Summary

  • The official IGF 2024 participation page reports 11,749 registrations, 7,343 people who collected badges and participated onsite, more than 2,800 online entities and a combined total above 10,143. These are four related but different measures.
  • The scale establishes demand for a global discussion venue. It also shows considerable host-country and regional mobilisation: half of reported entities were assigned to Asia Pacific, government accounted for 48 per cent, and 64 per cent described themselves as first-time entities, most from the host country and region.
  • Attendance does not create representation. The meeting had no global electoral roll, population-weighted sampling frame, delegation rule for ordinary users, credential from network operators or mechanism by which absent people could instruct and correct those present.
  • The official total does not disclose enough about intensity to distinguish a person who entered one online session from someone who organized, spoke, moderated or attended throughout. It cannot measure speaking time, proposal success, informal access, repeat institutional presence or downstream influence.
  • Better participation reporting should preserve the record total while adding denominator definitions, unique-person reconciliation, host effects, attendance intensity, programme roles, repeat participation, language and access barriers, speaker distribution, session-selection outcomes and evidence of institutional handoff.

A record can be real and still answer the wrong question

The 10,143 figure deserves to be taken seriously. Too much institutional criticism begins by treating every official number as publicity and then loses the useful evidence inside it. Riyadh was not an empty hall described with a large registration list. The official count says 7,343 people collected badges and participated at the venue. More than 2,800 participated online. The meeting carried 307 sessions, an exhibition village, high-level, parliamentary, youth, business and judiciary tracks, workshops, open forums and intersessional discussions. A gathering of that size requires demand, money, time and organizational capacity.

The error begins when a scale statistic is made to answer an authorization question. How many people took part? Who were they? How intensely did they take part? Whose experience could they speak from? Who asked them to speak? What body, if any, could they bind? These questions sit next to one another in a dashboard but are not interchangeable.

Large attendance can validate a venue in a market-like sense. People and institutions reveal that they value access enough to bear at least some cost. The host reveals a willingness to provide facilities and political attention. Organizers reveal the capacity to attract hundreds of proposed sessions. Remote participation reveals demand beyond the people able to enter the building. These signals can justify maintaining and improving the forum.

Mandate has a different structure. It requires a defined constituency, a rule for choosing or authorizing the person who acts, a known scope of authority, and some route for the constituency to receive an account and correct or replace the actor. The Riyadh total supplies none of those elements. It is a headcount of participation under the Forum's terms, not a transfer of authority from billions of Internet users.

That boundary protects the IGF. If the meeting is evaluated as a world election it will fail a test it was never designed to pass. If it is evaluated as an open forum, the size becomes impressive evidence while remaining open to a sharper question: how well did this temporary public discover issues, exchange knowledge and make its influence traceable?

Four numbers sit behind the headline

The official statistics provide a useful sequence. First came 11,749 registrations. Second, 7,343 entities collected badges and participated onsite. Third, more than 2,800 participated online. Fourth, onsite and online participation produced a total above 10,143.

Each number has a different denominator and evidential weight. Registration records intention or eligibility at some point before or during the meeting. It does not show that the registrant entered a room, opened a session or remained engaged. Badge collection is stronger evidence of physical arrival, but a badge does not disclose whether the holder attended for an hour or five days. Online participation is stronger than a livestream view if it records entry into a participation platform, but the published page does not explain the minimum activity that made a person count.

The combined total depends on deduplicating people who may have appeared both onsite and online.

Simple arithmetic clarifies why the distinctions matter. The 7,343 badge holders are about 62.5 per cent of the 11,749 registrations. The combined 10,143 is about 86.3 per cent of registrations, although the online term is explicitly stated as more than 2,800 and the total as more than 10,143. The difference between registrations and reported entities is at least 1,606. That gap is not a failure. Registration attrition is normal. It is a reason to label the measures precisely.

Nor can the 2,800-plus online figure simply be added to badge collection without a published identity rule. A person onsite might join a virtual session from a laptop because a room was full, because the speaker was remote or because the participation interface carried chat. If that person appears in both systems, the combined unique-person total requires reconciliation. The word "entities" may already reflect such reconciliation, but the public note does not describe it.

