dense. RIPE NCC's service-region page says it consists of more than 20,000 organisations acting as Local Internet Registries and that the service region covers more than 75 countries. Those organisations differ sharply in scale, language, legal environment, staffing, capital access and tolerance for procedural risk. A large European carrier can assign policy monitoring to specialists. A global cloud company can run the same analysis across multiple regions. A broker can reuse transfer knowledge on every deal. A small access provider in a frontier market may have one engineer who reads the list after midnight.

The economics of policy therefore begins before the first draft is published. It begins when operational pain is first noticed and somebody decides whether it is worth paying the cost to make that pain legible to the institution.

Open access is a starting condition, not equal influence

RIPE's policy system deserves credit for the principle it states. The published Policy Development Process in RIPE says the process is open to all, bottom-up, transparent, consensus-based and documented in RIPE Documents. It also says everyone interested in the well-being of the Internet may propose a policy and take part in the discussion that follows. On the RIPE policy-development page, prospective authors are told they do not have to be RIPE NCC members or regular meeting goers to propose policy.

Those commitments are important. A closed registry policy process would be worse. But economic equality is not produced by formal permission alone. A public library is open to all; it still helps to have time to read. A court filing system is open to parties; it still helps to have counsel. An open standards list is open to anyone; it still helps to know the archive, vocabulary and personalities. The same is true for registry policy. The right to speak is a starting condition. The ability to move text through the full policy life cycle is a scarce capability.

Institutional economics treats transaction costs as the costs of making a system work: search costs, bargaining costs, measurement costs, enforcement costs and monitoring costs. In RIPE policy, search cost means discovering the proposal, its history and its relevant working group. Bargaining cost means finding allies, narrowing language and responding to concerns without losing the core point. Measurement cost means proving that a rule creates real operational burden rather than mere annoyance. Enforcement cost means designing language that can be implemented by RIPE NCC without creating arbitrary discretion. Monitoring cost means watching what happens after adoption.

These costs do not disappear because the process is civil, open or well documented. In some cases documentation itself creates cost. A serious contributor must read policy text, proposal versions, mailing-list threads, meeting minutes, impact analysis and implementation notices. The more transparent the record, the more material there is to master. Transparency is essential; it can still reward people who have already paid the learning cost.

The difference between open access and equal influence is especially important in a consensus system. Consensus is not a simple count. It is a judgement about the quality, breadth and persistence of support and concern. That design avoids crude majoritarianism, but it also creates premium value for contributors who can formulate objections in the idiom chairs recognise as relevant. A small network may be harmed by a transfer delay or documentation rule, yet its first expression of the harm may be informal, late, emotional or tied to a specific case. A policy veteran can translate the same harm into a narrower claim: the proposed text lacks a cure period, creates reliance-risk for existing holders, shifts cost onto low-capacity LIRs and needs an implementation note.

What the RIPE policy life cycle asks people to buy

RIPE-781 makes the cost structure visible. It identifies four phases: creating a proposal, discussion, review and concluding. Before any text is drafted, the document encourages sharing the idea with as many community members as possible, presenting it to a relevant working group, researching whether similar ideas have been discussed, seeking advice from authors of successful proposals and making working-group chairs aware of the idea. Once the proposal exists, the process moves through public discussion, new versions where needed, a draft RIPE Document, RIPE NCC impact analysis, review, chair assessment of consensus, concluding-stage comment and possible implementation.

That sequence is procedurally sensible. It is also expensive. A contributor must pay the cost of pre-proposal consultation before receiving the status and visibility of a formal proposal. The contributor must locate old debates, not merely assert novelty. It must seek advice without surrendering the issue to better-networked actors. It must bring the chairs into view without asking them to write the case. It must be ready to answer whether a similar idea failed earlier, whether the right working group is involved and whether multiple working groups should be notified.

The next cost is drafting. The appendix to RIPE-781's policy proposal template asks for author details, a suggested working group, proposal type, policy term, summary, current policy text, new policy text and rationale including motivation plus arguments supporting and opposing the proposal. A document that asks for arguments against one's own proposal is a useful discipline. It is also a test of institutional literacy. A small operator that only knows its own pain must become fluent enough to describe the rule's opposing case before the process will treat the proposal as mature.

Then comes discussion. The RIPE NCC announces the proposal on the Policy Announce Mailing List, with discussion usually occurring on the relevant working-group mailing list. The discussion phase is at least four weeks. Significant comments can require a new version and a new discussion phase. Each proposal has a webpage with version history and status. Each version is publicly archived. A contributor must remain present as text moves, questions accumulate and opponents refine their case.

