Summary
- The founding mandate was wider than the main-stage programme in Athens. It expressly included public-policy issues concerning critical Internet resources, but the four principal themes were openness, security, diversity and access, with development and capacity building treated across them.
- The agenda emerged through a chain rather than a single vote: written contributions, open consultations, a Secretariat synthesis, deliberation by a Secretary-General-appointed Advisory Group, and recommendations accepted within the Secretary-General's convening authority.
- That chain created meaningful access to propose issues but left a weaker public record of elimination. The archive shows inputs, transcripts and the final programme more clearly than it shows which proposals were merged, demoted, rejected or deferred and why.
- Omitting a critical-resources main session reduced the risk that the new Forum would replay the most antagonistic WSIS arguments. It also removed from the central programme a question expressly listed in the Forum's mandate and important to many developing-country governments and civil-society entities.
- The success of Athens should not be measured only by attendance or civility. A founding agenda should also be judged by whether later readers can reconstruct the selection, identify missing interests and distinguish a temporary peace-making compromise from a settled limit on legitimate discussion.
A programme can become a constitution before anyone calls it one
Conference agendas look administrative. They allocate rooms, hours, moderators and speakers. In a new institution, however, the first agenda does something more consequential. It gives entities a shared map of the field. It announces which questions belong together, which terms deserve a main stage, which conflicts can be handled in smaller rooms and which demands will not define the institution's public face.
Athens was especially exposed to this effect because the Internet Governance Forum had no inherited annual practice. Its creators could not point to last year's headings and make incremental changes. They had to convert a very broad mandate, and the unresolved politics of the World Summit on the Information Society, into four days that governments, companies, technical institutions and civil society would all attend.
The result was durable. The IGF archive for the inaugural meeting records the overall theme as "Internet Governance for Development" and the principal sessions as openness, security, diversity, access and emerging issues, alongside opening, scene-setting, summing-up and forward-looking sessions. The next meeting retained the four principal headings and added critical Internet resources. Later programmes changed, combined and expanded the vocabulary, but Athens established the Forum's first recognisable grammar.
That grammar was not neutral. "Access" directs attention toward connectivity, price, interconnection and standards. "Diversity" directs it toward languages, local content and internationalised names. "Openness" can contain expression, information flows and access to knowledge. "Security" can contain trust, malware, spam, crime and privacy. Each is broad enough to host conflict. Together they make the Internet appear as a set of user-facing policy tensions rather than, for example, a struggle over institutional control of naming and numbering.
There were defensible reasons for that choice. The first Forum needed enough common ground to happen. Yet a prudent compromise at a founding meeting can harden into an assumption about the proper boundary of debate. That is why an agenda should be examined not only as a list of sessions but as a record of choices among competing definitions of Internet governance.
The Tunis mandate was deliberately wider than the Athens headings
The Tunis Agenda for the Information Society did not tell the Secretary-General to convene a conference on four preselected themes. Paragraph 72 supplied a wide mandate. It called for discussion of public-policy issues related to key elements of Internet governance; exchange of information and best practices; advice to stakeholders; engagement with existing mechanisms; facilitation of discourse; identification of emerging issues; reinforcement of capacity; promotion of WSIS principles; and attention to the needs of everyday users. It expressly included discussion of issues relating to critical Internet resources.
Other provisions constrained the Forum. It was to be multilateral, multistakeholder, democratic and transparent. It was not to replace existing institutions or supervise day-to-day technical operations. It was to have a lightweight and decentralised structure, use existing resources where possible, and operate through an open and inclusive convening process.
These provisions did not mechanically produce a programme. They contained tensions. A forum could discuss the public policy surrounding the root zone, domain names and Internet addresses without replacing ICANN, the IANA functions or the regional registries. But entities differed over where discussion ended and institutional challenge began. A call to identify emerging issues could justify almost anything. A commitment to development could be a dedicated debate about distribution or an adjective attached to every other subject.
