A Microphone Is Not a Lever

Summary

• The IGF was designed to support discussion, information exchange, advice, capacity and issue discovery while remaining non-binding and outside day-to-day technical operations. That boundary protects open dialogue, but it also means that speaking is only the first event in any implementation chain.
• A trackable recommendation must identify a requested act, a competent receiving body and the problem to be solved. A transcript, applause line, session summary or broad aspiration may be valuable, but none supplies a usable implementation denominator by itself.
• Forum follow-through should be reported as a sequence: eligible recommendation, acknowledged receipt, reasoned disposition, responsible owner, due date, authorized decision, implementation evidence and remedy. Rejection with reasons counts as disposition; it does not count as implementation.
• Different outcomes require different levers. Laws move through public institutions, registry policy through regional procedures, corporate obligations through contracts and boards, standards through standards bodies and implementers, disputes through review or courts, and routing changes through network operators.
• Resolution 80/173 created a clearer route for IGF outcomes into relevant UN processes after 2025. It improved transmission and institutional continuity, but consideration is still not execution. The receiving institution remains answerable for the act it adopts or refuses.
• The strongest forum is not one that claims every idea as an impact. It is one that shows where each concrete recommendation went, what happened next, what remains unresolved and which institution can be held to its own authority.

Athens created a public room, not a command chain

The first IGF met in Athens in 2006 under a mandate negotiated at the World Summit on the Information Society. The Tunis Agenda gave the forum a broad field: discuss public-policy issues, exchange information and good practice, advise stakeholders, identify emerging issues, strengthen capacity and publish proceedings. The same instrument set a precise boundary. Paragraph 77 denied the forum an oversight function, kept it from replacing existing arrangements, excluded day-to-day and technical operations and made its work non-binding.

That combination was not an administrative accident. It made the room possible. Governments could enter without conceding treaty power to a new body. Companies and technical institutions could explain practice without submitting every operational decision to a global vote. Civil society could challenge both without first obtaining standing before a regulator or court. Engineers could describe constraints while policymakers tested social consequences. The price of this openness was that no statement from the floor would operate as an order.

The official Athens record preserves consultations, programme decisions, transcripts and proceedings. Those records establish that people met and spoke. They can show when an issue entered the forum's public memory. They do not show that the person responsible for a registry, ministry, company or network accepted a duty to act.

The distinction should have framed impact claims from the beginning. The forum supplies visibility, encounter and public reasoning. A separate institution supplies jurisdiction, procedure and execution. When those stages are collapsed, the meeting receives credit for powers it does not have while the actual decision maker escapes scrutiny.

Speech has four possible afterlives

A forum intervention can disappear, remain a record, travel as influence or become part of an enforceable act. These are not points on an automatic progression.

The first afterlife is disappearance. The statement may be heard but omitted from the session report, lost among parallel sessions or left without a person willing to carry it further. This does not prove suppression. Conferences have finite attention, and many interventions are incomplete, repetitive or outside scope. It does mean that no follow-through claim should be made.

The second is preservation. A transcript, video, summary or message records the concern. Preservation matters because later actors can find the argument, compare years and challenge a false claim of consensus. Yet a record is evidence of speech, not evidence of institutional receipt.

The third is influence. A entity may take the idea into a ministry, company, standards group, registry community or research programme. Language may reappear in a consultation or proposal. Influence can be real even when no formal referral exists. It should be described with calibrated attribution: the forum informed, connected, clarified or accelerated the later work.

The fourth is authorized effect. A legislature enacts, a regulator orders, a board resolves, contracting parties amend, a policy process adopts, a court grants relief, a standards body publishes, or an operator changes production practice. At that point the legal or operational source of the result is the later act. The forum may belong in the history, but it is not the instrument that binds.

Measuring all four as “impact” makes the term useless. A serious follow-through account labels the stage and names the evidence.