The review phase adds another cost. RIPE NCC impact analysis is published before the proposal can move to review. That analysis may cover the registry and addressing systems, operations, service capacity and legal impact. A contributor must read it, understand what the registry believes the text means, and decide whether to revise, defend or narrow the proposal. If the analysis says implementation is costly, legally uncertain or operationally complex, the author must answer in a way that preserves consensus without ignoring the point.

The concluding phase adds the cost of late vigilance. RIPE-781 says that if there is no feedback at that stage, the absence is likely to be regarded as consensus and the earlier rough-consensus call still holds. This is reasonable as a method for closure; systems cannot wait forever. Yet it also turns silence into an economically meaningful signal. A small operator that misses the final window because the engineer was dealing with an outage may be counted as part of the quiet background against which the rule advances.

Discovery is the first fixed cost

Policy-discovery cost is easy to dismiss because the information is public. RIPE has policy pages, current and archived proposals, mailing lists, meeting sessions and public documents. The problem is not secrecy. The problem is attention. A small operator must know that a specific item on a specific list at a specific time may affect transfer timing, abuse-contact treatment, RPKI eligibility, database maintenance, IPv4 waiting-list expectations, documentation duties or resource registration. Discovery requires a theory of relevance before the costs are visible.

Consider a recurring transfer burden. A member may experience the burden as a slow deal, an unexpected document request or a buyer's discount. It may not immediately see that the friction has a policy dimension. The issue might sit in procedure rather than policy. It might be caused by contract language, legal review, anti-fraud control, registry staffing, sanctions exposure or an old policy compromise. The first cost is classification: what kind of problem is this?

Classification is cheaper for repeat players. A transfer broker sees many cases and can identify the pattern. A law firm sees repeated legal review. A large operator knows whether the same friction appeared in other acquisitions. A cloud platform can compare onboarding across regions. A long-time mailing-list contributor remembers which policy sentence created the current practice. A small LIR may see only its own case and hesitate to generalise.

The cost is not only cognitive. It is organisational. A network engineer may need permission from an employer before speaking publicly. A small company may fear that criticising a registry practice will complicate future requests. A larger member may require legal review before commenting. A public-sector operator may need approval from an agency hierarchy. A company exposed to sanctions, litigation or customer sensitivity may be careful not to put details on a public list. The open process does not remove those employer and legal costs.

For RIPE NCC, the challenge is that policy signals can be scattered across several operational surfaces. The What We Do pages describe resource distribution, tools for managing allocations and assignments, the RIPE Database, LIR Portal, DNS and reverse DNS, resource transfers and mergers, legacy resources, RPKI, policy implementation, measurements, analysis, meetings and training. A rule touching one surface can affect another. A transfer rule can affect database state. A database rule can affect abuse contactability. An RPKI practice can affect routing-security claims. Discovery requires connecting those surfaces before a formal proposal makes the connection for everyone else.

This is why plain-language notice matters. The institution should not assume that publication equals discovery. For high-consequence proposals, notice should explain in ordinary operational terms who might be affected: transfer buyers and sellers, small LIRs, legacy holders, holders using RPKI, networks with outsourced abuse handling, entities in complex jurisdictions, cloud-onboarding users, or members with limited administrative staff. A policy that cannot name its likely cost bearers is asking the wrong people to self-identify under pressure.

Evidence turns operational pain into institutional speech

The second fixed cost is evidence. A network can know that a rule is painful and still fail to prove that the pain should change policy. Institutions cannot revise policy for every anecdote. They need patterns, examples, counterfactuals and manageable remedies. Evidence is the bridge from operational irritation to institutional speech.

For policy-proposal transaction costs, evidence has several layers. The first is incidence: how often does the burden occur, and to whom? A single transfer delay may show bad luck; repeated delays across small sellers, cross-border deals or legacy holdings may show a rule problem. The second is mechanism: what exactly creates the cost? Is it policy text, implementation guidance, legal interpretation, database tooling, staffing, fraud control or poor member preparation? The third is distribution: does the burden fall evenly, or does it concentrate on small LIRs, non-English speakers, frontier regions, legacy holders, merger targets, resource-constrained engineers or members exposed to legal risk? The fourth is remedy: would a policy change reduce the cost without increasing fraud, uncertainty or administrative load?

Each layer requires work. Some evidence is confidential. Transfer prices, failed deals, corporate documents, customer commitments, legal memos and account disputes are not easily posted to a public list. Some evidence is fragmented. One operator knows its own delay; another knows a similar delay but not the first one. A broker may see a pattern but have commercial reasons to control the evidence. RIPE NCC may hold aggregate facts but be unable to disclose case details. The result is an evidence trap: the people most affected may lack the scale to show the pattern, while the people with scale may have private incentives in how the pattern is described.