This is not pedantry. A forum that asks the world to make evidence-based digital policy should model careful public statistics. It should define registration, onsite participation, online participation, unique participation, livestream viewing and remote-hub participation separately. The 20,000-plus livestream views reported for Riyadh belong in an audience measure, not silently inside the entity total. Views, sessions and people are different units.

The most defensible statement is therefore the official one: more than 10,143 people participated under the Forum's counting method, including 7,343 onsite badge holders and more than 2,800 online entities, following 11,749 registrations. The least defensible statement is that 10,143 equally engaged delegates collectively authorized a global position.

Demand is the strongest inference

What does the record total prove? First, it proves that a standing global venue for discussing Internet and digital policy can attract a substantial public. This is not obvious. Institutional calendars are crowded. Officials, engineers, advocates, researchers and businesses have many specialist meetings. Riyadh still drew thousands into a broad forum that could not promise a treaty, vote, licence, address allocation or regulatory order at the end.

That willingness reveals demand for translation. Digital-policy disputes cross professional languages. A regulator may understand legal powers but not routing practice. An operator may understand failure modes but not the distributional effect of a rule. A rights advocate may see a harm that a standards entity has never encountered. A company may know deployment constraints unavailable to a ministry. A forum creates encounters that no single operational institution can reproduce.

Second, the total shows demand for visibility. Session proposers, hosts, governments, firms and campaigns all value a global stage. Visibility can be used well or badly, but it is an institutional resource. The presence of 307 sessions indicates that the programme was not merely a sequence of speeches from one centre. Many actors wanted to frame a problem, present a practice, launch work or contest a dominant account.

Third, the scale shows that the IGF brand can mobilize newcomers. The official page says 64 per cent identified as first-time entities, with most coming from the host country and region. That is strong evidence that the meeting reached beyond a fixed travelling circuit. It also demonstrates the power of location. A host can expose a large local and regional population to institutions and debates that would otherwise remain geographically distant.

Fourth, the online number shows that participation demand exceeded the venue population. More than 2,800 people connected through the meeting's participation facilities, while more than 20,000 unique livestream views created a wider audience. Remote access is not the same as attendance in Riyadh, but it makes the meeting more than a local conference.

These conclusions are valuable enough. None needs the claim that the entities represented everyone. An institution becomes more credible, not less, when it states exactly what a statistic establishes.

The host effect is visible in the composition

Riyadh's scale cannot be interpreted without geography. The official statistics assigned 50 per cent of entities to Asia Pacific, compared with 16 per cent to Western Europe and Others, 11 per cent to Africa, 4 per cent to Eastern Europe and 2 per cent to Latin America and the Caribbean. Another 17 per cent were placed in an intergovernmental-organization category in the regional presentation, an unusual construction because institutional affiliation is not a geographic region.

The majority of onsite entities came from a list led by Saudi Arabia, followed by the United States, China, Brazil, Switzerland, Pakistan, the United Kingdom, the United Arab Emirates, Nigeria and other countries. Most online entities connected from Saudi Arabia, the United States, China, Germany and India. The IGF 2024 Progress Report adds that most first-time entities came from the host country and region.

This host effect is not evidence against international value. Rotation is partly meant to change who can enter. A meeting in Riyadh will be cheaper and institutionally closer for some people who would struggle to attend in Europe, Latin America or East Asia. Local universities, ministries, companies and professional networks can mobilize at a scale unavailable when the event is far away. A genuinely rotating forum should expect its composition to move.

But host mobilization limits what the total can stand for. A population shaped by the cost of travel, visa access, local institutional recruitment and geographic proximity is not a random or population-weighted sample of Internet users. The location can expand one region's presence while other regions fall. The official page reported Latin America and the Caribbean at only 2 per cent, five percentage points below 2023. That absence cannot be cured by saying the overall room was large.