The mandate therefore supplied authority and ambiguity at the same time. It authorised a broad conversation while leaving agenda makers to decide how to make that conversation manageable. The critical question is not whether the Advisory Group complied with a simple checklist. No such checklist existed. It is how the group exercised discretion where the founding text was broad, politically charged and internally balanced.
One useful test is fidelity by visibility. An issue named in the mandate does not need a plenary every year. But if it is omitted from the principal programme at the founding meeting, the public should be able to see that the omission was considered, why it occurred, whether the issue remained available elsewhere and when the choice would be reviewed. Without that record, a temporary programme choice can be mistaken for an interpretation of the mandate itself.
No single actor chose Athens, but authority was not evenly distributed
The first agenda came through several gates. The official history of the preparatory process identifies open consultations in Geneva on 16-17 February and 19 May 2006. The February meeting drew about 300 entities and addressed scope, substantive priorities, structure and functioning. Written contributions and a questionnaire added another route for participation. A synthesis paper summarised submissions and consultations.
That openness mattered. Governments did not meet alone to write a diplomatic agenda. Companies, civil-society organisations, members of technical institutions, academics and individuals could put issues into the record. Transcripts preserved disagreement in a way that a negotiated communique often does not. The process created a visible upstream field of claims.
The next gates were narrower. On 17 May, Secretary-General Kofi Annan announced an Advisory Group chaired by his special adviser Nitin Desai. The announcement described 47 people including the chair, drawn from government, business and civil society, including technical and academic communities. Its principal task was to prepare the substantive agenda and programme. It met on 22-23 May, after the second consultation, and again on 7-8 September.
The May meeting recommended the agenda, programme, structure and format. Those recommendations went to the Secretary-General. In September the group shortlisted panelists, selected workshops and finalised the programme. The Secretariat organised the consultations, collected and synthesised material, supported the Advisory Group and converted decisions into a functioning event. The Greek host supplied the venue and practical setting. The Secretary-General remained the legal convenor.
This was neither direct entity control nor unilateral Secretariat authorship. It was a delegated programme-making chain. Open participation influenced the pool; appointed advisers filtered it; the chair identified agreement; the Secretariat made synthesis and implementation choices; the Secretary-General's authority closed the chain.
The distinction matters because "multistakeholder" describes composition and participation, not one decision rule. The public could speak at an open consultation without possessing a vote on headings. Advisory Group members came from several sectors but served in their personal capacities. The chair's ability to state consensus was not the same as an election. The Secretary-General did not need to submit the final schedule to all contributors for ratification.
This design may have been appropriate for an experimental, non-negotiating forum. Its legitimacy depended less on a fictional claim that everyone jointly decided and more on faithful transmission: Did the smaller bodies accurately report the range of input? Did they explain consequential exclusions? Could a rejected proposer discover what happened? Those are agenda-accountability questions, not demands for parliamentary procedure.
The archive is rich at the entrance and thin at the point of elimination
The surviving record is unusually valuable for a 2006 international process. It includes consultation transcripts, entity data, written contributions, the synthesis, meeting programmes and main-session transcripts. The Athens proceedings page lets a reader compare the advertised architecture with what entities actually said. The publication Internet Governance Forum: The First Two Years, edited by Avri Doria and Wolfgang Kleinwächter in cooperation with the Secretariat, preserves accounts of the preparatory stages, meetings and early reflections.
Yet the record is asymmetric. It is easier to answer "What did people submit?" and "What appeared in Athens?" than "How did each material proposal travel between those points?" The public record does not provide a complete disposition table showing the proposal, proposer, decision, reason, conflict check, alternative placement and reviewer for every significant agenda request.
That missing middle is where agenda power operates. A proposal can disappear through explicit rejection, but it can also be merged under a broad heading, moved from a main session to a workshop, deferred to a later year, reframed until its original institutional claim becomes unrecognisable, or left without speakers able to carry it. All outcomes are not equivalent. Nor are all necessarily improper. They simply need different explanations.