This is where official support matters. RIPE-781 says RIPE NCC provides administrative support, including relevant facts and statistics and impact analysis. That support can lower the evidence burden. If the registry can publish aggregate transfer timing, common documentation failure points, request volumes, legal bottlenecks, rejected-case categories or implementation-cost ranges, the public debate becomes less dependent on private anecdotes. But even aggregate evidence has design choices. Which categories are used? Are small members separated from large ones? Are delays measured from first submission or complete file? Are legal-review pauses distinguished from member inaction? Are non-English support issues visible? Do withdrawn requests disappear from the picture?

The fair answer is not to treat every story as decisive. It is to create evidence templates that make small stories comparable. A transfer-friction proposal could ask contributors to provide, where safe, the request type, jurisdictional category, member size, delay source, downstream cost type and proposed remedy. An abuse-contact proposal could ask for ticket volume, false attribution rate, staffing cost, escalation outcome and customer effect. An RPKI proposal could ask for certificate-maintenance cost, route-risk scenario and recovery path. Structured evidence lowers the cost of becoming useful.

The important principle is that evidence design is itself governance. If the process only values evidence that large actors can produce, it will call the result meritocratic while pricing out the smaller base.

Drafting is a production function, not a clerical task

Policy drafting looks like word processing from the outside. It is not. It is the production function that converts a preferred institutional future into language that can survive discussion, consensus assessment, legal review and implementation. The first draft shapes the problem, names the affected class, defines the remedy and sets the default position from which later compromise occurs.

RIPE's proposal template makes this explicit. It asks for current policy text, new policy text, summary and rationale, including arguments supporting and opposing the proposal. That means the author must be able to read existing policy as an operative instrument. A small network may know what it wants: fewer repeated document requests, a clearer cure period, a narrower transfer restriction, a better database-maintenance path, a less burdensome abuse-contact rule. But policy requires text that tells RIPE NCC and the community exactly what changes and how the change will be interpreted later.

Drafting cost is high because the words sit near valuable systems. A phrase such as "must demonstrate need" can alter allocation behaviour. A phrase such as "may be transferred" can alter liquidity. A phrase such as "resource holder" can affect who is recognised. A phrase such as "accurate abuse contact" can become a service condition if tied to compliance practice. A phrase such as "may be revoked" can affect risk pricing even if rarely used. A transition clause can decide whether existing holders receive time to adjust or are placed under immediate uncertainty.

Repeat players hold drafting capital. They know previous language. They know how chairs read objections. They know which compromises are plausible. They know when to split a proposal, when to define a term, when to use a time-limited rule and when to leave implementation detail to RIPE NCC. They can ask friendly veterans to review text before publication. They can cite prior success. They can present a narrow amendment while holding a broader strategy in reserve.

Pre-proposal clinics are a practical fix. They should not turn RIPE NCC staff or chairs into authors for one side of a policy fight. But a structured clinic can help a contributor test whether the problem belongs in policy, identify prior debates, locate affected sections, define the cost, separate public evidence from confidential detail and choose whether a working-group discussion should precede a formal submission. That lowers the fixed cost of competent entry without giving anyone a substantive guarantee.

The aim is not to make proposals easy in the sense of casual. Bad policy should remain hard to pass. The aim is to make serious problems from low-capacity members less likely to die at the drafting desk.

Mailing-list stamina is an economic asset

Mailing lists are often defended as the core of open internet governance. They preserve a public record, allow asynchronous participation and reduce the need for travel. They are better than private lobbying rooms. Yet mailing-list influence has its own cost structure. It rewards those who can read long threads, answer quickly, remain calm under repetition, quote prior messages, distinguish old objections from new ones and return after every revision.

A four-week discussion phase is not simply four weeks on a calendar. It is four weeks of monitoring. A contributor must see questions as they arrive. It must decide which need answers and which can be left alone. It must avoid appearing evasive without letting opponents drag the discussion into every historical grievance. It must incorporate valid criticism without losing the policy's purpose. It must know when a chair's summary has framed the state of the debate correctly and when to ask for correction.

This is stamina, not mere access. A policy veteran can sustain a thread because the vocabulary is familiar. A consultant can bill time. A broker has a commercial reason to watch transfer language. A large operator can divide labour among staff. A small LIR may have one person reading after customer work. If that person misses two weeks, the thread may have moved from original problem to revised wording, impact concerns and emerging consensus.

The problem is not solved by asking everyone to behave politely. Civility is necessary, but transaction costs remain under civil conditions. The cost is in knowing when to speak, how much detail to provide, how to prove that a concern is general, how to handle a reply that reframes the issue, and when silence will be read as acceptance. A first-time contributor must learn all of that while the proposal clock runs.

