Summary
- The 2014 NETmundial statement did not merely announce broad values. It asked Internet-governance organizations to publish periodic progress reports, suggested continuing monitoring and information-sharing tools, and identified benchmarking systems and indicators as unfinished work. The memory problem was visible at the beginning.
- The 2024 consultation measured the views of its contributors and produced a rich record of proposals. It did not perform a representative, institution-by-institution audit of implementation since 2014. Agreement that principles remain relevant is evidence of continuing support, not evidence of compliance.
- NETmundial+10 proposed that the IGF serve as depositary or caretaker of the new Sao Paulo Multistakeholder Guidelines. Yet the statement left metrics, assessment systems, accountability arrangements and further guidance for possible future discussion, while acknowledging that the IGF would need new methods and resources.
- A durable memory function should preserve versions, define the process being assessed, assign evidence to each principle, publish dissent and negative findings, allow factual correction, track remedial commitments and distinguish self-reporting from independent review. It should not turn a non-binding forum into a regulator.
An anniversary should be an audit, not a ritual of renewal
Ten years is long enough for a principle to acquire a reputation without acquiring an implementation history. Institutions cite it in speeches, place it in opening declarations, and borrow its most agreeable adjectives. New entities encounter the principle as settled inheritance. The difficult evidence - who changed a procedure, who refused, who lacked money to participate, whose objection disappeared, and whether any remedy worked - becomes harder to reconstruct than the statement itself.
That is the institutional memory problem. It is not ordinary forgetfulness. Documents remain online, entities retain recollections, and anniversary meetings can recover the language. The failure occurs when no body is obliged to maintain a continuous account connecting the language to conduct. In that gap, survival of the phrase is treated as survival of the practice.
NETmundial+10 was well placed to expose the gap. The original meeting occurred in Sao Paulo in April 2014. Its Multistakeholder Statement described itself as non-binding and intended its recommendations to contribute to other forums and entities. It established no permanent NETmundial organization. The 2024 meeting was therefore not the tenth annual review of an operating institution. It was a new convening that chose to retrieve, reaffirm and extend the work of a one-time meeting.
That choice was valuable. It also revealed the weakness of relying on anniversaries. If no convenor had assembled NETmundial+10, there was no standing NETmundial office required to explain what had happened to the process principles. A durable principle cannot depend entirely on the continued interest of people capable of recreating its forum.
The appropriate standard is not whether the 2024 entities praised the 2014 text. It is whether the intervening decade left enough organized evidence to determine where the principles changed institutional behavior, where they failed, and who must answer for the next correction.
The 2014 text already knew that principles needed a memory
The ten 2014 process principles were more operational than a ceremonial declaration. They addressed multistakeholder participation, openness, consensus, transparency, accountability, inclusion, distribution, collaboration, meaningful participation, access and agility. The accountability principle expressly called for independent checks and balances, review and redress. The transparency principle required understandable decisions, documented processes and agreed procedures.
The roadmap went further. It said that organizations with responsibilities in the Internet-governance ecosystem should develop and implement transparency, accountability and inclusion principles. It asked them to prepare periodic public reports on progress and status. It suggested periodic reports, formal liaisons and timely feedback as coordination mechanisms. It recommended analyzing tools for continuing monitoring, analysis and information sharing.
The statement even named the missing measurement problem. Among subjects requiring further discussion, it listed benchmarking systems and indicators for applying Internet-governance principles. Its way forward encouraged existing organizations, forums and processes to consider the outcome, continue the listed discussions and report on their work at major Internet-governance meetings.
This was an embryonic memory architecture: periodic reports, coordination, monitoring, analysis, indicators and recurring public discussion. But the architecture had no named builder. "All organizations" could report voluntarily; existing entities were invited to continue work; indicators were deferred. No body had to define a common reporting period, compare institutional claims, identify non-reporters, preserve a versioned baseline or state when a principle had failed.