Workshop selection intensified the issue. The first retrospective reports that 36 workshop proposals were submitted by the deadline across subjects from multilingualism to expression. Workshops gave entities more room to organise focused discussion, and contemporary observer Milton Mueller found many of them more focused and mobilising than the plenaries. But the September Advisory Group meeting selected the workshops and shortlisted panelists. A topic's survival somewhere in the programme did not guarantee equivalent visibility, interpretation or influence.
An archive can therefore be open while selection remains partly illegible. Publishing every incoming contribution is not enough if the consequential act is the comparison among contributions. Conversely, releasing private deliberations verbatim is not necessary. Advisers may need space to test compromise. The accountability requirement is a structured explanation after selection, not permanent public performance during every negotiation.
Four headings converted political disputes into governable questions
The four principal themes were skilfully designed. Each took an area of conflict and translated it into questions that actors with different authority could discuss without first settling who governed the Internet.
Openness joined freedom of expression, information flows and access to knowledge. That allowed human-rights advocates, governments, platforms, publishers and technical entities to address censorship and content without requiring the Forum to issue a binding free-speech code. In Athens, expression became prominent both in the main session and in workshops. Amnesty International's campaign against Internet repression supplied a vivid public focus.
Security was framed around trust and confidence, including spam, phishing, viruses, crime and privacy. Pairing protection with privacy resisted a purely law-enforcement definition. It also made room for operational and user evidence. The heading was broad enough for actors to disagree about whether security required control, cooperation, design changes or rights protection.
Diversity linked multilingualism, local content and internationalised domain names. This connected cultural inclusion with technical systems. It made language a matter of participation and infrastructure rather than ceremonial representation. It also gave naming questions a place without opening the larger institutional conflict over control of critical resources.
Access centred availability, affordability, connectivity policy, interconnection costs, interoperability and open standards. It brought development into concrete relationships: who could connect, at what price and through which dependencies. For entities from countries with low penetration or high transit costs, this was not a secondary social concern. It was the practical condition for joining every other discussion.
These translations were a genuine achievement. They let entities speak across professions. A minister, an engineer, an activist and a company executive could enter the same session through different evidence. The headings did not demand consensus on institutional sovereignty before dialogue began.
But translation also edits. "Diversity" could absorb internationalised domain names while excluding the authority structure around the DNS. "Access" could discuss interconnection cost without confronting control of address distribution. "Openness" could debate expression without resolving which institutions could impose remedies. The programme foregrounded effects and practices; it backgrounded control over the systems that produced some of those effects.
That was not concealment in every case. It was a choice to begin with tractable public-policy questions. The problem arises only when tractability becomes the test of whether a subject belongs. A global forum is needed partly because some important questions are not easily made comfortable.
Critical Internet resources were within the mandate and outside the main frame
The clearest omission was critical Internet resources. The Tunis Agenda named the subject. The WSIS negotiations had made authority over the DNS root, domain names, Internet addresses and related institutions one of the most contested parts of global Internet politics. Yet Athens had no dedicated principal session on it.
Later evidence makes the decision difficult to dismiss as accidental. In a retrospective chapter on the first IGF years, MAG entity Jeanette Hofmann wrote that the group could not agree to include critical Internet resources as a main programme item and that some members feared the session would become an attack on ICANN. Her account is entity testimony, not a neutral minute, but it supplies a plausible explanation consistent with the public sequence. In 2007, Rio added critical Internet resources as a fifth principal theme.
The fear was understandable. WSIS had exposed deep disagreement about the United States' role, intergovernmental authority, ICANN's legitimacy and the meaning of internationalisation. Replaying fixed speeches could have crowded out every other subject. A failed confrontation in the first year might have convinced businesses or technical organisations that the Forum was a venue for institutional attack, while governments dissatisfied with existing arrangements might have treated procedural restraint as bad faith.
Avoiding the dispute bought room for relationship building. Athens entities who had previously addressed one another through rival blocs sat on common panels. The Forum demonstrated that an international gathering could discuss Internet policy without negotiating a treaty or issuing instructions to technical operators. That achievement should not be trivialised.