The category structure also complicates comparison. If 17 per cent are removed from geographic groups because they work for intergovernmental organizations, regional percentages do not tell the reader where those people came from or whose conditions they understand. A staff member of an international body has both a location and an institutional affiliation. Mixing the dimensions conceals rather than resolves that dual identity.

The correct interpretation is conditional: Riyadh achieved exceptional reach, especially in its host country and region. It broadened the Forum's constituency of experience. It did not create a miniature demographic image of the Internet's global population.

Stakeholder labels reveal imbalance, not authorization

The official stakeholder distribution was striking: 48 per cent government, 25 per cent private sector, 12 per cent civil society, 7 per cent technical community, 7 per cent intergovernmental organizations and 1 per cent press. Government and business therefore accounted for nearly three quarters of reported participation.

This pattern says something important about the meeting. Riyadh attracted state and commercial institutions at unusual scale. It may have improved access to officials and executives whose presence is often requested at multistakeholder forums. It may also have tilted room capacity, speaker supply and informal attention toward institutions with salaries, travel budgets and formal delegations.

The percentages do not show what "government" meant in practice. A minister, regulator, municipal officer, diplomat, public-company employee and administrative support official may all enter a broad government category while holding very different authority. Private sector can include a global platform, a local supplier, a network operator, a consultancy or a person who selected the nearest label. Civil society contains rights groups, development organizations, professional associations, community initiatives, foundations and many other forms.

The technical community may include operators, standards contributors, researchers and Internet institutions whose interests diverge.

Nadia Tjahja, Trisha Meyer and Jamal Shahin's study of civil-society participation mapped 2,830 organizations across fifteen years of IGFs and showed why one stakeholder label cannot be treated as a homogeneous constituency. The same caution applies across the classification. A percentage can describe the labels chosen by people in the records. It cannot establish that the category selected them, that they consulted the category or that every relevant interest had access.

The Research ICT Africa report Beyond Multistakeholder Tokenism found that IGF participation had increased and become partly more diverse, while participation from least developed countries remained consistently low and available data could not answer basic questions about underrepresented groups, disability, language, online constraints or active involvement. Its most important lesson for Riyadh is methodological: top-line stakeholder proportions are a beginning, not a measure of meaningful equality.

A 48 per cent government share does not give governments 48 per cent of a global mandate. A 12 per cent civil-society share does not mean civil society consented to one eighth of an outcome. These were participation categories in a forum without weighted voting. They should be used to investigate access and concentration, not to allocate fictional shares of sovereignty.

A first-time majority is a success with an unresolved second half

Sixty-four per cent first-time participation is perhaps Riyadh's most interesting statistic. It weakens the claim that the Forum is only a closed annual reunion. It suggests successful local outreach and a significant injection of new experience. A newcomer may bring a problem that long-serving entities have normalized, a local practice absent from global debate or a constituency that had no previous route into the room.

New entry also changes the meaning of raw attendance. First-time entities need orientation, institutional maps, accessible language and visible routes from listening to contribution. If they arrive after themes and sessions have been selected, their presence expands the audience more than the agenda. If they enter a packed programme without knowing which institutions can act, they may leave informed but unable to intervene later.

The official percentage does not say how many newcomers organized sessions, spoke, moderated, served as rapporteurs, entered working groups or returned to a national initiative. It does not say whether their questions reached the record or whether they spent most of the meeting navigating acronyms and parallel rooms. Nor does it show whether first-time status was distributed across stakeholder and regional categories.

This is why retention should not be measured as a simple demand that everyone return. A healthy forum needs turnover. Some people obtain the knowledge or contact they need and should not become permanent conference professionals. The more useful follow-up asks whether a first-time entity acquired a durable route into the relevant institution, local network or future consultation.

Riyadh proved that the IGF can open a large door. It did not publish enough evidence to show where newcomers went after crossing the threshold. That is the unresolved second half of access.

Gender statistics identify a severe gap

The participation page reports 69 per cent male, 31 per cent female and less than 1 per cent another self-identification. The Progress Report gives 68 per cent male and 31 per cent female, a one-point difference that may result from rounding, timing or classification. Either version records a substantial imbalance.

