ICANN Accountability Metrics Should Count Corrections, Not

Summary

ICANN already publishes a large public record: bylaws, board materials, annual reports, public-comment files, accountability-mechanism pages, review pages and participation channels. That record proves that activity is visible. It does not, by itself, prove that mistakes are corrected.
The decisive accountability measures should be outcome measures: corrections made after challenge, reversals or withdrawals, time to remedy, cost shifted away from the harmed party, repeated failure by the same function, and follow-through after review findings.
Counting meetings, documents, public comments and attendance can reward institutional motion even when the affected party receives no usable remedy. A dashboard can look healthy while the underlying decision remains wrong, late or unrepaired.
A stronger ICANN accountability scorecard would keep the existing transparency record but add a remedy ledger: each claim of error, the decision point affected, the review path used, the elapsed time, the final institutional change, and the evidence that the change reached the people who bore the cost.

The wrong denominator

The easiest thing to count in a multistakeholder institution is activity. Meetings have dates. Agendas have links. Public-comment windows open and close. Reports move from draft to final. Webinars draw attendance. Mailing lists produce messages. Board minutes are posted. Review teams convene. Staff prepares updates. The result can be a handsome accountability narrative in which the institution appears responsive because it is always publishing, consulting and gathering.

That is a useful transparency record, but it is the wrong denominator for accountability. Accountability is not the number of opportunities to speak. It is the relationship between an institutional error and the correction of that error. If a decision was made on a false premise, did the record change? If a public-comment summary compressed a material objection, was the objection restored to the decision file? If a board rationale omitted a material factor, was the rationale corrected? If a harmed party won a review, did the remedy arrive soon enough to matter?

If the same defect appeared in later decisions, was the failure treated as repeated institutional learning failure or as another isolated case?

The question matters especially for ICANN because the institution claims legitimacy through open, transparent and bottom-up participation rather than through ordinary electoral control. Its bylaws require detailed explanations of the basis for decisions, including how comments influenced policy considerations, and public disclosure of rationales for board and community decisions. They also describe accountability mechanisms such as reconsideration, independent review, the Ombudsman, specific reviews and Empowered Community powers. That architecture is substantial.

Yet the architecture can still produce an accountability illusion if its public measures emphasize how many channels exist rather than whether those channels change outcomes.

A meeting count can answer one question: was there a chance to attend? A document count can answer another: was there something to read? A participation count can answer another: did people show up? None of those counts answers whether the institution corrected itself when challenged. A decision can be preceded by open comment, discussed in public meetings, supported by a thick report and still be wrong. A remedy can be available in theory and still be too slow, too expensive or too narrow to restore the position of the affected party.

A review recommendation can be accepted in principle and still never alter the recurring behavior that produced the problem.

The better denominator is the universe of plausible errors. That universe is not easy to count, because it includes mistakes that were never challenged, challenges abandoned because of cost, objections buried in consultation records, complaints resolved informally, and repeat defects that affected different parties at different times. But difficulty is not an excuse to count only what is administratively convenient. If ICANN reports accountability mainly through meetings held, comments received and reports issued, it measures the size of the stage rather than the effectiveness of the remedy.

Transparency is not the same as correction

ICANN deserves credit for maintaining a large public record. Its bylaws require a public website with information about meetings, policy matters, accountability mechanisms, budgets, audits, comments and public forums. Its annual report page links years of reports. Its reconsideration page lists requests and status information. Its accountability mechanisms page explains that reconsideration, independent review and the Ombudsman are intended to provide review of ICANN action.

Its specific-review provisions require periodic assessment of accountability and transparency, including public input, decision explanations, acceptance by the Internet community, policy development and the Independent Review Process.

That record is not trivial. It gives journalists, operators, governments, lawyers, civil-society groups and ordinary Internet users a way to reconstruct at least part of the institutional file. Many public bodies and private platforms disclose less. The problem is not absence of paper. The problem is that paper can become a substitute for remedy.