The wording created a collective expectation without a counterparty. Every relevant body could say that another forum was the natural place for synthesis. A meeting could cite examples without covering the field. An institution could report activity without explaining outcomes. The roadmap made memory desirable but not compulsory.
That omission did not invalidate the principles. A non-binding meeting could not simply appoint itself supervisor of unrelated states, companies, standards bodies and community organizations. The omission does explain why a later meeting could find the same implementation question unresolved. A request addressed to everyone is often an assignment accepted by no one.
Relevance, adoption and performance are different claims
The first discipline of an implementation review is to separate three propositions. A principle may remain relevant because the problem it describes still exists. It may be adopted because an institution incorporates it into rules, terms of reference or routine practice. It may perform because the adopted mechanism changes participation, decisions or remedies in the intended direction.
These propositions require different evidence. Continuing relevance can be supported by current examples of exclusion, opacity or uncorrected error. Adoption requires a dated institutional act and a defined scope. Performance requires observations before and after adoption, or another credible comparison, showing what happened in practice.
NETmundial+10 was strongest on relevance. Its 2024 Multistakeholder Statement reaffirmed all ten process principles and said they remained valid amid technical, social and economic change. It was candid that they had not yet been fully implemented. It identified persistent concern that governance processes failed to include all relevant stakeholders meaningfully.
That is an important finding, but it is broad. "Not fully implemented" does not identify which institutions were examined, which principle each failed, what evidence supported the finding, or whether the failure was absence of a rule, weak execution, inadequate resources or deliberate refusal. It is a diagnosis at ecosystem level without a case register.
The risk appears when continuing support is treated as proof of performance. Entities can strongly agree that transparency matters while processes remain opaque. An institution can adopt a participation policy while setting deadlines that exclude less-resourced communities. A forum can publish every submission while never explaining how any submission affected the result. A formal appeal can exist while delay makes the remedy useless.
An anniversary report must therefore resist the most flattering inference. The principle's rhetorical durability is not the result to be measured. It is the starting hypothesis.
The 2024 consultation measured a community, not a decade
NETmundial+10 produced a substantial consultation record. Its Final Report explains that an open online consultation received 154 contributions. The organizers used Likert-scale questions, rankings and open questions; coded 2,766 content units as proposals, comments, examples and other categories; published contributions; and used the material to prepare a draft for refinement during the two-day event.
This is serious procedural evidence. It records what a self-selecting group of contributors thought in 2024, what improvements they proposed, and which options they ranked highly. It helps reconstruct the drafting environment. It is also not an implementation evaluation of the preceding decade.
The distinction follows from the questions. Contributors were asked whether the 2014 principles remained relevant, whether limited public evidence inclusion contributed to persistent difficulties, whether most processes applied flexible stakeholder roles, whether opportunities for non-governmental participation had improved, and what improvements should be made. Those are perception and design questions. They do not sample a defined population of institutions or inspect decisions against published criteria.
A consultation respondent may have deep experience, but the response is still testimony. An institutional audit would ask for the rule in force, the notice period, participation data, financial support, draft history, decision record, dissent, appeal, remedy, implementation date and later outcome. It would distinguish a national regulatory consultation from a standards working group, an address-policy process, an intergovernmental negotiation and an industry code because their authority and participation models differ.
The consultation did not claim to be a representative global survey. Its 154 submissions were inputs to collective drafting. Percentages and rankings describe those submissions; they should not be generalized into compliance rates for all digital-governance processes. The Final Report's transparency about method makes this limit easier, not harder, to respect.
NETmundial+10 therefore recovered entity memory and design knowledge. It did not recover a complete implementation ledger. That is precisely why its proposed caretaker role matters.
The Sao Paulo Guidelines convert values into observable steps
The strongest substantive advance in 2024 was the move from ten high-level principles to thirteen guidelines and twelve recommended process steps. The Sao Paulo Multistakeholder Guidelines ask processes to address power asymmetries, support informed deliberation, treat stakeholders fairly, disclose interests, respect linguistic diversity, share responsibility, adapt to change, resolve conflict, consider long-term effects, build capacity, coordinate across forums and pursue practical outcomes.