This matters for more than visual diversity. Digital-policy burdens, access conditions, surveillance risks, labour markets and online harms are not evenly distributed. A meeting with more than twice as many men as women cannot use its total size to neutralize the skew. Ten thousand entities can contain more varied experience than one thousand while still reproducing a major exclusion.

The total also does not show where the imbalance sat. Entity registration, session organization, speaking roles, moderation, main-stage visibility and informal access are separate layers. A meeting could improve the overall gender ratio while leaving agenda or high-level roles concentrated. It could also have better speaker balance than the registration population. The published aggregate cannot answer either possibility.

Gender categories alone do not measure socioeconomic position, disability, age, language, race, rural access, care responsibilities or the safety of participation. The point is not to demand an unlimited identity questionnaire. It is to stop a single large denominator from concealing known differences in the cost and quality of access.

A credible record announcement would therefore pair the achievement with the gap: the Forum reached its largest reported participation while recording a heavily male population. Both facts belong in the institutional memory.

Participation is not representation

Representation is often used loosely to mean that different kinds of people were visible. In that descriptive sense, Riyadh contained representatives of governments, companies, organizations and technical institutions. Many entities could accurately speak for an employer or official office within the scope of their instruction. Some advocates could describe communities with which they had long relationships. Some individuals could speak only for themselves, which is entirely legitimate in an open forum.

The stronger claim is different. To say that the meeting represented global users or the Internet community implies an authorization relationship. No worldwide voter list selected the 10,143. Internet users did not receive equal or population-weighted seats. Network operators did not credential one delegation to speak for operations. People affected by a particular platform, shutdown, cybersecurity rule or identity system did not necessarily know that a relevant session existed.

Registration was open, but openness is not a sampling method. It removes one formal barrier and leaves many material ones: information, language, connectivity, travel cost, visas, time, organizational support, confidence, disability access and the opportunity to leave paid or care work. Self-selection tends to favour people already connected to institutions that recognize the meeting's value.

Nor does professional affiliation cure the problem. An employee attending as private sector may have authority to explain a company's practice but not to bind all companies. A government official may speak for a ministry but not for other states or for every citizen at home. A civil-society entity may bring deep community evidence without being elected by a global civil-society electorate that does not exist. A technical specialist may understand one layer of the network while lacking authority over autonomous operators.

The IGF does not need these fictional mandates. Its founding Tunis mandate makes it a forum for multistakeholder policy dialogue, exchange, advice, capacity and identification of emerging issues. It expressly excludes oversight and day-to-day technical operation. Entities contribute knowledge and argument. The institutions that later act retain their own authority and accountability.

Riyadh's legitimacy therefore rests on openness, plurality, transparent programme choices, evidence quality and honest handoff. It does not rest on turning attendance into consent.

The world's operators were not sitting as one body

The mandate error is especially dangerous when discussion concerns Internet operations. The Forum can host excellent debate about routing security, domain abuse, access networks, infrastructure resilience or protocol deployment. Operators and technical bodies can learn from affected communities and public authorities. None of that makes the meeting an operational command centre.

The Internet is run through many separate institutions and networks. Autonomous network operators make routing and security choices. Regional registries administer number resources under their own policies. Domain registries and registrars operate contracts and technical systems. Standards organizations use their own contribution and consensus rules. Governments exercise legal powers within jurisdiction. Platforms control services through corporate governance and contracts.

A entity from one of these institutions may carry real organizational authority. The attendance total does not pool those authorities. Ten thousand people in one venue cannot, by presence alone, reassign an address block, validate a route, change a protocol standard, alter a registry policy or compel a platform. Any later action must pass through the responsible institution's rules.

This is not a weakness in the meeting. The separation lets an operator hear criticism before a technical change, lets a regulator learn the operational cost of a proposal and lets users contest a design assumption. The Forum's value lies partly in convening authority holders without absorbing their authority.

Public accounts should therefore use verbs carefully. Riyadh discussed, surfaced, compared, recommended and connected. A government adopted, a standards body approved, a network deployed or a company changed a service only when evidence shows that separate act. Scale can amplify an idea; it cannot execute it.