Transparency answers the first accountability question: can the public see what happened? Correction answers the second: did the institution change what happened after error was shown? The two are related but not identical. A fully visible error is still an error. A board rationale can be posted quickly and still omit the reason that later proves decisive. A public-comment archive can contain thousands of submissions and still fail to show whether a minority warning was treated as evidence. A reconsideration request can be listed and denied with reasons and still reveal that the remedy channel is too narrow for the harm alleged.

The distinction is familiar outside ICANN. A court does not prove fairness merely by publishing hearing dates; it proves review capacity by issuing orders, reversing decisions, remanding records and enforcing deadlines. A regulator does not prove consumer protection merely by counting complaint intake; it proves effectiveness by showing restitution, corrected practices, penalties, closed cases, recurrence rates and time to resolution. A company does not prove privacy accountability merely by publishing a policy; it proves it by showing breach correction, deletion completion, appeal outcomes and repeat-incident controls.

Internet governance should not use a weaker standard just because its institutions are unusual.

ICANN's public record already contains the ingredients for stronger measurement. Reconsideration requests have requesters, challenged actions, timing and outcomes. Independent review matters have claims, panels, declarations, costs and implementation consequences. Board resolutions have rationales and votes. Public-comment records have submissions, summaries and decisions. Reviews have recommendations, board action, implementation status and later assessment of intended effect. The missing step is to tie these records into an outcome ledger that asks what changed.

That ledger should not punish ICANN for every challenge. Many complaints will be weak. Some will misunderstand the mission. Some will ask for remedies outside ICANN's authority. Some will be tactical attempts to reopen a settled bargain. A serious correction metric can preserve those distinctions. It can count denied claims, out-of-scope claims and withdrawn claims without pretending that every rejection is a failure. What it cannot do is treat the existence of the channel as success regardless of whether the channel ever produces a timely, usable correction.

Meetings measure access, not accountability

Meetings are necessary in a multistakeholder environment. ICANN cannot coordinate global identifier policy through private memos alone. Face-to-face meetings, remote sessions, community forums and public board sessions create shared information and make it harder for decisions to disappear into private negotiation. They also impose costs: travel, time zones, language burden, policy-staff advantage, agenda endurance and the ability to follow several topics at once.

Counting meetings therefore says little unless the count is connected to decision influence. A hundred sessions can still leave a small operator unable to alter a harmful decision if the important text was drafted before the meeting, if the objection required legal expertise, if the final summary treated dissent as background, or if the board had no realistic option left when the comment closed. A meeting count can even reward a bad design. The more complicated the decision path becomes, the more meetings it generates, and the more the institution can claim engagement while entities bear rising participation costs.

A remedy-based metric would ask different questions. How many material objections raised in meetings were reflected in later text? How many were rejected with specific reasons? How many were marked as outside ICANN's mission, and was that mission finding later tested? How many issues recurred across several meetings before any action? How many late-stage meeting concerns showed that earlier consultation had failed? How many commitments made in a public forum were later tracked to completion? How often did staff or board materials change after a meeting, and which decision points moved?

This does not require a mystical measure of influence. It requires a discipline of traceable changes. If a meeting produces no change, the record can say so. If a meeting produces a narrower clause, a delayed effective date, a public-interest analysis, a request for further evidence or a referral to another forum, the record can identify that result. If a concern is rejected, the record can give the reason. The count then becomes a count of decision effects, not a count of chairs filled or microphones used.

Participation metrics also need denominator discipline. "More entities" sounds healthy until the denominator is named. More compared with what? The global Internet public? Affected contracted parties? Number-resource holders? Civil-society groups? End users? Governments? Technical operators? Volunteers who attended the previous proceeding? A metric that reports attendance without class, region, subject-matter stake and repeat-player concentration can hide capture. A small number of well-resourced actors may dominate a proceeding while the aggregate participation line rises.

A remedy metric should identify not only who spoke but whether the affected class obtained correction when its evidence was strong.