The process steps are even more assessable. A process should define its issue, identify affected stakeholders, engage them continuously, share information, secure equitable participation, facilitate dialogue, prepare drafts, explain how wider feedback was considered, make decisions openly, allow the wider community to react to inconsistent outcomes, implement decisions with accountability, and monitor and adapt.
This sequence supplies many of the entities a reviewer needs. It is possible to inspect whether a scope was published, whether affected groups were identified, whether drafts changed, whether reasons accompanied rejected input, whether remedial powers existed and whether monitoring occurred. The guidelines reduce the temptation to call any diverse meeting multistakeholder merely because several sectors appeared.
Their construction also shows that NETmundial+10 understood the continuity problem. The statement says multilateral processes should publish timelines, report progress or lack of progress regularly, document contributions and divergent views, set steps and deadlines for implementation, and reflect on impact and status. The final process step is explicitly to monitor and adapt.
Yet a principle can demand monitoring without itself being monitored. The guidelines tell adopting processes what to do; they do not automatically create an external institution that determines whether the process did it. "Community powers" are valuable only after the relevant community, standing, procedure and remedy are defined. Accountability for commitments requires a register of commitments and an actor responsible for following them after the meeting ends.
The guidelines therefore make a future audit possible. They do not by publication alone make it happen.
A caretaker was proposed, but custody is not yet accountability
NETmundial+10 came closer than the 2014 meeting to naming an owner. It recommended that the IGF act as depositary or caretaker of the Sao Paulo Guidelines and invited the Forum to consider their implementation in its own work, further discussion and evolution. Possible future subjects included prioritizing or clustering the guidelines, developing metrics, constructing systems for assessment and accountability, and producing explanatory tools.
The verbs are exact. NETmundial+10 "recommend[s]" a caretaker and "look[s] forward" to IGF consideration. Metrics and accountability systems "may" be discussed. The statement did not record an IGF decision accepting the function, amend the Forum's mandate, appropriate resources, specify a reporting date or designate an office to answer inquiries.
Custody is also narrower than accountability. A depositary can preserve the authoritative text and its versions. A caretaker can convene discussion and coordinate improvements. An auditor needs access to evidence and a method. An appeals body needs jurisdiction, standing and remedial power. A regulator needs legal authority. These functions should not be merged by implication.
The IGF is particularly suitable for some of them. It can keep an open repository, host comparative reviews, connect national and regional initiatives, invite evidence from institutions, and expose disagreement without pretending to issue binding judgments. Its annual and intersessional work can shorten the ten-year memory cycle.
Other functions may be inconsistent with the Forum's design. The 2005 Tunis Agenda made the IGF non-binding, denied it an oversight role and kept it out of day-to-day and technical operations. A caretaker cannot compel an unrelated government, company or technical body to produce records unless that institution has separately accepted the obligation. Nor should the IGF acquire covert supervisory power through an anniversary recommendation.
The correct response is not to abandon the caretaker proposal. It is to define the function narrowly enough to fit the Forum and require each adopting institution to supply its own enforceable accountability.