There was no vote hiding inside the applause

The IGF does not produce negotiated outcomes. Its annual messages and summary materials synthesize discussion across many sessions. That editorial work is useful because no reader can absorb 307 events. It also creates a risk that recurring statements from a self-selected programme are heard as consensus of a global public.

Attendance cannot supply the missing vote. A person who registered did not vote on every proposition. A badge holder may not have entered the relevant room. An online entity may have listened without using the floor. A livestream viewer is part of the audience, not a signatory. Silence can mean agreement, disagreement, absence, language difficulty, lack of time or a decision to listen.

Even within a session, applause and chat reactions have no stable denominator. The people present were drawn by the subject and schedule. Organizers chose speakers. Moderators allocated time. Parallel sessions removed potential entities. A popular statement can be meaningful evidence of room sentiment without becoming a plebiscite.

The responsible formulation is bounded: a view appeared in a named session; multiple session reports recorded a similar concern; a synthesis identified a recurring theme; contributors to a stated process endorsed a recommendation. Each description preserves where the evidence came from.

The irresponsible formulation invokes "10,143 stakeholders" as if each had reviewed and approved the statement. The number then becomes a source of consensus capture: the prestige of the whole crowd is attached to a conclusion made by a smaller and often unmeasured subset.

Riyadh's record should make the Forum more careful, not less. The larger the audience, the more tempting it is to borrow their presence for claims they never made.

Programme power was concentrated before most people arrived

Most entities entered after the principal agenda architecture had already been built. The Multistakeholder Advisory Group reviewed thematic inputs, agreed the four subthemes, evaluated workshop proposals and balanced the programme. The Secretariat screened submissions and supported the process. Session organizers selected proposed questions, speakers and formats. The host supplied the venue and much of the conference environment.

The 2024 workshop review rules show the sequence. Proposals passed an initial Secretariat screening, received scores from MAG members, and then entered a MAG discussion in which overall topic balance could alter the ranked result. Proposals that did not score highly could receive further consideration if they helped balance the programme. This is a reasoned allocation mechanism, not a crowd vote.

Open thematic calls and session submissions broaden access to agenda formation. They do not mean every eventual attendee had equal influence. A person learning of the Forum at registration cannot retroactively propose a subtheme. A first-time attendee may contribute from the floor, but floor time occurs within a question, speaker set, room and schedule already chosen.

The MAG's power is real and bounded. It allocates scarce programme visibility. It does not decide the substantive policies discussed. Its accountability should match its actual role: published criteria, conflicts, selection statistics, reasons for balance adjustments, accessible calls and analysis of whose proposals succeed.

The attendance record is not evidence that programme authority became dispersed among 10,143 people. It is evidence that a programme selected through a much narrower year-long process attracted a very large public.

Counting bodies misses intensity and role

Imagine four entities. One organizes a main session, moderates another, attends five days and holds meetings with officials. One speaks remotely on a panel and answers questions for an hour. One badge holder visits the exhibition village during an afternoon. One online registrant joins a workshop for twenty minutes. All may properly count as entities, but their opportunity to shape the meeting differs radically.

Headcounts flatten those differences. That is acceptable for reach and unsafe for influence. A full account needs role and intensity measures without turning participation into surveillance. It can count organizers, speakers, moderators, rapporteurs and general attendees; the number of sessions joined; days active; floor interventions; and whether a contribution entered a report. These can be published in aggregate.

Repeat participation matters too. Newcomers expand access, while long-serving entities hold institutional memory, proposal skill and dense relationships. The concern is not that experienced people return. It is whether the same organizations repeatedly control agenda roles while headline attendance grows around them.

Speaking time is another missing layer. A session may list diverse speakers but allocate the most time to one category. Online questions may be visible in chat but never voiced in the room. A moderator may invite an affected entity only after senior speakers consume the schedule. Entity totals cannot reveal these patterns.