The cost was also real. Agenda makers protected the new institution by excluding from its central stage the subject most closely associated with why many governments had demanded a global process. This affected not only symbolism but evidence. A main session determines which questions receive prominent speakers, shared attention, official summing-up and press coverage. Workshops can preserve a conversation, but they do not necessarily preserve equal agenda status.
The omission also created a precedent about civility. If the most contested institutional question is excluded because it could reproduce conflict, actors benefiting from the status quo gain a structural advantage. They need only describe challenge as repetitive or destabilising. Those seeking change must first prove they can discuss power without making powerful institutions uncomfortable.
The better interpretation is narrower. Athens made a temporary peace-making choice for a first meeting. Rio's addition of the subject shows that the choice did not become a permanent jurisdictional bar. A transparent disposition record in 2006 could have made that temporariness explicit: no dedicated main session this year; related workshops accepted; reasons stated; review promised for the next programme.
Development was everywhere, which risked making it answerable nowhere
The overall title, "Internet Governance for Development", gave development elevated status. Secretary-General Annan's message to Athens described it as an overarching priority. Capacity building and development were intended to run across the programme.
That design had merit. Development is not one topic beside security or openness. Affordability shapes access; language shapes diversity; institutional capacity shapes security; rights shape openness. Isolating development in one session can allow every other panel to assume wealthy-country infrastructure, travel, expertise and market conditions.
Cross-cutting status can nevertheless dilute responsibility. When a subject is everywhere, no session chair owns the task of showing whether it was actually addressed. A programme may mention developing countries repeatedly while avoiding distributional questions: who pays for connectivity, whose standards become mandatory, which regions supply speakers, whose operational evidence counts, and whether capacity funding lets entities influence planning rather than merely attend.
The same problem affects everyday users. The Tunis mandate identified their concerns. Athens spoke about users across openness and security, but users were not a coherent appointing constituency with assured stage time. Representatives of organisations could plausibly describe user effects without being selected by users. A cross-cutting label did not solve that authority gap.
Auditing a cross-cutting priority therefore requires a matrix, not a slogan. Each main session and workshop should identify the development question it addresses, the affected communities represented, evidence from low-resource conditions, and any issue the organisers could not cover. The final report should compare the promise with the programme. Without that step, "for development" risks becoming an atmosphere rather than a measurable orientation.
This is particularly important for first agendas. The headings chosen in Athens influenced who built expertise around the Forum. If development appears mainly through access and capacity, future organisers may underweight competition, labour, taxation, procurement, infrastructure ownership or institutional power even though these also shape development. Early categories attract repeated proposals and familiar speakers. Unnamed dimensions must fight their way in as exceptions.
Main sessions and workshops formed two different levels of recognition
Athens is often praised for its workshops and dynamic coalitions. The praise is justified. Entity-organised sessions let narrower communities pursue subjects that broad panels could only introduce. Dynamic coalitions gave volunteers a way to continue work on privacy, open standards, access and other issues after the meeting. The less formal spaces produced relationships and initiatives that a negotiated conference might have suppressed.
But a distributed programme is not flat. Main sessions had common visibility, simultaneous attention from the full meeting, central transcription and a role in the Forum's public narrative. Workshops competed for rooms, times, speakers and audiences. Dynamic coalitions depended on voluntary energy and resources after entities went home.
Moving a proposal from the main programme to a workshop can be a sensible response to limited time. It can also alter the question. A plenary on control of critical resources asks the whole institution to acknowledge the dispute. A technical workshop on one component may produce better detail while leaving the institutional conflict outside. Both have value; they are not substitutes.
The first Forum's structure allowed agenda makers to claim breadth because many issues appeared somewhere. An audit should instead track level of recognition. Was the proposal a principal theme, main session, workshop, coalition, speaker intervention or merely an archived submission? Did it receive a response? Was it summarised accurately? Could its proponents continue work?