The recommendation must exist before it can be followed

Not every view expressed at a forum is a recommendation. “We need a safer Internet” states an objective. “Governments should do more” identifies a broad class of actor. “The community must act” often identifies nobody at all. Such statements may orient discussion, but they cannot be assigned, timed or completed.

A recommendation becomes eligible for follow-through measurement when it contains three minimum elements. It names the requested act. It names a receiving body that is capable, at least plausibly, of taking that act. It states the public or operational problem the act is meant to address. Evidence and a proposed date make it stronger; they are not substitutes for the three essentials.

Consider a session on route-origin security. “RPKI should be more resilient” is a finding. “The relevant registration service provider should publish a tested continuity plan for its hosted certification service before the next annual review” is a recommendation. “Every network must reject invalid routes tomorrow” is concrete but may be unsafe and directed to no single body with authority over every network. Precision exposes both feasibility and institutional limits.

This eligibility rule protects the forum from a perverse metric. If every sentence enters the denominator, the follow-through rate will be meaningless and easily manipulated. Organisers could improve the number by counting only ceremonial commitments or depress it by counting every aspiration. The denominator should consist of concrete, attributable requests whose due dates can eventually be assessed.

Outputs should preserve broader findings separately. A forum can reveal uncertainty, disagreement or a question that is not ready for action. The discipline is not to force every insight into a recommendation. It is to stop calling an insight implemented when no implementation claim can be tested.

A session report is a record, not a receipt

The IGF 2025 outputs page shows the variety of records produced around one annual meeting: Lillestrom Messages, a summary report, session reports prepared by organisers, transcripts, videos, youth outputs, a parliamentary track and intersessional publications. The meeting contained 262 sessions. This is a substantial public archive. It is not one institutional decision.

Each format answers a different question. A transcript asks what was said. A session report asks how organisers summarized their session. Annual messages synthesize themes across many discussions. A policy brief or coalition paper may develop a narrower argument over months. A parliamentary output may be intended for legislators. None should silently inherit the authority of the others.

Receipt is a later event. The named receiving institution must acknowledge that the recommendation reached a place where it can be considered. Evidence might be a public docket, a numbered consultation submission, an agenda entry, a letter with a stable date, a policy-proposal page, a board-paper reference or a court filing. The form will differ by institution. The essential fact is that responsibility no longer rests only with the speaker and session organiser.

The presence of an official in the room is not receipt. A civil servant may attend without authority to commit a ministry. A registry employee may listen without controlling a community policy process. A company delegate may participate personally or within a narrow brief. Even the head of an institution may need a board, legislature, membership or formal notice procedure before the institution acts.

Counting attendance as receipt allows every body to claim it was informed while nobody owns the next step. A public acknowledgement prevents that ambiguity.

The recipient must be competent for the requested act

Naming a recipient is not enough. The recipient must have authority over the entity of the recommendation.

Internet governance language often uses large nouns: the United Nations, the technical community, industry, civil society, the private sector or the global Internet community. These labels describe fields of participation. They do not identify the person who can sign, vote, configure, register, pay or adjudicate. A recommendation addressed to “the technical community” may be less actionable than one addressed to a specific operator's routing team, RIR policy group or standards working group.

Competence has several forms. A legislature can enact within constitutional limits. A regulator can issue rules or orders within delegated jurisdiction. A court can decide a dispute brought by a party with standing. A corporate board can direct the company within its governing instruments. Contracting parties can change duties between themselves. A Regional Internet Registry can administer records and services under its policies and agreements. A standards body can publish a specification. A network operator can decide how its routers treat available data, subject to law and contract.

The functional analysis of multistakeholder institutions developed by Mark Raymond and Laura DeNardis is useful because it refuses to treat Internet governance as one process. Different tasks require different authority relations. The right forum for discussion is not automatically the right body for execution.

A recommendation receipt should therefore include an authority note. It should state the provision, contract, policy, corporate power or operational control through which the recipient could act. If no competent recipient exists, the first recommendation may need to be institutional: create or designate a body with bounded authority and review.