A better design would preserve list-based deliberation while making the cost of following it lower. Each proposal page should show a living issue register: concern raised, current answer, text affected, status, and whether the concern remains unresolved. Chair summaries should separate strong objections, answered concerns, implementation questions and out-of-scope points. Version changes should include plain-language deltas, not only text differences. Contributors who join late should be able to see the live controversy without reading every message from the beginning.

Mailing lists are public infrastructure. Public infrastructure needs usability, not only access.

Meetings still price voice

Meetings add another price layer. RIPE community work uses mailing lists, but RIPE meetings and working-group sessions remain important because real-time exchange clarifies misunderstandings, reveals the intensity of concern and allows chairs to read the room. A meeting can rescue a proposal from slow textual confusion. It can also convert travel budgets, time zones, confidence, spoken English, employer permission and social familiarity into influence.

Remote access reduces some costs. It does not remove them. A small network must still know the session matters, clear time during operational hours, follow specialised discussion live, decide whether to speak publicly and frame the comment in a form that affects the record. In-person attendance adds flights, visas, hotels, lost engineering time and the informal value of hallway discussion. People who attend regularly know each other. They share shorthand. They know which objections are likely to land. A small operator arriving once to warn of a documentation burden enters a social market that already has prices.

The region makes this more consequential. RIPE NCC serves more than 75 countries, with large differences in language, income, travel access, legal systems and operational maturity. A meeting in Europe is not equally close to every member. A session scheduled conveniently for some time zones is inconvenient for others. A contributor whose employer sees policy as part of strategic affairs has a different opportunity cost from a contributor whose employer sees it as non-billable overhead.

None of this means meetings are bad. The alternative would be a thinner public record and slower correction of misunderstanding. The point is narrower: attendance should not be mistaken for representativeness. A room can be full of capable, public-spirited people and still underrepresent those who cannot afford to be present. Consensus judgement should ask not only who spoke, but who likely bears cost and whether those cost bearers had a realistic chance to understand and comment.

Meetings are markets in attention. The governance task is to prevent the price of attention from deciding the distribution of policy benefits.

Language and legal variety raise the price of seriousness

English is the working language of much RIPE policy discussion, but the RIPE NCC region is multilingual and legally diverse. The same policy sentence may be read by a Dutch lawyer, a Turkish network engineer, a Gulf telecom manager, a Central Asian ISP founder, a Ukrainian operator, a Spanish hosting company, a German compliance team and a British consultant. They do not arrive with the same legal vocabulary, regulatory assumptions or translation cost.

Language cost is not only translation in the literal sense. It is the cost of moving from operational experience into policy English. "We cannot get the documents quickly" becomes a claim about evidentiary burden and proportional verification. "This transfer rule makes deals fail" becomes a claim about liquidity, reliance and closing risk. "The abuse mailbox is overloaded" becomes a claim about false attribution, response incentives and the difference between contactability and liability. "RPKI changes need more care" becomes a claim about certificate-state continuity and operational recovery.

Some contributors can make that move because they have staff, counsel or policy experience. Others cannot. The result is not only quieter non-English voices. It is a narrowing of what counts as serious. A contribution may be discounted because it lacks terms that the policy culture recognises, even when the underlying cost is real.

Legal variety adds another layer. A documentation requirement that is simple in one jurisdiction may be slow or expensive in another. A corporate extract, notarised document, merger record, bankruptcy certificate, beneficial-ownership proof or sanctioned-party review may carry different cost depending on local law, bureaucracy, banking system and conflict exposure. A rule designed around the average member can therefore become regressive: it charges more, in time and risk, to members that already have less administrative capacity.

This is particularly relevant for IPv4 transfers, legacy resource recognition, merger handling, abuse-contact obligations and database maintenance. These are not only technical matters. They sit near corporate identity, legal authority, service obligations and commercial transactions. If a proposal changes the evidentiary burden or timing of those processes, its effect cannot be measured only by asking whether the text is technically clear.

Legal-impact explainers should also be more accessible. RIPE NCC impact analysis can identify legal impact, but members need to know what that means operationally. Does the concern relate to Dutch law, sanctions, data protection, contract enforceability, due diligence, fraud risk, or the limits of registry authority? Does it increase staff workload, member burden or external legal uncertainty? A one-line legal caveat can shift a debate without helping small members understand how the cost will reach them.

Policy seriousness should not require fluency in every discipline at once. The more a process demands technical, legal, economic and English-language fluency from the same person, the more influence moves toward the few who can buy that bundle.

Impact analysis is a help, not a substitute for cost equality

RIPE NCC's impact-analysis role is one of the most important equalising tools in the policy system. RIPE-781 says RIPE NCC publishes an impact analysis before a proposal moves to review, with the goal of providing supporting information and projections about possible impact. The listed areas include RIPE NCC's understanding of the proposed policy, impact on the registry and addressing systems, impact on operations, services and capacity, and legal impact.