This is not an argument that every proposal deserves a main session. Scarcity is real, and an agenda without hierarchy is unusable. The obligation is to make hierarchy visible. A proposer should know whether an issue was declined because it duplicated another session, lacked a multistakeholder format, exceeded the Forum's role, could not secure speakers, arrived late, or lost in a substantive priority choice.
Reasons also discipline selectors. "Too controversial" is not enough where controversy concerns a matter named in the mandate. "Already handled elsewhere" should identify the venue and explain why IGF dialogue would add no value. "Limited public evidence diversity" should distinguish a curable panel-design problem from a reason to exclude the issue itself. A reason code cannot replace judgment, but it can expose inconsistent judgment.
Emerging issues offered flexibility without a clear entrance test
The emerging-issues session helped prevent the four headings from closing the field completely. A young institution needs an open category because technology and public concern move faster than annual planning. The session recognised that the agenda makers' taxonomy could not be exhaustive.
Yet "emerging" can privilege novelty over unresolved power. Critical resource governance was not new in 2006; that was precisely why some entities wished to avoid it. An emerging-issues track can absorb fashionable technologies while long-standing structural disputes remain classified as old battles. Novelty then becomes another route around accountability.
The appropriate test is not age but institutional need. An issue deserves attention when it has new evidence, newly affected groups, changed authority, increased consequence or a gap among existing venues. A long-running problem can re-emerge because conditions change. A new technology may not require a global discussion if capable institutions already address it transparently.
Athens also showed that entities can create salience during the event. Expression became more prominent through workshops, campaigns and public response. This bottom-up correction is valuable, but it depends on resources. Organisations able to bring speakers, organise events and attract media can overcome a weak formal position. Less funded communities cannot rely on spontaneous attention.
Agenda legitimacy therefore cannot be delegated entirely to the marketplace of conference interest. Organisers should leave room for entity energy while correcting for predictable resource differences. Travel support, remote access, early notice, translation and assistance with session design are not ancillary. They determine which proposed issues become viable programme entities.
The absence of a negotiated outcome increased the importance of selection
The IGF was designed as a forum for dialogue, not a body that routinely adopts binding decisions. That restraint reduced the immediate stakes of any one panel. No Athens moderator could allocate address space or change the root-zone process. Entities could explore disagreement without bargaining over final text.
It would be a mistake to infer that agenda choices therefore had little power. In a non-binding institution, attention is the principal institutional currency. The Forum influences vocabulary, relationships, problem recognition, policy learning and the formation of coalitions. Selecting a subject can validate it as a global public-policy concern. Repeating a heading can create a professional community around it. Exclusion can leave an issue to institutions whose authority is itself contested.
Non-binding discussion can also travel. Officials return to ministries; executives return to companies; engineers return to standards and operational bodies; activists return to campaigns. They carry frames and contacts from the Forum into places where decisions are made. The IGF does not need to issue an order for its agenda to shape policy.
This makes selection evidence more, not less, important. A formal legislature records votes because the vote changes law. A forum should record agenda dispositions because attention is one of the main things it allocates. The public need not pretend that a session equals a mandate. It should be able to see why the institution devoted common attention to one issue and not another.
The distinction also protects the Forum from inflated claims. A well-attended main session does not establish global consensus. A dynamic coalition does not represent all stakeholders. A summing-up is not a negotiated resolution. Accurate labels let discussion influence others without being misrepresented as authority it did not acquire.
An agenda audit should reconstruct both the selected programme and its shadow
A credible audit of Athens would start with the mandate and create a list of substantive duties. It would then catalogue written contributions, consultation interventions and workshop proposals using the proposers' own language as far as possible. Each item would be assigned a public disposition: selected as a main theme; selected as a main session; merged; accepted as a workshop; redirected; deferred; withdrawn; ineligible; or not selected.
For merged proposals, the record should show where their central question survived. For rejections, it should give a concise reason. For deferrals, it should identify the review point. Panel and moderator selection should be included because a heading can remain while its range of positions narrows through speaker choice.