The original ten principles fail different memory tests
A useful review should not collapse the ten principles into one score. Each creates a distinct evidence problem:
| 2014 principle | Evidence a memory function would preserve | Unresolved 2024 question |
|---|---|---|
| Multistakeholder | Stakeholder map, decision stages, actual influence, role justification | Which processes were examined, and who was excluded from which stage? |
| Open, participative, consensus driven | Notice, access, draft history, objections, declared decision rule | What counted as consensus, and where did unresolved objection remain? |
| Transparent | Procedures, agendas, records, reasons, affiliations, decision trace | Was information merely published, or usable in time to affect the result? |
| Accountable | Responsible actor, independent check, review, redress, remedy completion | Who can challenge failure, and what happens after a finding? |
| Inclusive and equitable | Participation by region and affected group, support, language and disability access | Did formal openness overcome cost, time, language and power barriers? |
| Distributed | Function and authority map across organizations and levels | Did distribution improve resilience, or diffuse responsibility beyond recovery? |
| Collaborative | Liaisons, joint work, conflict handling, non-duplication and hand-offs | Which collaboration produced an outcome rather than another meeting? |
| Enabling meaningful participation | Capacity support, timing, response to input, influence over drafts | Could newcomers and underrepresented groups alter the decision? |
| Access and low barriers | Affordability, connectivity, public access and discriminatory barriers | Which institution owns the access outcome, rather than only discussion? |
| Agility | Review triggers, revision time, emergency process, sunset and adaptation | Was speed compatible with notice, evidence and correction? |
This table is not a retrospective grade. The available NETmundial materials do not justify a global compliance percentage. It is a map of evidence that should have been accumulated continuously.
The exercise reveals why principle-level affirmation is limited public evidence. Transparency may improve while equitable participation deteriorates. A process can be agile because it excludes deliberation. Distribution can preserve plural authority or conceal the absence of a responsible party. Collaboration can broaden knowledge while no institution commits to execution.
Institutional memory must retain these trade-offs. Otherwise every anniversary selects visible improvements and forgets the costs that appeared elsewhere.
A baseline must identify the process, not the brand
No global scorecard can be credible without a defined unit of analysis. "Internet governance" contains intergovernmental negotiations, domestic rule-making, technical standards, network-resource policy, company governance, civil-society coalitions and discussion forums. Applying one participation model to all of them would be both inaccurate and contrary to the 2014 principle that roles vary with the issue.
The unit should be a decision process with a named institutional home, scope, authority, start and end point. A reviewer can then ask who set the agenda, who drafted text, who could decide, who had to implement, and who could seek correction. A conference series is too broad if its programme, funding and outputs follow different rules. A corporation is too broad if board selection, policy development and operational service each have separate procedures.
The baseline should record what existed when the institution first claimed alignment with NETmundial. Later reviews can compare changes. If no baseline was taken in 2014, the first review should say so and use the earliest recoverable evidence rather than inventing precision.
This discipline prevents branding from becoming proof. An organization may describe itself as multistakeholder while limiting decisive stages to one constituency. A state consultation may be legitimately government-led but still provide meaningful, reasoned participation. The question is not whether the label appears. It is whether the institution's actual authority model satisfies the relevant guideline and explains deviations.
The 2024 statement recognizes this problem when it says the guidelines can assess processes that are multistakeholder only in name. A caretaker should turn that sentence into a repeatable test. It should publish both the common criteria and the contextual reason why a criterion applies differently to a legislature, a standards body or a voluntary coalition.
Evidence should be attached to claims, including negative claims
Institutional reports often list meetings, submissions and entity counts because those facts are available. They say less about rejected proposals, unanswered dissent, missed deadlines, abandoned remedies or communities that could not enter. The result is activity reporting rather than accountability.
A memory function should require claim-level evidence. If an institution says participation became more equitable, it should identify the affected groups, the prior imbalance, the intervention and the observed change. If it says consensus was reached, it should publish the decision rule and material objections. If it says an appeal worked, it should show whether the remedy arrived before the contested action became irreversible.
Negative evidence deserves a first-class place. A cancelled translation, unfunded travel programme, unavailable record, unfilled review office or non-response from an adopting institution is not an empty cell to be removed. It is evidence about the implementation environment. So is the inability to determine who was responsible.
The Final Report of NETmundial+10 is useful here because it does not hide the consultation machinery. It explains coding, rankings and the route from inputs to a draft. A future review should extend that transparency from drafting to disposition: which material proposal entered the text, which was combined, which was rejected and why.