That is valuable because it prevents a proposal from being judged only on the author's framing. It lets the registry say how the text might work in practice. It can reveal implementation costs that supporters missed. It can identify legal uncertainty. It can show effects on resource consumption, aggregation or fragmentation. It can help contributors focus on actual trade-offs rather than imagined ones.

But impact analysis does not by itself eliminate transaction-cost inequality. First, the analysis arrives after earlier costs have already been paid. The problem had to be noticed, framed, drafted and discussed before the analysis became available. A small operator that could not pay those earlier costs never reaches the stage where the registry's analysis helps.

Second, impact analysis can create a new interpretation cost. A technically precise assessment may still be difficult for small members to map onto their own risk. If RIPE NCC says implementation would require additional checks, changed tooling, legal review or support capacity, who pays? Staff through budget? All members through fees? Transfer users through delay? Small LIRs through repeated documentation? Buyers through closing risk? The institutional impact and the member-side impact are related but not identical.

Third, impact analysis is naturally centred on what RIPE NCC can observe. The registry can describe its operations, systems and legal concerns better than it can describe every downstream contract, financing condition, customer SLA, address-market discount or internal staffing cost borne by members. Those downstream costs are often where policy matters most. A proposal that looks cheap to implement can be expensive for small members; a proposal that is costly for RIPE NCC may reduce larger market uncertainty. Without a member-side cost note, the distributional picture remains incomplete.

A small-member impact note would help. It could ask a fixed set of questions: Does the proposal create new documentation duties? Does it change the timing of transfers, certifications, database updates or eligibility checks? Does it require legal review by members? Does it affect members differently by size, jurisdiction, language or business model? Does it create ongoing monitoring obligations? Does it include transition time, cure paths and post-implementation review? The note need not decide the policy. It would force distributional cost into the record before review.

Impact analysis is therefore part of the answer, not the whole answer. It lowers uncertainty about implementation. It should be complemented by tools that lower the cost of understanding who bears the policy's fixed and variable costs outside RIPE NCC.

Revisions are where value moves quietly

Policy value often moves between versions. A proposal may begin as a narrow fix and become a broader rule. It may begin as a strict obligation and become a softer requirement. It may gain a transition period, lose a grandfathering clause, add a definition, move a phrase from rationale into operative text or respond to impact analysis with language that changes implementation discretion. Each change may be reasonable. Together they create revision-management cost.

Version history is public in the RIPE process. Each proposal receives a webpage, status and archived versions. That is necessary transparency. Yet public version history is not the same as understood change. A small operator must compare text, read the rationale for change, track which concerns were answered, decide whether its own concern survived, and comment again if needed. This is legal and policy work disguised as reading.

Large repeat players can manage revisions strategically. They know when to accept a compromise that preserves the core economic effect. They know when a word in a transition clause is worth more than a paragraph in the rationale. They know how to wait until review to narrow an objection or press for an implementation note. They know which changes are cosmetic and which move value.

Small members often respond to the visible headline rather than the operative change. They may support a proposal called "clarification" without realising that the clarification alters burden of proof. They may ignore a proposal called "database accuracy" until a revision makes it harder to maintain records through outsourced operations. They may read a summary and miss a change to enforcement timing.

The fix is not to freeze proposals. Revision is how policy learns. The fix is to make change cheaper to understand. Every new version should include a plain-language change log that answers: what changed, why it changed, which concerns it answers, which concerns remain open, and which groups may face a different cost because of the change. Redline comparisons help experts; narrative deltas help everyone else.

Chair summaries can also reduce revision cost by keeping an issue register. If a concern about small-LIR documentation burden was raised in discussion, the next version should say whether it was addressed by text, rejected as out of scope, deferred to implementation or still unresolved. Without that trail, contributors must police their own concerns across drafts, which again favours those with time.

Policy legitimacy depends not only on whether the final text was public, but whether meaningful changes were visible to those who bear the cost.

The repeat-player advantage is structural

The repeat-player advantage is not mainly a story of bad motives. It is a story of lower marginal cost. Once an actor has learned the RIPE policy archive, working-group culture, proposal template, impact-analysis style, chair expectations and implementation pathways, that knowledge can be reused. Each new proposal becomes cheaper to track. Each new list thread becomes easier to interpret. Each new revision is easier to price.

Large operators gain from internal scale. They can spread policy knowledge across transfer needs, RPKI operations, database maintenance, abuse desks, legal review, public affairs and regulatory strategy. A cloud platform can treat policy monitoring as capacity planning. A telecommunications group can connect registry policy to spectrum, customer growth, merger activity and security obligations. A large hosting company can compare effects across many customers.