Meetings should remain visible. They should not be treated as the proof of accountability. The proof comes when meeting input can be followed into changed text, corrected reasoning, delayed action, compensation, reversal, remand or a public explanation of rejection.

Documents can bury decisions as well as reveal them

Documents are another tempting accountability count. ICANN produces consultation papers, public-comment summaries, board rationales, annual reports, review documents, implementation updates, correspondence, presentations and briefing materials. A public record that thick can look incontestable. It can also become hard to audit.

Document volume creates two risks. First, it can bury the decisive choice. The public may see a proposal, a comment summary, a staff report, a board rationale and later implementation materials, but the exact point where an objection was accepted or rejected remains unclear. Second, document volume can dilute responsibility. When every stage has a document, no single document appears to own the decision. Staff summarized. A committee reviewed. The board considered. A later team implemented. If the outcome is defective, each file can point to another.

Correction metrics cut through document inflation. They require each consequential document to answer a limited set of questions. What claim of error or risk was raised? Where was it considered? Who had authority to correct it? What changed? What did not change? What reason was given? What deadline was attached? What evidence later showed completion?

This approach is particularly important for public-comment summaries. A summary can comply with publication expectations while compressing disagreement into themes chosen by the institution. The right metric is not simply whether the summary was published on time. It is whether the summary preserved material objections, separated duplicate support from distinct evidence, identified minority arguments, recorded rejection reasons and tied the comments to the final decision. If public comment is part of legitimacy, the measure should be "comments converted into accountable reasons," not "comments received."

Board rationales deserve the same treatment. ICANN's bylaws point toward detailed explanations and rationales. A rationale metric should not only count whether a rationale exists. It should count later correction: how often rationales were amended, supplemented, contradicted by review findings, followed by remand, or cited as limited public evidence by a review body. A rationale that survives challenge is not automatically correct, but a rationale repeatedly found incomplete is an institutional signal. The dashboard should capture that signal.

Even annual reporting needs a remedy lens. Annual reports are valuable for finance, operations and institutional memory. But an annual accountability page that lists projects, engagement, reports and strategy can still avoid the negative ledger: decisions corrected, delay paid for by others, matters withdrawn after challenge, review findings left open, repeat defects and cost barriers. The missing negative ledger is exactly what affected parties need. Accountability reporting should make institutional discomfort visible because discomfort is where learning happens.

Corrections are the primary unit

The first serious accountability metric should be corrections. A correction is not merely a changed sentence. It is a demonstrable institutional act that repairs an error in record, reasoning, decision, timing, implementation or affected-party position. Corrections can be small or large. They can involve a public correction to a factual statement, a revised comment summary, a changed board rationale, a reopened consultation, a delayed effective date, a revised contract clause, a remanded decision, a corrected eligibility determination, or a public admission that an earlier path was wrong.

Corrections should be counted by type. Factual corrections are the simplest: the record said something false or incomplete, and the public file was amended. Reasoning corrections are more important: the institution accepted that a decision rationale omitted or misread a material factor. Procedural corrections occur when the institution reopens, extends, remands or repeats a step because the earlier step was defective. Substantive corrections occur when the outcome changes. Remedial corrections occur when an affected party receives money, fee relief, deadline relief, restored status, application reprocessing or another practical repair.

The dashboard should then separate voluntary corrections from compelled corrections. A voluntary correction after staff review shows internal responsiveness. A correction after reconsideration shows the review channel had effect. A correction after independent review shows externalized review pressure. A correction after litigation or settlement shows that the internal system may have failed earlier. These distinctions matter because the cost and delay borne by the affected party differ.

Corrections also need severity bands. A typographical fix in a board rationale is not the same as reversal of a decision after an independent review. A delayed report is not the same as a rejected application being revived. A public-interest analysis added before final action is not the same as a corrected record after irreversible harm. Severity bands should not be used to hide small errors; small errors matter when repeated. But the public should know whether the correction changed the outcome, the reasoning, the timing, the record or only the presentation.