Disposition is valuable even when the answer is no

Follow-through should not reward automatic agreement. A forum can produce weak, unlawful, unaffordable or technically dangerous proposals. The receiving body has a duty to use its own evidence and mandate, not to ratify meeting enthusiasm.

Disposition means that the recipient has made the recommendation legible inside its own procedure. It may accept the request, accept part of it, modify it, refer it to another competent body, reject it, defer it pending evidence or close it as outside scope. The reasons matter more than the label. A rejection should identify the authority limit, evidentiary gap, cost, conflict or alternative route. A referral should name the new owner rather than send the request into institutional fog.

This is why disposition rate should be reported separately from implementation rate. If ten recommendations are received and all ten receive reasoned decisions, the receiving system is responsive even if only four are adopted. Calling the other six failures would pressure decision makers to accept for appearances. Conversely, if ten recommendations are repeatedly celebrated but eight never receive an answer, a later success story about two should not conceal the silence.

The disposition record also protects dissent. A session summary may compress disagreement into a common direction. The receiving body should see material objections and answer them where they affect feasibility or rights. The 2012 CSTD improvement report called for more tangible outputs while preserving converging and diverging views. That is not just a reporting preference. It helps prevent an energetic room from becoming fictitious unanimity downstream.

A responsible body needs a responsible owner

Institutions act through people and offices. “The ministry will consider” is not assignment. “The registry community will discuss” is not ownership. A receiving body should identify the office, committee, working-group chair, board sponsor, contract manager or operational team responsible for the next procedural event.

The owner does not have to promise the final result. A policy officer can be responsible for registering a proposal without deciding consensus. A legislative committee can schedule scrutiny without guaranteeing enactment. A standards chair can manage review without dictating technical agreement. Ownership means accountability for movement and accurate status, not unilateral control over a collective outcome.

This distinction is especially important in open processes. A forum entity may be told that “anyone can submit a proposal.” Formally that is access. Practically it can become abandonment if nobody helps translate a broad concern into the receiving procedure's required form. The owner should identify missing evidence, applicable scope and the next decision point. Assistance does not guarantee adoption; it prevents procedural opacity from serving as a silent veto.

Public ownership also limits reputational arbitrage. Senior figures often endorse principles at global meetings while implementation responsibility sits with a less visible office. The institution receives the benefit of association with the principle but can later say that no actionable request reached the correct unit. A receipt that names both the institution and its responsible owner closes that gap.

When responsibility changes, the record should change with it. Staff turnover, elections and reorganizations are normal. The recommendation should remain attached to an office and mandate rather than disappear with the person who first accepted the handoff.

A date turns attention into a test

Without a date, “under consideration” can last longer than the problem. A deadline does not force a particular answer, but it creates a point at which delay becomes visible.

The relevant date depends on the act. A regulator may publish a consultation calendar. A legislature may work within a session. A registry policy proposal may have discussion, review and last-call periods. A contract dispute may require notice and cure before escalation. An emergency service case may need hours or days, while a standards change may reasonably take months or years. Imposing one universal clock would be performative rather than rigorous.

Every recommendation should nevertheless have a next-event date. If final implementation cannot be predicted, the recipient can state when it will decide scope, publish evidence, open consultation, issue an impact assessment or report another status. Contingencies should be explicit. “Within sixty days after the consultation closes” is testable. “In due course” is not.

The RIPE Policy Development Process illustrates the value of staged time. It identifies proposal, discussion, review and concluding phases, requires actual timescales to be documented when defaults vary, and connects accepted policy to implementation where necessary. The process has its own legitimacy questions, but it demonstrates that open participation and deadlines can coexist.

A forum follow-through register should record original due dates and revisions. Moving a date may be justified by evidence or consultation. Erasing the first date destroys the ability to distinguish prudent extension from habitual delay.

Implementation evidence must show a changed state

An institution has not implemented a recommendation merely because it repeated the language. Implementation changes a legal, contractual, financial, registry, technical or administrative state.