The solution is not to weaken experts. A policy system without experts would be worse. The solution is to treat expert advantage as a design fact. If repeat players have lower marginal cost, the process should lower fixed entry costs for occasional contributors. That means better summaries, clinics, evidence templates, issue registers, small-member notes and review windows. It also means chairs should be alert to the difference between repeated presence and broad cost-bearing.

In institutional terms, RIPE policy should make expertise productive without allowing procedural fluency to become a quiet property right.

Small LIRs pay the highest marginal price

The highest marginal price is often paid by the member least able to afford it. A small LIR may have enough address dependence to be affected by policy, but not enough staff to monitor policy continuously. It may need transfers, database accuracy, RPKI reliability and abuse-contact clarity just as much as a large network, but it lacks scale. It pays the full learning cost for each issue and cannot amortise that cost across many deals, departments or client files.

This creates a paradox. Small members are often the actors that bottom-up governance most needs to hear from because they reveal how rules operate under capacity constraints. Yet they are least likely to convert that knowledge into policy influence. A large operator can say a documentation rule is manageable because it has staff. A small operator may know the same rule will consume scarce engineering time, delay customers and increase dependence on consultants. If the small operator stays silent, the rule looks cheaper than it is.

The marginal cost appears in several forms. There is time cost: hours spent reading archives, drafting comments and joining calls. There is managerial cost: persuading an employer that public policy contribution is work, not hobby. There is legal cost: checking whether public comments disclose customer or transaction information. There is translation cost: writing in policy English. There is reputational cost: worrying that visible criticism will be remembered during later registry interactions. There is opportunity cost: every policy hour displaces customer work, security work or sales support.

Distributional effects are not limited to member size. Non-English speakers, newer LIRs, operators in jurisdictions with complex documents, networks close to sanctions or conflict risk, holders with legacy records, small hosting firms, regional access providers and time-poor engineers all face higher fixed costs. Some large actors face high costs too, especially when legal review is heavy, but they are more likely to have systems for absorbing those costs.

If policy ignores this distribution, it may produce rules that look efficient because the inefficiently burdened did not appear. That is not consent. It is selection. A small operator's absence may mean the proposal is harmless, but it may also mean the cost of objection exceeded the expected benefit of speaking.

Small-member impact notes are therefore not symbolic charity. They are a method for correcting measurement error. They ask whether the observed debate undercounts the cost borne by members with low policy capacity. A mature registry should want that correction because unmeasured cost becomes resentment, workaround, non-compliance or exit into intermediated services.

The specific surfaces where cost turns into power

Policy-proposal transaction costs matter most where policy text sits close to market value or operational continuity. In the RIPE NCC setting, several surfaces deserve particular attention.

Transfers are the obvious one. IPv4 scarcity gives transfer rules economic weight. A change in eligibility, documentation, holding period, publication, recognised holder status or inter-RIR compatibility can affect deal timing and price. Brokers and large buyers may understand these effects quickly. Small sellers, merger targets and occasional buyers may discover them only during a transaction.

RPKI is another. Route-origin authorisation is a security improvement, but policy or implementation changes around certification can create recovery burdens, signer-authority questions and customer-risk scenarios. Large operators can build internal RPKI teams. Small networks may rely on a few people or outsourced help. A proposal framed as security can still shift operational cost unevenly.

Database accuracy is a third. Accurate registry data is a public good. But accuracy duties can become costly when companies restructure, contacts change, abuse handling is outsourced, or legacy records do not fit modern documentation expectations. A proposal that improves data quality may be worthwhile; the question is whether it includes realistic cure paths and support for low-capacity members.

Abuse-contact policy is a fourth. Everyone wants abuse desks to be reachable. Yet the cost of handling misdirected, automated or low-quality complaints falls unevenly. A small provider may receive tickets that do not map neatly to its actual control surface. A large provider may automate triage. Policy that treats contactability as simple may miss the operational cost of being contactable without being responsible for the reported harm.

IPv4 waiting-list and allocation rules are a fifth. Scarcity makes queue design and eligibility rules politically and economically sensitive. A small new entrant may depend on a modest allocation. A larger actor may treat allocations as marginal compared with transfers. Policy language around need, waiting periods and resource return can shape entry conditions.

Across these surfaces, the pattern is the same. A proposal that looks technical can allocate cost. A proposal that looks administrative can alter bargaining power. A proposal that looks open can still reward those who can pay to persist.

Openness without equalisation can concentrate influence

There is a tempting but flawed defence of any open policy process: if people care, they can show up. That statement is formally true and economically incomplete. People show up when expected benefit exceeds cost. If the cost is high, low-capacity actors will rationally stay away even when the rule matters to them. Their silence is not proof of indifference; it may be proof that the process is too expensive relative to their ability to affect the outcome.