Finally, corrections need a "who benefited" field. A correction that improves future text may not repair the person who suffered under the prior text. A correction that helps a contracted party may not help registrants. A correction that helps a supporting organization may not help end users. A correction that improves a number-related record may not affect DNS actors. If the stated accountability system is for the Internet community, its metrics must show whose position was restored and whose was merely acknowledged.

Reversals and withdrawals are not embarrassments

Institutions often dislike counting reversals because reversals look like failure. That instinct is backward. A system that never reverses itself may be perfect, but it is more likely defensive. If ICANN receives substantial challenges over many years and almost never reports a changed outcome, the public should not automatically infer flawless decisions. It should ask whether the remedy channels are too narrow, too costly, too slow, too deferential or too reluctant to disturb institutional choices.

Reversals should be separated from withdrawals. A reversal occurs when ICANN or a review body changes the result or requires a new result. A withdrawal occurs when a requester abandons a challenge, a proceeding becomes moot, or ICANN changes course before a formal decision. Withdrawals can mean many things. They may show settlement, cost exhaustion, mootness, procedural narrowing, tactical retreat or successful informal correction. Treating all withdrawals as neutral hides important accountability information.

Each reversal or withdrawal should identify the reason category. Did the challenger prevail on a false-information claim? Did ICANN accept that material information was not considered? Did a board action exceed the mission? Did a deadline make the claim moot? Did ICANN voluntarily change the challenged action? Did the requester withdraw after receiving private assurance? Did the requester withdraw because the remedy would arrive too late? Some of those answers require confidentiality in individual cases, but aggregate categories can still be public.

Reversal metrics are also valuable because they identify learning points. If reversals cluster around public-comment summaries, the summary method is defective. If they cluster around staff implementation, the policy-to-implementation handoff is weak. If they cluster around new gTLD evaluation, the evaluation rules may be unstable. If they cluster around board rationales, the board file may be too thin. If withdrawals cluster after long delay, the review channel may be functionally exhausting challengers rather than resolving claims.

Reversals should not be weaponized against every staff member or director involved. Fear of blame makes institutions hide error. The goal is not a personnel scoreboard. It is an institutional-control scoreboard. Reversals reveal where authority met reality and had to adjust. That is exactly what accountability is supposed to show.

Delay is a remedy metric

Delay is often treated as an administrative inconvenience. In accountability systems, delay is a remedy metric. A correction that arrives after the commercial opportunity, application window, public-comment decision, contract term, election, budget cycle or delegation moment has passed may be formally correct and practically empty. For affected parties, time is part of the remedy.

ICANN's own accountability architecture recognizes time in several places. Reconsideration requests have filing windows. Board materials have posting expectations. Public-comment periods have minimum periods and decision sequences. Independent review involves procedures, panels and sometimes preliminary relief. The public record can show dates. What is missing is a consistent public measure of time-to-remedy across accountability channels.

A serious dashboard should publish at least five delay measures. First, time from challenged action to filing. That shows whether affected parties could understand and react quickly. Second, time from filing to admissibility or initial response. That shows whether the channel opens promptly. Third, time from filing to final decision. That shows whether review arrives in a useful window. Fourth, time from decision to implementation. A favorable declaration without implementation is not yet a remedy. Fifth, time from repeated warning to institutional correction.

If the same issue appears in public comments, ombuds reports, reconsideration requests and review findings before any change, the delay is systemic.

Delay measures should be tied to harm. A six-month delay in a long-term policy review may be tolerable. A six-month delay in an application, delegation, contract, eligibility or emergency decision may be decisive. The dashboard should therefore identify the decision type and practical deadline. It should not merely report average days. Averages hide outliers, and outliers are where accountability often fails. Median time, longest open matters, age bands and time-sensitive categories would be more informative.