Informal access is harder to count and no less important. People onsite meet during breaks, meals and bilateral appointments. They learn which institution is preparing a decision and whom to contact. Remote entities can gain knowledge while missing the corridor in which a future coalition forms. The official total combines both modes but should not imply equal opportunity.

The policy lesson is simple: use headcount to measure reach. Use role, intensity, interaction and follow-through to measure participation quality. Use authorization evidence to measure mandate. No single number can do all three jobs.

The affected population is the missing denominator

Internet governance discussions often concern populations far larger than the meeting. A session on connectivity may affect rural communities unable to join a high-bandwidth platform. A debate on content governance may concern speakers of languages absent from the panel. A cybersecurity proposal may shift cost to small operators without policy staff. An identity measure may affect migrants, children or people lacking standard documents.

The relevant denominator therefore changes by issue. The question is not only whether Africa, Asia Pacific or Latin America appeared in the whole meeting. It is whether the people materially affected by a particular proposal were present in that discussion, could understand it, had time to contest it and had a route into the institution that might act.

A global attendance total can coexist with a session missing its affected public. It can also coexist with excellent small sessions in which organizers recruited the right expertise and lived experience. Scale at meeting level tells us little about fit at issue level.

This is where representation claims should become specific. An organization may say it consulted 200 members under a published method. A public official may carry a legal portfolio. A community advocate may disclose the relationships through which evidence was gathered. A network operator may explain which systems it runs. These claims can be evaluated.

"The global community was present" cannot be evaluated because the category has no roster, selection rule or scope. It converts an aspiration into a fact. Riyadh's ten thousand should not be used to fill the evidential gap.

The report needs a denominator ledger

A better annual participation statement would begin with a short ledger. How many unique people registered? How many received approval? How many collected a badge? How many entered at least one online meeting room? What minimum activity defined online participation? How were people using both modes deduplicated? How many joined through remote hubs? How many only watched a stream? Which totals are exact and which are thresholds or estimates?

The second layer would describe intensity. Report active days and sessions in broad bands, not personal histories. Distinguish programme contributors from general entities. Show how many questions or speaking interventions came from onsite, online and hub entities. Record whether online contributions were carried into session reports.

The third layer would examine agenda access. Publish thematic-input counts, proposal numbers, acceptance rates and organizing roles by region, stakeholder group, gender and first-time status where privacy permits. Show whether a large newcomer population also entered the programme or mainly observed it.

The fourth layer would record barriers. Survey people who registered but did not participate. Ask about visa, cost, timing, language, accessibility, connectivity, safety and schedule. Include those findings beside the success total rather than in a separate improvement exercise few readers see.

The fifth layer would trace handoff. For major recurring recommendations, identify the government, company, standards body, operator, registry or international institution with authority to respond. Later reports should state whether that body considered the issue. This converts participation into accountable influence without inventing a forum mandate.

Such a ledger would not diminish Riyadh. It would make the record more durable. Future meetings could distinguish growth from changes in counting and compare access quality across hosts.

What 10,143 permits the Forum to say

The Forum can say that IGF 2024 was its largest reported annual meeting. It can say the hybrid event drew more than 10,143 entities, with 7,343 onsite and more than 2,800 online following 11,749 registrations. It can say government and private-sector participation was especially strong, that Asia Pacific was the largest regional grouping, and that a large first-time population was mobilized around the host country and region.

It can say the numbers demonstrate appetite for a global, open discussion venue. It can reasonably infer that location changes access. It can use the record to argue for resources, better remote facilities, multilingual support, newcomer orientation and stronger data collection.

It cannot say the count is a vote of confidence in every session, output, host policy or institutional position. It cannot assign absent people to entities who share a country or stakeholder label. It cannot infer equal voice from equal admission. It cannot treat global reach as authority over technical operation.

Nor should critics say the number proves nothing. It proves a great deal about mobilisation and convening power. A careful analysis resists both inflation and dismissal.

A six-part test for future record claims

Every record participation announcement should answer six questions.

First, what exactly was counted? Separate people, registrations, badges, platform entries, sessions and views. Second, how unique was the count? State the rule for duplicate identities across modes and days. Third, who had access? Show region, stakeholder affiliation, gender, development context, language and relevant barriers without confusing these dimensions.