For public law, the evidence may be an enacted statute, final rule, licence condition, appropriation, judgment or published executive act. For a company, it may be a signed contract amendment, board decision, service commitment or completed corrective measure. For number-resource policy, it may be an adopted policy text, an implementation announcement and observable service behavior. For standards, it may be a published specification followed by independent implementations. For a network, it may be a deployed configuration, tested control or measured operational outcome.

Publication alone is sometimes the requested act. If the recommendation is to disclose a continuity plan, a complete and dated disclosure may be implementation. If the recommendation is to improve continuity, publishing a plan is only an intermediate output. The requested entity must remain stable or the institution can downgrade an outcome while claiming success.

Evidence should also identify scope. A pilot in one office is not national implementation. One operator deployment is not industry adoption. A policy announcement is not complete if the necessary service remains unavailable. A judgment may declare a right without securing compliance. These distinctions do not diminish progress; they keep the public record accurate.

The most credible status labels are plain: not received, received, under review, referred, rejected with reasons, adopted, partly implemented, completed, superseded or overdue. A separate evidence link should support each transition. Adjectives such as landmark, historic and transformative do not establish state change.

Remedy completes the authority chain

Implementation can harm as well as help. A recommendation may become a rule that burdens an operator, a registry action that impairs continuity, a procurement condition that excludes a provider or a technical default that creates new failure modes. The authority chain is incomplete unless an affected party can challenge the later act.

The remedy belongs primarily to the receiving institution's legal or contractual setting. A regulation may be reviewed by a court. A registry service decision may have an appeal, arbitration or contractual dispute route. A corporate act may be challenged under an agreement or governing law. A standards decision may have a process appeal, while deployment remains a choice for implementers. An operator configuration may be rolled back or reviewed through the operator's incident and change controls.

The IGF should not become the appellate body for every downstream act. Its founding design gives it neither universal standing rules nor enforcement machinery. It can, however, refuse to describe an adopted recommendation as complete when the resulting institution provides no correction path. A recommendation affecting rights should name the expected review route before adoption, not after the first injury.

Remedy also disciplines attribution. If a ministry adopts a forum-informed rule, the ministry cannot invoke the multistakeholder origin to avoid legal review. If a registry adopts a proposal discussed at the IGF, the registry remains responsible under its own agreements and procedures. “The community wanted it” is not a defence against a claim that the authorized body exceeded its power or ignored evidence.

Influence without responsibility is easy to celebrate. Remedy makes the later institution own the consequences.

The public-law path runs through constitutional authority

Some forum recommendations are directed to governments: amend a law, fund access, protect rights, regulate a platform, strengthen cybersecurity or improve procurement. A minister's conference statement can signal intent, but enforceability depends on domestic authority.

The path may require cabinet approval, legislative text, committee scrutiny, appropriation, regulatory consultation, publication and judicial review. In another constitutional system, the sequence will differ. The common principle is that the forum neither supplies the electorate nor substitutes for public-law procedure. A government representative can carry evidence home; the state must still act through the institutions authorized to bind people.

This boundary protects both democracy and the forum. Without it, critics can portray every global discussion as an attempt to legislate at a distance. With it, entities can be ambitious about ideas while exact about their legal status. The recommendation receipt should identify the domestic body and procedural entry point: ministry review, bill, regulatory docket, budget request or judicial claim.

Public-law follow-through also needs territorial scope. A national act may affect companies and networks within jurisdiction; it does not become a global Internet rule because the discussion was global. Cross-border effect may arise through market size, treaty, conflict-of-laws rules or voluntary adoption. Those are separate mechanisms requiring separate evidence.

A reasoned government rejection can be legitimate. The proposed act may violate constitutional rights, exceed jurisdiction or impose costs unsupported by evidence. The follow-through test asks whether the government owned and answered the request, not whether it obeyed the room.

The number-policy path begins with a proposal, not applause

Internet number policy is often discussed at global and regional meetings, but a forum intervention does not alter allocation, transfer or registration rules. The receiving route is the applicable regional or global policy procedure.