Delay also interacts with cost. A party with substantial resources can survive a long review. A small applicant, volunteer group, civil-society organization or regional operator may not. If the only parties able to reach correction are those able to fund extended counsel and policy presence, the accountability channel is unequal. Time-to-remedy should therefore be reported with claimant type and cost indicators where available.

The standard should be simple: if the institution was wrong, the public should know how long the wrong position governed and who paid for that interval.

Compensation and cost shifting are accountability evidence

ICANN accountability discussions often focus on decisions and reasons, but compensation is part of accountability. Compensation does not always mean damages in a court sense. It can mean fee relief, cost reimbursement, renewed opportunity, restored priority, deadline extension, reprocessing at no additional charge, or bearing the cost of a review that the claimant should not have had to bring.

Why does compensation matter? Because a system that corrects future behavior but leaves the harmed party with the entire cost of proving the error teaches affected parties to stay silent. If challenging an error costs more than the benefit of correction, errors will be underreported. The dashboard will then show few successful challenges, not because the institution is accurate, but because correction is uneconomic.

Independent review can involve serious cost. Reconsideration may be less expensive but still requires expertise, time and a record. Public-comment advocacy also has cost. A meaningful metric should show when ICANN bore review costs, when costs were allocated to a claimant, when the claimant received practical relief, and when a favorable finding produced no material restoration. It should also show whether cost consequences were predictable. If a entity cannot estimate the cost and possible recovery before challenging an error, accountability becomes a gamble.

Compensation metrics also expose institutional incentives. If ICANN can make an error, resist correction, lose or partially lose, and still leave the affected party bearing most costs and delay, the incentive to correct early is weak. If early correction reduces cost, and late correction is visibly counted against the institution, the incentive changes. Accountability metrics should reward early repair and make late repair more visible.

There is a legitimate caution. ICANN is not an ordinary public compensation tribunal, and many disputes involve policy choices rather than private injury. Not every disappointed stakeholder deserves payment. But the absence of a universal damages model is not a reason to ignore cost. At minimum, ICANN can report cost allocation, fee waivers, reprocessing relief, implementation expenses and whether review costs were shifted when ICANN's position failed. The public can then distinguish principled denial from uneconomic accountability.

Repeated failure is more important than isolated error

Every institution makes mistakes. The accountability question is whether mistakes recur after warning. Repeated failure should therefore be one of the highest-weight indicators in an ICANN accountability scorecard.

Repeated failure can take several forms. The same office may miss the same type of deadline. Staff may rely on similar unsupported assumptions across proceedings. Board rationales may repeatedly omit how public comments changed, or failed to change, the decision. Public-comment summaries may repeatedly merge distinct objections into broad themes. Review recommendations may repeatedly be accepted but left without measurable effect. A remedy channel may repeatedly reject claims as outside scope even though the underlying harm remains unaddressed somewhere else. In each case, the second and third occurrence is more serious than the first.

The dashboard should therefore track recurrence by defect type, not by scandal label. "Incomplete rationale" is a defect type. "Material information not considered" is a defect type. "Late implementation after accepted recommendation" is a defect type. "Unclear authority to correct" is a defect type. "Out-of-scope dismissal with no alternate remedy" is a defect type. These categories let the public see whether ICANN is learning at the level of institutional control.

Repeated failure metrics also protect ICANN against unfair anecdote. Critics often rely on memorable cases. A recurrence dashboard can show whether a complaint is a one-off grievance or part of a pattern. If a defect appears once in a decade, the remedy may be case-specific. If it appears every year, the remedy should be structural. The public debate becomes less emotional and more empirical.

Recurrence should include positive recurrence too. If an early-correction practice works, count it. If a new board-rationale template reduces later challenges, count that. If a public-comment response table reduces claims that comments were ignored, count that. If the Ombudsman resolves a recurring fairness concern before formal dispute, count that. Accountability should not be a catalogue of shame. It should be a learning system. But learning cannot be shown unless repeated problems and repeated improvements are both visible.