That requirement should not turn editors into stenographers. Consensus documents need synthesis. But synthesis becomes accountable when a entity can trace a consequential concern to a reasoned outcome. Without that trace, preservation of every submission can coexist with disappearance of its substance.
The evidence record should remain accessible after links change and organizers leave. A durable archive needs stable identifiers, dates, versions and custody. Institutional memory cannot rest on personal inboxes or the continuing operation of one event website.
Correction is different from updating the next edition
NETmundial+10 calls the Sao Paulo Guidelines a living document. That is sensible. Multistakeholder methods should evolve as remote participation, language technology, funding models, geopolitical constraints and forms of decision-making change. A living document, however, needs rules for changing memory rather than replacing it.
At least three forms of change should be separated. A factual correction fixes an error in the record while preserving what was originally published. An interpretive note clarifies application without rewriting the adopted text. A revision changes the guidance through a declared process and produces a new version.
If those categories are merged, a caretaker can silently alter the baseline against which earlier conduct was assessed. Entities cannot know whether an institution complied with the rule in force at the time or with a later preference. Critics may discover that the wording they challenged has disappeared without a decision record.
The correction channel should be open to anyone who can identify a specific error or omission. The caretaker should acknowledge the request, publish non-sensitive evidence, assign a reviewer without a disabling conflict, decide within a stated period and preserve the original, request and disposition. Material disputes over interpretation should be recorded rather than disguised as typographical fixes.
Revision requires broader participation. The process should state who can propose changes, how drafts are published, how underrepresented communities are supported, how consensus is assessed, and who confirms the final version. A revision should include a change log explaining not only what changed but what implementation evidence made the change necessary.
That is how institutional memory becomes learning. Without a correction path, "living" can mean that every generation restates the principle in its preferred language and leaves no accountable history of why.
The IGF can host memory without becoming the world's inspector
A legitimate design would allocate functions among the IGF, adopting institutions and independent reviewers rather than placing all authority in one forum.
The IGF could maintain the canonical text, version history, assessment method, public register of adopters and repository of reports. It could convene an annual evidence session, invite challenges and publish a synthesis that identifies gaps. Its national and regional initiatives could contribute contextual evidence without being treated as branch offices subject to global command.
Each adopting institution would identify the decision processes covered, name an accountable officer or body, publish a baseline, provide evidence on the applicable guidelines, disclose exceptions, and state remedial commitments. Voluntary adoption would become specific enough to evaluate. Institutions with legal or membership duties could embed these requirements in their own enforceable rules.
Independent reviewers could examine a rotating sample, test self-reports and publish methods and conflicts. Independence need not mean a permanent global tribunal. Universities, audit bodies, civil-society researchers and peer institutions could conduct reviews under common evidence standards, provided funding and selection are disclosed.
The IGF synthesis should not purport to invalidate a law, reverse a technical decision or sanction a company. It should say what the evidence supports, what remains disputed and which competent institution owns correction. Public comparison can create pressure while preserving authority boundaries.
This model also protects the IGF. If it becomes the sole judge of whether every institution is legitimately multistakeholder, disputes over its own representation, funding and mandate will overwhelm the repository function. A distributed evidence system with a common ledger fits the ecosystem better than centralized certification.
The caretaker's authority would be epistemic and procedural: preserve, compare, question and report. Remedial authority would remain where the relevant law, contract, membership rule or institutional mandate places it.
Funding is part of the accountability design
NETmundial+10 acknowledged that a strengthened IGF would require new financial and human resources. Its consultation record also reported concern that the Forum lacked sufficient resources to perform its mission. This is not a secondary administrative detail.
An unfunded caretaker will preserve the text and defer the difficult work. Comparative assessment requires researchers, archives, translation, accessible participation, data maintenance, conflict review and follow-up with non-reporting institutions. If the function depends on a short grant, memory will again disappear when attention moves elsewhere.