The APNIC Policy Development Process moves from pre-meeting discussion through open policy meeting consensus, post-meeting discussion, confirmation and Executive Council endorsement. RIPE uses documented proposal, discussion, review and concluding stages. ARIN's PDP has its own proposal, consultation, Advisory Council and Board structure. The procedures differ, and none treats an IGF panel as a regional policy vote.

A speaker who wants a number-policy change therefore needs a named proposer, draft text, affected policy clauses, impact analysis, discussion record and implementation owner. The forum can improve the problem statement and recruit entities from outside the usual policy circle. It can reveal consequences that a regional list has missed. Those are substantial contributions. They still enter the policy system as evidence and proposals, not as instructions.

Scope must be watched carefully. Some matters called policy are actually service practice, contract interpretation or individual disputes. A policy group may be unable to order emergency restoration for one operator. A service team may be unable to change a community-adopted eligibility rule. A board may have implementation duties but no legitimate power to manufacture consensus. The receipt should identify which branch owns which part.

This is the difference between access and leverage. Anyone may be invited to speak. The lever is the defined procedure capable of changing the rule.

The standards path ends in implementation, not publication

Technical standards offer another common source of confusion. A forum may identify an interoperability or security problem and recommend a standard. The recommendation must then enter a standards organization with a charter, technical review, consensus practice and publication process.

RFC 2026 describes the Internet standards process. RFC 7282 explains why rough consensus is not applause, a simple vote or the absence of objection. Technical objections must be understood and answered. This discipline is valuable precisely because a compelling conference narrative may conceal deployment costs, incompatible implementations or security regressions.

Publication remains an intermediate event. An RFC does not reconfigure every network. Vendors must implement, operators must deploy and relying systems must use the result. Market adoption, procurement or regulation can make a standard practically important, but each route has a different authority and remedy. A standards body should not claim deployment merely because the document exists; a forum should not claim a standard merely because a working group later addressed the same topic.

The follow-through record should therefore separate standards receipt, chartering, draft adoption, publication, independent implementation, interoperability testing and operational use. It should also preserve failure. A technically sound idea may not be adopted because the installed base is too large or migration risk too high. That is not necessarily institutional neglect.

The meaningful lever in engineering is a changed, tested behavior. The microphone can help identify why that behavior is needed. It cannot provide running code by acclamation.

The contract path creates duties between identifiable parties

Many Internet-governance effects arise through private agreements: registry service terms, registrar contracts, cloud commitments, peering arrangements, procurement clauses and platform terms. A forum can expose an imbalance and propose better language. The duty begins when authorized parties execute or validly amend the agreement.

Contract follow-through needs specificity. Who are the parties? What service or resource is covered? Which obligation changed? When does it take effect? Can one party amend it unilaterally? What notice and cure apply? Which law and forum govern? What relief is available? A public pledge that answers none of these questions may shape reputation, but it is not a contractual lever.

Contracts also have limited reach. A company can bind itself and, within law, its counterparties. It cannot bind every user, operator or competitor merely because its representative joined a multistakeholder discussion. Industry codes can broaden coverage when companies adopt them, while legislation or regulation may give some duties wider effect. The source of each obligation should remain visible.

Remedy distinguishes a commitment from a slogan. If a service provider promises continuity but excludes every meaningful consequence for breach, the obligation may be thinner than the public language suggests. If an affected party lacks standing, evidence access or timely relief, nominal enforceability can still fail in practice.

A forum follow-through record should therefore link to the executed instrument or public terms, not just the announcement. It should state who can invoke the obligation and whether the recommended protection survived negotiation.

The corporate-accountability path has defined standing and scope

Some recommendations concern the conduct of a coordinating corporation such as ICANN. Here too, the name of the forum is less important than the instrument that authorizes review.