Participation can become consent theater

The phrase "multistakeholder" carries moral force in Internet governance. It suggests that decisions are legitimate because affected communities can participate. That is partly true. It is also incomplete. Participation can become consent theater when the institution counts invitations, sessions and comments while the decision path remains practically closed.

Consent theater has several signs. The proposal is already negotiated before public comment. The comment window is open, but the deadline leaves little room for meaningful change. The summary identifies themes but not rejection reasons. The board rationale says comments were considered without showing how. The same commenters return year after year because only repeat players understand the path. Affected users appear only at the end, after the technical or contractual choice is nearly fixed. The final report thanks entities and proceeds.

None of those signs proves bad faith. Complex coordination often requires early drafting, expert negotiation and legal constraints. But they do prove that participation counts are limited public evidence. If the institution invites participation but cannot change the decision, it should say so. If only implementation details remain open, it should say so. If a policy choice is already settled by a supporting organization, it should say so. If the board has discretion but the counterparty is unlikely to reopen a term, it should say so. Honest limits are better than inflated participation metrics.

An accountability dashboard should therefore include "remaining discretion at participation stage." For each major consultation, what could still change? Policy objective, legal text, timing, implementation detail, monitoring, budget, or nothing material? How many comments addressed each open element? Which comments asked for changes outside the remaining discretion? How many objections were rejected because the decision was already constrained? This would make participation truthful.

The same principle applies to community meetings. A meeting before a decision has different accountability value from a meeting after the decision path is fixed. A listening session is not the same as a deliberative session. A briefing is not the same as a remedy hearing. A dashboard that reports all participation events together hides these differences.

Participation remains essential. But the metric should be influence under stated constraints, not attendance under a banner of openness.

The accountability channels need outcome tables

Reconsideration, independent review, the Ombudsman and specific reviews should each have outcome tables that use comparable categories. The categories do not need to flatten the legal differences among the channels. They need to let the public follow the remedy path.

For reconsideration, the table should identify the challenged action or inaction, the asserted standard, timeliness, preliminary handling, final outcome, whether any record was corrected, whether any decision was changed, whether the request exposed a broader defect, and time to completion. It should distinguish denial because the claim failed from denial because the channel could not reach the harm. That distinction is central. A claim outside reconsideration may still show an accountability gap.

For independent review, the table should identify claim type, mission or bylaw provisions at issue, preliminary relief, final declaration, cost allocation, implementation action, time from filing to declaration, time from declaration to implementation, and whether the same issue appeared later. The public does not need confidential legal strategy to see whether review changed institutional behavior.

For the Ombudsman, the table should identify complaint categories, recommendations, acceptance, follow-through and recurrence. Because the Ombudsman often works through fairness, facilitation and recommendation rather than binding order, follow-through is the key metric. A recommendation without response is not equivalent to a recommendation implemented.

For specific reviews, the table should identify recommendations, board action, intended effect, implementation status, evidence of effect and later reassessment. ICANN's bylaws already contemplate review of whether prior Accountability and Transparency Review recommendations have been implemented and whether implementation produced the intended effect. That is the seed of the right metric. The public dashboard should make intended effect the central unit, not only implementation status. A recommendation can be "implemented" in a narrow project sense without curing the recurring accountability problem that prompted it.

Comparable outcome tables would let readers ask whether different channels converge. If reconsideration rarely changes outcomes, independent review changes some outcomes but slowly, the Ombudsman identifies fairness concerns without order power, and specific reviews produce broad recommendations with uncertain effect, the public can see the shape of the accountability system. If early staff correction resolves many matters quickly, the public can see that too. Either way, the conversation moves from institutional self-description to remedy evidence.

Number-resource governance needs the same metric

ICANN's number-related role is narrower than its DNS contracting role, but it is not irrelevant. The bylaws describe ICANN's mission for Internet Protocol numbers and Autonomous System numbers as coordinating allocation and assignment at the top-most level, providing registration services and open access for global number registries as requested by the IETF and RIRs, and facilitating global number registry policies by the affected community and related tasks agreed with the RIRs.