This is not a call for artificial quotas or for weakening consensus. It is a call to recognise selection bias. The visible debate is a sample of the affected community. Samples can be biased. In a policy setting, the bias is toward those with time, language, confidence, archive knowledge, commercial incentive, employer permission and repeated exposure. That group may contain much of the expertise needed to make good rules. It may still understate the costs borne by those outside the sample.

The danger increases when policy affects scarce resources. If a rule concerns a low-stakes technical convention, under-participation may be regrettable but limited. If it concerns IPv4 transfers, registry recognition, RPKI, database records, abuse duties or eligibility for scarce resources, under-participation can shift value. Those who can afford to shape the policy can lower their own costs or raise rivals' costs without ever closing the door to others.

This is the institutional-economics critique, not a moral accusation. A process can be open, transparent, civil and still structured in a way that favours repeat players. The remedy is to reduce the cost of informed contribution and to measure where contribution remains expensive.

Cost equalisation should be targeted. It should not subsidise noise or make every late complaint decisive. It should help members understand what is at stake, provide evidence in comparable form, track unresolved issues, translate summaries where useful, expose implementation costs, and invite input from groups likely to bear cost before the proposal becomes too mature to change.

Chairs have a central role. They should assess consensus with attention to cost-bearing, not only list volume. A small number of informed objections from directly affected members may deserve more weight than a larger number of general expressions of support. Conversely, repeated unsupported objections should not block policy merely because a visible actor can keep posting. Consensus judgement is partly a discipline of reading transaction costs.

RIPE NCC also has a role, even though the community owns policy. As secretariat, publisher, deadline tracker, data provider and impact-analysis author, the registry can lower information cost without deciding the substantive debate. It can make proposal pages easier to understand, publish aggregate data, support drafting, produce member-cost templates and record implementation review. Administrative support is economic infrastructure.

The aim is not perfect equality; no process can produce that. The aim is to prevent formal openness from becoming a cover for predictable concentration.

Constructive design starts before the proposal exists

Most policy-cost reduction should start before formal submission. By the time a proposal enters review, much of the agenda has already been set. Pre-proposal design can make it cheaper for operational pain to become usable policy without making every complaint a formal proposal.

The first tool is a plain-language problem statement. Before detailed text, a would-be author should be able to publish a short note: what operational problem is recurring, who appears affected, what evidence exists, what current policy or practice may be involved, what is not being proposed, and what kind of input is needed. This invites early correction. It also lets small members recognise themselves in the issue before a draft becomes technical.

The second tool is a pre-proposal clinic. The clinic should be procedural, not substantive advocacy. It could help contributors identify whether the issue belongs in policy, which working group is relevant, which prior debates matter, what evidence can be shared safely, and what alternatives exist. It should not write the policy for them or give them a privileged route. It should lower the fixed cost of responsible entry.

The third tool is an evidence template. Templates should be tailored to common surfaces: transfer delay, database accuracy, abuse-contact load, RPKI continuity, waiting-list effects, legal documentation, implementation burden and post-adoption cost. A member should be able to submit structured experience without exposing sensitive transaction details. Aggregated evidence can then be used by authors, chairs and RIPE NCC impact analysis.

The fourth tool is a prior-history map. If a similar idea was discussed before, the proposal page should summarise the prior issue, outcome, decisive concerns and unresolved questions. A link to old archives is not enough for low-cost access. The institution should not require every contributor to rediscover the same history from scratch.

The fifth tool is affected-group notice. A proposal touching transfer rules should explicitly flag buyers, sellers, brokers, small LIRs, legacy holders and merger cases. A proposal touching abuse contact should flag small access providers, hosting firms, outsourced abuse desks and security reporters. A proposal touching RPKI should flag networks with limited automation, customers relying on hosted certification, and operators with complex authorisation chains. Notice should be operational, not only categorical.

These tools preserve bottom-up control. They do not move authority to RIPE NCC staff or to a closed committee. They make it less likely that the only people capable of starting policy are those who already know how the game is played.

Better records lower the cost of staying involved

Once a proposal is active, the main design need is continuity. Contributors should not have to reconstruct the whole debate after every list burst, meeting session or version change. The proposal page should become a navigation instrument, not only an archive.

A living issue register would be the strongest improvement. Each significant concern would be captured in a short table: issue, raised by, affected text, current answer, status and next decision point. The register would not decide the issue; it would preserve it. If a concern is answered by version two, the register says so. If it is rejected as out of scope, the register says why. If it remains open for review, contributors can see that without scanning every message.

Plain-language version histories would also help. A redline shows wording changes but not always institutional meaning. A narrative change log should explain what changed in operational terms: "adds a six-month transition period for existing holders"; "narrows the rule to future transfers"; "moves documentation detail to implementation"; "adds legal review trigger"; "removes the proposed exception for merger cases." This is the difference between transparency for experts and transparency for members.