The ICANN Bylaws define a mission, commitments and accountability mechanisms including reconsideration and independent review. These processes can test specified acts against corporate obligations. They have rules of standing, timing, scope, evidence and remedy. They do not review every Internet institution merely because ICANN participates in global governance discussions.

A forum recommendation aimed at ICANN may proceed through public comment, a supporting-organization or advisory route, a board paper, a contractual negotiation or an accountability mechanism. Each path asks a different question. Public comment seeks input before or during decision. Reconsideration examines specified action or inaction under its standard. Independent review addresses compliance with the constituting commitments. A contract route concerns the parties and clauses involved.

The follow-through record should identify which route was chosen and why. Sending the same document to several inboxes can create the appearance of activity without establishing standing anywhere. Conversely, a narrow procedural challenge should not be described as a referendum on the whole governance model.

This discipline also protects critics. If they lose, the record can show whether the reason was lack of jurisdiction, missed time, weak evidence or a decision on the merits. “The institution listened” is not enough; “the institution decided under this power and this review standard” is the testable statement.

The operator path ends at the network that carries the risk

The Internet Numbers Registry System maintains globally unique registration through a hierarchy involving IANA, Regional Internet Registries, Local Internet Registries and consumers. The same RFC is explicit that route announcement and advertisement are operational matters outside the registry system. This division matters whenever a forum discusses routing security or reachability.

A registry can publish records, provide certification services and implement resource policy. A resource holder can authorize certain statements. A relying party can validate available data. The network operator decides how validated information affects route selection, filtering, monitoring and incident response, subject to its legal and commercial obligations. No forum chair can make all those networks accept one policy.

An operator-facing recommendation should therefore identify the actual control: create a ROA, change a validation policy, diversify caches, test stale-data behavior, improve incident escalation or publish a continuity objective. It should state which operator or class of operator is asked to act and what evidence would demonstrate safe deployment. “Adopt routing security” is too broad to measure.

The operator also owns the downside. A bad filter can remove legitimate reachability. A rushed certificate transition can create invalid states. A recommended security control may interact with customers, upstreams and emergency procedures. Follow-through cannot be reduced to adoption count; it must examine reliability and correction.

This is why implementation authority belongs close to the risk-bearing network. Forum discussion can widen the evidence. Operational responsibility cannot be transferred to the audience.

A worked example: from an RPKI concern to accountable action

Suppose a small operator describes a recurring risk at an IGF session: its hosted RPKI service depends on one registration provider, and the operator cannot identify a published continuity commitment for a prolonged institutional outage. The intervention receives attention and appears in the session report. What happens next?

First, organisers formulate an eligible recommendation rather than paraphrase the concern: the named provider should publish and test a continuity plan covering certificate issuance, repository availability, holder communication and recovery, with a defined response date. The operator confirms that this wording reflects the problem.

Second, the provider acknowledges receipt and identifies whether the request belongs to service operations, contract terms, technical policy or community policy. It names an owner and a date for scope determination. If part requires policy change, a proposer enters the applicable regional procedure. If part concerns a current contractual duty, the operator may use the service dispute route without waiting for community consensus.

Third, any technical proposal is checked against RPKI architecture, repository and relying-party behavior. The provider may publish a plan, run a documented exercise and correct weaknesses. Operators independently decide how to configure validation and fail-safe behavior.

Fourth, the record links the forum output, receipt, provider decision, policy proposal if any, implementation evidence and review route. If the provider rejects the request, it states whether the reason is security, cost, authority or existing coverage. If a later act impairs the operator, the relevant contract, arbitration or court route remains available.

The forum has influence at the beginning and may convene review at the end. The levers in the middle belong to named institutions.

Follow-through is a family of rates, not one vanity score

A single impact number can hide every institutional failure. The better method reports a sequence for each annual cohort of eligible recommendations.

The receipt rate is acknowledged recommendations divided by eligible recommendations. The disposition rate is recommendations accepted, modified, referred, rejected or closed with reasons divided by acknowledged recommendations. The on-time decision rate is dispositions reached by the stated date divided by dispositions due. The adoption rate is recommendations converted into an authorized decision divided by eligible recommendations due for decision. The implementation rate is completed authorized acts divided by adopted recommendations due for completion.