After the IANA stewardship transition, number-related accountability often runs through agreements, service-level expectations, the RIR communities and the Number Resource Organization environment rather than through ICANN's domain-name revenue base.

That makes remedy metrics more important, not less. A narrow function can still fail in ways that matter: late delegation, unclear escalation, weak public reasoning, uncertain replacement authority, poor status reporting, or confusion about which body can correct which problem. If the institution counts coordination meetings and service reports but not correction, the number community receives activity evidence without remedy evidence.

A number-related accountability table should therefore identify top-level registry-service corrections, missed service commitments, escalations, repeated issues, public explanations, customer-impact windows and whether any correction required action by ICANN, PTI, the RIRs, the NRO or another body. The point is not to import DNS politics into number governance. It is to ensure that the narrower role has proportionate accountability.

This also protects institutional boundaries. If a number-related complaint is outside ICANN's authority, the dashboard should say where the corrective authority actually sits. "Outside scope" should not be the end of the public record. It should be a routing instruction: RIR community, NRO process, IETF request, contract escalation, PTI service channel, or another path. An accountability system that repeatedly tells number-resource actors that their claims are outside scope without identifying the effective remedy path is not accountable in practice.

The same principle applies in reverse. ICANN should not be blamed for every number-resource dispute merely because it sits in the IANA chain. Remedy metrics can show which body had authority, which body acted, and where the correction occurred. Clear attribution is a protection against both under-accountability and over-accusation.

A practical scorecard

A remedy-centered ICANN scorecard could be built without replacing the existing public record. It would add a cross-channel layer that reads the record through accountability outcomes.

The first section would be intake. It would count material claims of institutional error by channel: public comment, reconsideration, independent review, Ombudsman, specific review, Empowered Community petition, board correspondence and staff correction. The metric would not treat every claim as valid. It would identify the claim type and the authority asked to act.

The second section would be disposition. Claims would be marked accepted, partially accepted, rejected on merits, rejected as outside authority, withdrawn after institutional change, withdrawn without institutional change, moot, settled, unresolved or pending. Each category would require a short public explanation. This alone would improve the record because withdrawals and outside-authority findings would stop disappearing into ambiguity.

The third section would be correction. It would count record corrections, reasoning corrections, procedural corrections, substantive reversals, remands, delayed effective dates, compensation or cost shifting, future-control changes and public commitments. Each correction would be tied to the original claim.

The fourth section would be time. It would show age bands, median time, longest open matters, time-sensitive matters, time from decision to implementation and matters where remedy arrived after the relevant decision window. Delay would be visible as a governance cost.

The fifth section would be recurrence. It would identify repeated defect types, repeated offices or decision stages, repeated late corrections, repeated out-of-scope findings, and repeated recommendations without demonstrated effect. It would also identify successful reductions in recurrence after a corrective measure.

The sixth section would be affected class. It would show whether the matter involved contracted parties, applicants, registrants, number-resource actors, governments, civil society, end users, technical operators or internal community bodies. The purpose is not identity politics; it is incidence. If some classes can obtain correction and others mainly receive thanks for participating, the public should know.

The seventh section would be institutional response. It would identify whether the correction was voluntary, board-directed, staff-directed, review-directed, settlement-based, court-related or external. Early voluntary correction should be treated as a strength. Late correction after expensive review should be treated as a cost signal.

None of this requires ICANN to admit wrongdoing in every row. It requires ICANN to report the fate of claims of error in a way the community can audit.

Guardrails against bad metrics

Bad remedy metrics would create new problems. If every correction is treated as shame, staff will avoid corrections. If every denial is treated as victory, the dashboard becomes public relations. If every claim is counted equally, tactical complainants can distort the picture. If every metric is reduced to a single score, nuance disappears. The design needs guardrails.