Funding can also distort the record. A donor may not dictate findings but may influence which institutions or regions receive study, which evidence can be collected, and whether criticism of the donor's sector is sustained. The memory function should disclose cash and in-kind support, conditions, selection roles and conflicts. Core preservation and reporting should have predictable support not contingent on favorable outcomes.
Participation funding is equally important. A global review based on institutions able to send staff to annual meetings will reproduce the asymmetry it measures. Remote access lowers some costs but not time zones, language, bandwidth, expertise or the opportunity cost of sustained review. Regional evidence partners need support to collect and challenge information locally.
The budget should be tied to a work programme: repository maintenance, number of assessments, translation coverage, correction deadlines, outreach and independent review. Annual reporting should show what was not done for lack of resources. That negative disclosure prevents institutional ambition from being mistaken for delivery.
A recommendation without a budget can still guide voluntary action. It cannot support a claim that a global memory function now exists.
Public-sector continuity makes delayed review costly
The memory problem has practical consequences for public institutions. Governments, regulators, schools, hospitals and public networks participate in or depend on decisions made across technical and policy forums. They often have long procurement cycles, statutory consultation duties, language obligations and limited specialist staff.
When a principle is restated without implementation evidence, a public body cannot easily tell which process offers meaningful participation, which consultation will explain its use of input, or where a remedy can be pursued. Staff repeatedly reconstruct institutional histories. Smaller administrations depend on claims made by better-resourced entities. The cost falls most heavily on those whom capacity-building language intends to include.
Continuity also matters when representatives change. A ministry may support a statement in one year and send different officials to the next review. Civil-society organizations lose grants; company teams reorganize; technical volunteers rotate. A public record should allow successors to understand commitments without relying on personal attendance.
This is why regular reporting must include lack of progress, as the 2024 statement itself says. A missed milestone is actionable only if it remains visible to the next office holder. If every meeting begins with a fresh consultation about whether the old principle is still relevant, institutional turnover consumes the time that should be spent correcting implementation.
Public-sector continuity does not justify a rigid global checklist. National legal duties and resources differ. It does justify a stable evidence format that allows institutions to explain context and allows affected communities to see whether the explanation changed.
The best capacity-building tool may be a reliable institutional memory: a record showing where to intervene, what evidence matters, who must answer and when the next decision occurs.
A practical implementation ledger can remain non-binding
The minimum viable memory system would contain nine linked records.
First, a principle record would preserve the 2014 text, the 2024 guidelines, interpretations, versions and change history. Second, a process record would identify the institution, authority, decision type, timetable and affected community. Third, an adoption record would show the dated act by which the institution accepted specific guidelines.
Fourth, an evidence record would attach documents and observations to each applicable guideline. Fifth, a dissent record would preserve material disagreement, the response and any unresolved question. Sixth, a commitment record would name the actor, action, deadline and expected evidence of completion.
Seventh, a review record would disclose the reviewer, method, funding, conflicts, findings and uncertainty. Eighth, a correction record would preserve challenges, factual amendments, remedies and implementation status. Ninth, a continuity record would identify the next review date, custodian and succession arrangement.
None of this requires the IGF to bind an external institution. A body can join voluntarily and be assessed against the claims it chooses to make. The public ledger can distinguish "endorsed," "scope declared," "self-report filed," "independently reviewed," "remedy committed" and "remedy completed." It should never compress those states into a seal of approval.
Non-participation should also be recorded carefully. An institution that never endorsed NETmundial has not breached a voluntary pledge merely by declining the ledger. But if it publicly claims adherence, the caretaker can note that no evidence was supplied. If a government or membership body has separately made the principles mandatory, enforcement belongs to that authority.
The ledger's value is comparability and recall. It makes overstatement harder, identifies where evidence is thin, and gives future meetings something more rigorous than collective memory to review.
Measurement must not reward the appearance of inclusion
Metrics can improve accountability and also create new theatre. Counting sectors, countries, speakers, submissions or translated pages is easy. Those measures matter, but they do not establish that affected people influenced a decision.