Fourth, what could entities do? Distinguish watching, attending, asking, speaking, organizing, selecting and deciding. Fifth, whom could they speak for? Record formal office, organizational instruction, consultation or personal capacity without assuming stakeholder-wide delegation. Sixth, where did influence go? Identify the institution that received an idea and the authority under which it later acted or declined.

Riyadh performs strongly on scale and partial composition. It performs less strongly on uniqueness, intensity, role and follow-through because the published statistics do not provide enough information. On global mandate it does not perform at all, because no such mandate was offered or created.

The test is reusable because it does not depend on whether attendance rises or falls. A smaller meeting could have deeper participation. A larger one could broaden discovery. A remote-heavy one could transform access while leaving selection power unchanged. The measures should reveal those differences rather than rank every year on one line.

Turnout cannot validate the host's wider record

A record meeting also creates reputational value for its host. The city appears in the name of annual messages, leaders welcome an international public and local institutions gain access to a global professional network. That benefit is a normal part of hosting. It becomes misleading when attendance is treated as entity approval of the host government's unrelated laws, policies or public record.

The 10,143 did not answer a referendum on Saudi Arabia. People attended for different reasons: a session topic, an employer, a speaking role, professional relationships, local opportunity or access to officials and specialists. Some may have held strong views about the host; others may have had none. Registration did not ask them to endorse a political proposition, and presence cannot be converted into one after the fact.

The converse is also true. Concern about a host does not erase the work of entities or prove that every session was controlled. A session can produce serious evidence inside a politically contested setting. A local first-time entity can gain useful access without becoming an instrument of official reputation. Analysis should test specific restrictions, exclusions, programme interventions or access failures rather than infer them from the host's identity alone.

Host legitimacy and forum legitimacy therefore need separate evidence. The Forum should publish the terms governing access, the division of programme responsibility, any restrictions encountered, entity-safety arrangements and how complaints were handled. The host should claim credit for facilities and support actually provided. Neither should describe the crowd as endorsement.

This boundary matters for future rotation. If attendance automatically blesses a host, critics may treat nonattendance as the only responsible act and local entities lose an important opening. If attendance is understood as participation in an independent forum, people can engage, scrutinize and contest without surrendering the meaning of their presence.

Conclusion: the crowd authorized only its own participation

Riyadh's record matters. More than ten thousand entities showed that Internet governance is not an esoteric concern confined to a few technical institutions. Governments, companies, advocates, specialists, international bodies, journalists and newcomers devoted attention to a forum whose principal outputs were discussion, knowledge, networks and recommendations. The host-country effect brought many first-time entities into view. Online access extended the meeting beyond the venue.

The same evidence contains warnings. Government and business dominated the stakeholder distribution. The regional pattern was heavily shaped by the host and left Latin America and the Caribbean with a very small share. The gender gap was wide. Registration, onsite attendance, online participation and viewing were not equivalent. The public statistics did not measure influence, authorization or equality of voice.

The 10,143 therefore created no hidden legislature and no operational command. Entities authorized their own presence. Some carried authority from a government, employer or organization, limited to that institution. Others brought expertise, testimony or personal judgment. Together they formed an unusually large deliberative public, not an electorate for the Internet.

That is enough to justify pride and scrutiny at once. The IGF can claim demand, reach and convening power. It can claim that its record exposed thousands of people to ideas and institutions they might not otherwise encounter. It can improve the chance that a neglected concern reaches someone able to act.

The institution should stop there unless evidence supports the next link. When an operator deploys, a regulator rules, a company changes practice or a standards body agrees, that institution must own the decision. When an annual message summarizes debate, it must retain the modest status of a synthesis. When participation grows, the report must disclose whose participation, in what mode and with what opportunity.

The number becomes more impressive when it is not asked to impersonate a mandate. Riyadh proved that the Forum was wanted. It did not prove that the Forum, its programme or its entities spoke for everyone the Internet connects.

Sources