First, the scorecard should avoid a universal grade. Accountability is multi-dimensional. A year can show faster correction but more errors, fewer challenges but higher cost, more public-comment influence but weaker implementation follow-through. Compressing that into one number would reproduce the same problem the scorecard is meant to cure.

Second, the scorecard should preserve qualitative reasons. A correction without reasons is hard to learn from. A denial without reasons is hard to trust. A withdrawal without context is hard to interpret. Short reason categories, backed by documents, are better than silent counts.

Third, the scorecard should distinguish claim validity from channel adequacy. A weak claim can be properly denied. A strong claim can be outside the wrong channel. A recurring outside-channel denial can reveal a missing remedy even when each individual denial is formally correct. This distinction is essential for ICANN because its mission limits are real. The institution should not act outside mission, but mission limits should not become a black hole where accountability evidence vanishes.

Fourth, the scorecard should report uncertainty. Some settlements are confidential. Some informal corrections may not disclose all details. Some records may not identify claimant cost. Some harms are hard to quantify. The answer is to mark limits, not to omit the category. A public "unknown or confidential" count is still more honest than no count.

Fifth, the scorecard should be reviewed independently. If the same office whose performance is measured controls categories, coding and presentation, the dashboard may become another public-comment summary problem. Independent review does not need to be adversarial. It can be a periodic audit by a cross-community group with access to the public record and authority to challenge category choices.

The aim is to make accountability measurable without turning it into a game. The best metric is one that changes behavior before the next challenge arrives.

The cultural change

The hardest change is cultural. ICANN's legitimacy language often emphasizes openness, participation, global diversity, bottom-up policy and the multistakeholder model. Those are important values. But they can make correction feel like an attack on the model. If a decision is reversed, critics may say the model failed. Defenders may say the model worked because it corrected itself. Both reactions miss the practical point. The public needs to know whether correction is normal, timely and effective.

A mature accountability culture treats correction as institutional maintenance. A corrected rationale is not humiliation. It is better public reasoning. A remanded decision is not collapse. It is a control working. A public admission that a comment summary missed a material issue is not weakness. It is evidence that consultation means something. Cost shifting after a successful challenge is not generosity. It is recognition that accountability should not be financed entirely by the person harmed.

This culture would also change how ICANN reports success. Instead of saying only that more people participated, it could say that three material objections changed final text, two were rejected with published reasons, one was deferred to a named forum, and one late-stage defect led to a new check before board action. Instead of saying that review recommendations were implemented, it could say that the recurrence rate for the defect the recommendation targeted fell, stayed flat or could not yet be measured.

Instead of saying that accountability mechanisms remain available, it could say how often they produced correction, how long they took and who bore the cost.

Such reporting would not satisfy everyone. ICANN is a contested institution. Some critics want stronger governmental control. Some actors want less central authority. Some disputes are really market disputes, geopolitical disputes or private commercial fights dressed in accountability language. A remedy dashboard will not settle those arguments. It will, however, make one question harder to avoid: when ICANN was shown a correctable error, did it correct it?

That is the question meetings cannot answer.

Sources

This article uses the current ICANN Bylaws page, which identifies the bylaws as amended on 3 July 2026 and sets out ICANN's mission limits, transparency commitments, website disclosure obligations, reconsideration, independent review and Accountability and Transparency Review provisions: https://www.icann.org/en/governance/bylaws.

It also uses ICANN's Accountability Mechanisms page, which describes accountability and transparency as safeguards for the multistakeholder model and summarizes reconsideration, independent review and related mechanisms: https://www.icann.org/resources/pages/mechanisms-2014-03-20-en.

The reconsideration discussion relies on ICANN's Requests for Reconsideration page, including its public status update, timelines and request list: https://www.icann.org/resources/pages/accountability/reconsideration-en.

The annual-report discussion relies on ICANN's Annual Report page, which links complete reports across multiple years and shows how regular reporting is already part of ICANN's public record: https://www.icann.org/resources/pages/governance/annual-report-en.