A balanced review would examine access, influence, decision quality and remedy. Access asks who received timely notice, materials, language support and resources. Influence asks whose proposals changed scope, drafts or outcomes and whether rejected input received reasons. Decision quality asks whether evidence, trade-offs and dissent were visible. Remedy asks whether an error or procedural failure could be corrected in time.
The review should also examine power. A nominally equal speaking slot can coexist with unequal research staff, lobbying access, travel budgets, legal authority and control over implementation. NETmundial+10 rightly calls for realistic analysis of asymmetry. That analysis should identify the mechanism by which power affected the result, not assign moral status to stakeholder categories in the abstract.
Outcome measures need caution. A technically sound or popular policy is not proof that the process was inclusive. An inclusive process can produce a contested or unsuccessful result. Process legitimacy and substantive performance overlap, but one should not be used to erase the other.
Qualitative findings are therefore unavoidable. The solution is not to replace judgment with a composite score. It is to publish the evidence, criteria, reasoning and uncertainty behind the judgment. Metrics should direct attention, not manufacture precision.
The caretaker should also test its own process under the same guidelines. Who selected reviewers? How did wider feedback alter the method? Can an assessed institution challenge a factual error without suppressing criticism? Reflexive assessment would give the guidelines credibility that another declaration cannot supply.
Later reforms should not be credited to a principle without a causal record
The decade after 2014 contained real institutional change. The IGF developed more intersessional work and received a further ten-year mandate through the 2015 WSIS review. The IANA stewardship transition produced new contractual and community-accountability arrangements. Remote participation became routine in many forums. Governments and organizations opened additional consultations to non-state contributors.
These developments may be consistent with NETmundial. Consistency is not proof that NETmundial caused them. Each reform had its own mandate, entities, negotiation, evidence and political conditions. The IANA transition, for example, began with a United States government announcement, proceeded through function-specific community proposals and required concrete institutional instruments. The 2015 General Assembly resolution extended the IGF under the Tunis mandate. Neither result should be entered in a NETmundial success column solely because the 2014 statement favored accountability or a stronger Forum.
A causal record would identify when an authorized decision maker cited a NETmundial principle, what proposal the citation supported, whether opponents addressed it, and how the final act differed. Even then, the principle may be one influence among several. The ledger should distinguish direct implementation, express influence, compatible independent reform and unconnected change.
The same rule applies to failure. A repressive law, inaccessible meeting or opaque corporate decision does not prove that NETmundial made conditions worse. It may demonstrate continuing relevance and non-implementation, but attribution requires evidence. A credible memory function should not inflate either achievements or disappointments to make its founding document appear more consequential.
This restraint strengthens the principles. Institutions are more likely to report honestly if review is not a search for ceremonial victories. It also helps future researchers understand how soft norms travel. Some enter formal rules; some alter expectations; some supply vocabulary after a decision has already been made; some are ignored. Those pathways are different forms of influence and should remain visible.
An anniversary that claims every favorable change invites skepticism. An anniversary that traces only what the evidence can support creates a record worth inheriting.
Unfinished commitments need expiry, renewal and escalation states
The 2014 roadmap contained requests with different degrees of specificity. Organizations should publish periodic progress reports. IGF improvement recommendations were suggested for implementation by the end of 2015. Stable and predictable funding was described as essential. Benchmarking systems required further discussion. Existing entities were invited to report at major meetings.
A memory system should not leave all of these in a permanent category called "ongoing." Each commitment needs a state. It may be proposed, accepted, funded, started, partially completed, completed, superseded, refused, expired or unverifiable. The state should name the actor that made the determination and link the evidence.
Deadlines require a default consequence. Missing a non-binding deadline need not trigger a sanction. It should trigger a visible status change and a request for explanation. The responsible institution can renew the commitment, narrow it, replace it or state that it will not proceed. Silence should not be rendered as progress merely because discussion continues.