The remedy-availability rate is rights-affecting implementations with a published review route divided by all rights-affecting implementations.

The end-to-end follow-through rate is narrower: completed implementations divided by eligible recommendations whose implementation dates have passed. It should never replace the component rates. A low end-to-end rate may reflect poor receipt, slow procedure, repeated rejection or difficult implementation. Those conditions require different remedies.

Time measures matter as much as percentages: median days to receipt, disposition and completion, plus the oldest overdue item. Medians should be accompanied by tail measures because a few operators can bear severe harm while the typical recommendation moves quickly.

The unit of analysis must remain stable. One recommendation cannot be split into ten easy subtasks to inflate completion. Ten distinct recommendations cannot be merged into one indefinite theme to conceal non-performance. Material changes require a public note.

Most importantly, no current long-run IGF rate should be invented. The public records reviewed here do not provide the complete denominator and status fields needed to calculate one.

Attribution requires more than chronology

If a policy changes after an IGF session, the forum may have contributed. “After” is not “because of.” The later institution may have been working on the issue already, responding to litigation, market events, elections, technical incidents or another consultation.

A credible attribution record looks for links. Did the receiving body's paper cite the forum output? Did an official acknowledge that the intervention changed scope or timing? Did the same proponent carry the recommendation into the authorized process? Did meeting entities form a project that produced new evidence? Did decision makers describe the forum as one input among several? These signals support contribution without requiring an implausible claim of sole causation.

Independent verification is especially useful for celebrated success stories. The institution running the forum has an incentive to show value; the receiving institution has an incentive to show consultation. Their accounts may both be accurate and still overstate causality. A dated sequence, document comparison and interviews with decision entities can test the link.

Failure deserves the same care. A recommendation may not proceed because the receiving body lacked authority, evidence changed, the proponent withdrew, another reform superseded it or implementation created unacceptable risk. Labeling every unimplemented idea “ignored” can be as misleading as labeling every later change “impact.”

The goal is not to deny the forum credit. It is to give credit a defensible size. Agenda influence, translation, coalition formation, policy adoption and operational execution are different achievements.

Resolution 80/173 strengthens transmission, not automatic conversion

The permanent-IGF decision adopted in December 2025 changes the follow-through environment. General Assembly Resolution 80/173 asks the IGF to report annual and intersessional outcomes to relevant UN entities and processes and asks those entities to take the outcomes into account. It also strengthens intersessional work, participation, staffing and sustainable-resource objectives.

This is a clearer institutional bridge than a speaker hoping that the right official remembers a panel. The UN DESA follow-up table maps requested actions to asked entities and provides a model of explicit assignment. Paragraph 124's biennial reporting can create another place to record progress across WSIS commitments.

The verbs remain bounded. Reporting is not adoption. Taking into account is not implementation. A UN entity may consider an IGF outcome and reject it, adapt it, refer it or act under a separate mandate. The responsible entity should say which. Otherwise the new bridge risks becoming a better distribution channel for documents without a stronger account of results.

Permanent status makes longitudinal measurement more feasible. The forum no longer needs to treat every cycle as potentially final. It can maintain recommendation cohorts across years, publish overdue items and compare receiving bodies. Stable capacity can support this work without turning the Secretariat into an enforcement agency.

The opportunity after 2025 is therefore procedural: make transmission visible enough that institutions cannot use the forum as proof of inclusion while leaving concrete requests ownerless.

The forum should publish a recommendation receipt

Each eligible recommendation should leave the meeting with a short public receipt. The receipt would name the session and authors, state the requested act, identify the receiving body, preserve material objections, name the receiving office if acknowledged, record the authority route, set the next-event date and link later evidence.