Impact analysis should be paired with a response window that ordinary members can understand. The analysis should include a short "what this may mean for members" section when relevant. If legal impact is material, the practical question should be stated. If operations or capacity impact is material, the likely member-facing effects should be described. If implementation requires new data from members, the expected burden should be clear.

Chair reason memos are another useful device. When chairs assess consensus, especially for high-impact proposals, they should explain not only that consensus was or was not reached, but why certain concerns were treated as resolved, weak, out of scope or still material. This protects chairs as well as members. A clear reason memo reduces later claims that silence was misread or that a narrow group dominated the judgement.

Finally, post-implementation review should be part of the original design for costly policies. If a policy is expected to lower transfer friction, reduce abuse-contact failure, improve database accuracy or strengthen RPKI continuity, there should be a review window after implementation. Did costs fall? Did small members face unexpected burdens? Did RIPE NCC support volume rise? Did market actors adapt in ways the proposal did not anticipate? Policy should learn from its own cost record.

Records do not need to be long to be useful. They need to be structured around the questions members actually face: what changed, why, who pays, what remains open and when can it still be influenced?

What cost-of-participation metrics should show

If transaction costs shape influence, they should be measured. Not every cost can be quantified, and measurement should not become another burden. But a registry that tracks deadlines, proposal status and impact can also track basic cost-of-participation signals.

One metric is proposal-origin diversity. How many proposals come from large operators, small LIRs, consultants, brokers, vendors, individual community veterans, academics or staff-supported groups? The point is not to rank authors by virtue. It is to see whether proposal initiation is concentrated among those with low fixed costs.

Another metric is first-time contributor survival. When a new voice raises a substantive concern, does that voice continue into later phases, or disappear after the first exchange? Disappearance may mean the concern was answered. It may also mean the cost of persistence became too high. Chairs and RIPE NCC should be able to distinguish those cases by looking at whether the issue, not only the person, remained in the record.

A third metric is affected-group response. For each high-impact proposal, the proposal page should identify likely affected groups and record whether input was received from them. If a transfer proposal receives comments from brokers and large buyers but little from small sellers or merger targets, that is not fatal, but it is information. If an abuse-contact proposal receives security-community support but little from small access providers, the cost picture is incomplete.

A fourth metric is implementation surprise. After adoption, did RIPE NCC support volume, legal review, transfer timing, database tickets or member confusion exceed expectations? If so, the policy process underpriced implementation cost. That information should feed back into future proposals.

Metrics should not become a new gate that blocks policy. They should be published as governance signals. A proposal can pass even with imperfect participation if the substantive case is strong and costs are understood. But the institution should know when it is relying on a narrow, high-capacity sample of the community.

What gets measured will not fully decide what gets fixed. It will at least make the price of influence visible.

The right standard is cost-aware openness

RIPE policy should not apologise for being demanding. Internet-number policy affects routing, registry integrity, scarce resources, security, markets and member obligations. A good proposal should require evidence, drafting discipline, public reasoning and revision. A process that made policy too easy would create instability. The question is not whether to reduce every cost. The question is which costs are necessary for quality and which costs merely protect insiders from challenge.

Necessary costs include proving the problem, writing clear text, answering objections, considering implementation and accepting public scrutiny. Unnecessary costs include forcing every newcomer to rediscover archive history, hiding distributional effects in technical language, making version changes hard to understand, treating English policy fluency as a proxy for seriousness, and regarding silence from high-cost members as evidence that they are unaffected.

Cost-aware openness keeps the strengths of RIPE's model. It preserves bottom-up initiation, working-group discussion, public archives, consensus and RIPE NCC impact analysis. It adds a more explicit concern for who can afford to use those tools. It asks whether the process is not only open, but usable by the affected range of members.

For RIPE NCC, this is also a strategic legitimacy issue. The registry's public value rests on trust in records, services and community policy. IPv4 scarcity, transfer markets, RPKI reliance, database accuracy and regional diversity make that trust more economically consequential than it was when address abundance reduced the stakes. If small members believe policy is formally open but practically owned by repeat players, trust becomes more expensive. If repeat players believe every late operational objection can derail policy, governance becomes unstable. Cost-aware design is the middle path.

The practical test for future proposals is simple. Can a small LIR that discovers a recurring transfer, RPKI, database or abuse-contact burden understand the current rule, find the relevant history, describe the problem, submit usable evidence, draft or support text, follow revisions, comment on impact analysis, appear remotely if needed and check implementation without becoming a policy professional? If the answer is usually no, then formal openness is doing less work than the institution thinks.

The better goal is not frictionless policy. It is policy whose friction is justified, visible and fairly distributed. In a scarce-address world, that is not administrative housekeeping. It is part of the economics of registry legitimacy.