Escalation also has to be bounded. If a self-report is late, the caretaker can mark it missing and invite independent evidence. If a factual dispute persists, it can publish both positions and the material available. If an adopting institution breaches its own enforceable rule, the matter should be referred to the appeal, membership, contractual or legal mechanism that has authority. The caretaker should not invent a remedy it cannot deliver.
Supersession is especially important over a decade. A 2014 recommendation may be overtaken by a later mandate or a different institutional design. The record should not call that failure automatically. It should show the later act, explain which objective survived and identify any residual gap. Conversely, a new initiative should not erase an older unmet commitment merely by using updated vocabulary.
Clear states make delayed work politically legible. They prevent organizations from renewing the prestige of a principle while keeping every practical obligation indefinitely unfinished. They also give future entities a precise starting point: not whether the community still cares, but which commitments remain open, who owns them and what evidence would close them.
The next review should begin now, not at NETmundial+20
The 2024 statement already contains the elements of a work programme. It identifies the IGF as a possible caretaker, treats the guidelines as living, contemplates metrics and accountability systems, calls for regular reporting and emphasizes implementation, monitoring and adaptation. The missing act is to convert those elements into dated responsibilities.
Within the first cycle, the caretaker function should publish an acceptance statement defining scope and limits. It should preserve the canonical 2014 and 2024 texts, issue a draft evidence framework, and invite a small, diverse group of institutions to declare processes for pilot review. The pilot should include different authority models rather than only organizations already confident of a high score.
A second cycle should publish findings, objections and method changes. It should identify evidence that could not be obtained, costs of participation and any guideline that proved too vague. Institutions should state remedial commitments with dates.
Annual updates can then track completion and corrections. A fuller review every three or four years would allow comparison without forcing every process into the rhythm of one meeting. The next decennial gathering would assess accumulated evidence and revise guidance where necessary; it would not begin by rediscovering the absence of a baseline.
The schedule should survive cancellation of an annual event. Repository custody, report deadlines and correction channels need continuity independent of a host country or conference programme. If the IGF cannot accept or fund the function, the record should say so and another coalition can propose a bounded alternative rather than assuming silent adoption.
Institutional memory begins with an answer to a simple question: who is late if the report does not appear? NETmundial+10 did not supply that answer. Its own guidelines explain why it now should.
Principles need a responsible memory, not a permanent sovereign
The case for ownership can be misunderstood as a demand for one global authority to enforce every principle. That would contradict the distributed character of Internet governance and exceed the mandate of a non-binding forum. It could also let a central assessor impose one participation model on institutions with different legitimate functions.
The needed ownership is narrower. Someone must own the integrity and continuity of the evidence. Each institution must own correction within its authority. Independent reviewers must own their methods and conflicts. Entities must be able to see where one responsibility ends and another begins.
This arrangement preserves pluralism while preventing diffusion from becoming evasion. A standards body, regulator, company and forum can each apply the guidelines differently, but each should identify the accountable process, evidence and remedy. The caretaker can compare the explanations without becoming their superior.
That is also the protection against consensus capture. A later institution should not cite NETmundial as proof that its preferred governance model was globally ratified. It should show which principle it adopted, how the adoption changed conduct, what objections remain and which authority can correct failure. The public record should preserve minority and non-entity perspectives rather than converting attendance into universal consent.
NETmundial+10 deserves credit for refusing to pretend the 2014 principles were complete merely because they survived. It admitted incomplete implementation, supplied more concrete steps and identified a plausible home for future care. The decisive work starts where the anniversary statement ends.
If metrics remain a possible discussion, custody an unaccepted recommendation, resources an aspiration and correction an abstract guideline, the next decade will produce another eloquent diagnosis. If those items acquire owners, dates, evidence and remedies, the 2024 meeting can become the moment NETmundial developed an institutional memory rather than merely remembering itself.