The receipt should also state status. It is a entity recommendation, organiser synthesis, coalition output, parliamentary statement or other defined product. It should never imply IGF-wide consensus unless a procedure actually supports that claim. The receiving body should be able to correct an inaccurate description of its mandate without editing the recommendation itself.

Privacy and safety require limits. A person disclosing abuse, repression or a security vulnerability may need protected handling. The public record can describe the institutional issue and status without exposing a vulnerable speaker, exploit detail or confidential evidence. Openness should illuminate accountability, not create another harm.

The receipt is not a command. A receiving body does not acquire a duty merely because organisers named it. The body can reject competence and identify the appropriate route. That response is useful because it reveals an authority gap that discussion alone often conceals.

Nor should every session be forced to issue recommendations. Some sessions exist to compare evidence, surface uncertainty or build capacity. Their outputs should say so. A smaller set of concrete recommendations with complete follow-through is more informative than hundreds of synthetic action points detached from willing owners.

What a speaker should do after leaving the stage

A entity seeking an enforceable result should treat the intervention as the beginning of a second, more specific task.

First, write the requested act in one sentence and identify the institution that controls it. Second, ask the session organiser to preserve the statement accurately and distinguish it from general discussion. Third, submit the request through the receiving body's recognized procedure and obtain a dated acknowledgement. Fourth, identify the legal, contractual, policy or operational authority under which the body could act. Fifth, ask for the next decision date and the evidence required. Sixth, preserve the review route before accepting an implementation that may affect rights.

These steps may feel less dramatic than a plenary. They are where leverage appears. A carefully drafted regional policy proposal can matter more than a widely shared clip. A contract notice can protect continuity more directly than a declaration. A regulator's docket, court filing or operator change record can reveal progress that conference communications miss.

The speaker should also be prepared for a legitimate no. Advocacy is not entitlement to adoption. The aim is a reasoned decision by the right institution, followed by accountable implementation where adopted.

Organisers can help by inviting receiving bodies to state procedures before the session. A panel on registry services should tell entities where policy, service and dispute requests go. A session on standards should identify the relevant working groups and implementation constraints. Access to a microphone becomes more useful when access to the next institution is intelligible.

Meeting energy is an input, not an outcome

Large attendance, a full room, remote participation, social attention and senior speakers can show relevance. They can widen discovery and impose reputational pressure. They cannot show that a recommendation was received, decided or implemented.

Energy metrics are attractive because they are immediately available. Registration totals can be announced at closing. Session counts and online views can be displayed. Follow-through is slower, less flattering and distributed across institutions that the forum does not control. It requires organizers to preserve uncertainty and share credit.

That is precisely why it is a better test. A forum devoted to governance should reveal the path from voice to responsibility. It should show when influence stops and another mandate begins. It should make non-action visible without pretending to command action.

The microphone remains valuable. It can let a small operator place a continuity risk before institutions that would not otherwise meet. It can let an advocate challenge a comfortable technical assumption, or an engineer expose the operational cost of a political slogan. The mistake is not speaking. The mistake is treating the opportunity to speak as the remedy.

A lever changes state through a fulcrum, force and defined point of application. In institutional terms, those are authority, procedure and responsibility. The forum can help a person find them. It should never claim that the sound system supplied them.

Sources and scope

The mandate and historical boundary are grounded in the Tunis Agenda, the first IGF archive and the 2012 CSTD improvement report. Current output and follow-up analysis uses the IGF 2025 outputs, Dynamic Coalition guidance, 2025 stocktaking suggestions, Resolution 80/173 and the UN DESA follow-up table.

Receiving-body examples are bounded to the RIPE, APNIC and ARIN policy procedures, the Internet standards process, ICANN Bylaws and RFC 7020. Functional interpretation is informed by Raymond and DeNardis, DeNardis on infrastructure control and Kettemann on the IGF's normative role.

No complete public dataset reviewed through 15 July 2026 supports a historical IGF-wide follow-through percentage. The rates and receipt design in this article are proposed evaluation methods. They should be tested prospectively with published denominators rather than reconstructed into a false measure of past performance.