The audit should compare the people who proposed subjects with the people who selected and presented them. Geography, gender and stakeholder category are relevant but limited public evidence. Affiliation, funding capacity, language, institutional role and exposure to the policy consequence may reveal different concentrations. These data need privacy safeguards, but aggregate patterns can be published.
It should also examine the "shadow agenda": mandate items and strongly supported proposals that did not receive equivalent programme status. Critical Internet resources would be the clearest case. Development would be tested across sessions rather than accepted from the title. Concerns of everyday users would be traced to speakers and evidence. The audit would ask whether controversy, duplication, resource constraints or institutional deference best explains each absence.
Finally, it should compare the official programme with entity behaviour. Which workshops filled? Which questions repeatedly entered sessions despite not being scheduled? Which dynamic coalitions formed? Which omissions were corrected in Rio? Athens cannot be judged solely by what its planners predicted. The event itself generated evidence about demand.
Such an audit would not retry every 2006 judgment. It would show how judgments were made and what they produced. Historical accountability is valuable because the first agenda still influences institutional memory. It also offers a model for future cycles: publish a disposition record before the event, then a correction report afterward.
Transparency must preserve deliberation without preserving obscurity
Agenda makers will entity that candid discussion is difficult if every tentative compromise is attributed. That concern is legitimate. Members may need to say that a proposed session lacks balance, that a prominent speaker is unsuitable or that a confrontation could overwhelm the meeting. Immediate publication could turn programme design into positional theatre.
The answer is not to release every private exchange. It is to separate deliberative privacy from decisional traceability. The names attached to sensitive comments can remain protected while the institution publishes the criteria, material alternatives, final disposition and reason. Where members have conflicts because they are affiliated with a proposed organiser or an institution directly affected by a session, recusal or declared participation can be recorded without exposing confidential debate.
Aggregate voting is unnecessary if the group uses rough consensus. The chair should state how consensus was assessed, whether material objections remained and which matters were decided administratively. If no agreement was possible, the record should say who held final authority. "The programme emerged from consultation" is too vague when the actual closing act belongs to an appointed group, a chair, the Secretariat or the Secretary-General.
Selectors also need an appeal that fits conference time. Full adjudication would paralyse planning. A short correction window could allow proposers to identify factual errors, undeclared conflicts, inconsistent criteria or mistaken merging. The remedy might be reconsideration, a published explanation, an alternative session or priority in the next cycle. Not every disagreement requires reversal.
The goal is institutional memory. New Advisory Group members should not have to rediscover why a subject was excluded years earlier. Entities should be able to distinguish a mandate boundary from a logistical choice. Scholars and affected communities should not need personal access to former members to understand the programme.
Speaker selection was a second agenda inside the first
Headings did not determine the full range of Athens. The September Advisory Group meeting also drew up a shortlist of panelists, and moderators had to turn very broad themes into questions. This created a second selection layer after the choice of subjects.
A session called "openness" can be a debate about state censorship, corporate content control, intellectual property, open standards or access to public information. Which version appears depends on who receives the microphone, what expertise their biography is said to carry and which first question the moderator asks. Similar choices shape security. A panel dominated by law-enforcement and large-provider experience will define trust differently from one that includes incident responders, vulnerable users, privacy advocates and small networks.
Speaker diversity is not a cure by itself. A panel can contain several regions and sectors while all speakers share the same view of the disputed institution. It can include a person from a developing country whose professional role is remote from the costs being discussed. Conversely, a technically specialised panel may be appropriate for one question even if it does not reproduce the full meeting in miniature.
The audit requirement is therefore functional. For each principal session, organisers should identify the claims that need testing, the forms of knowledge required, the affected interests that cannot be inferred from expertise and the material positions left unfilled. A speaker matrix should distinguish geographic and stakeholder characteristics from the role a person performs in the debate.
Moderators deserve similar attention. They decide whether a question from the floor receives an answer, is reframed or is ruled outside the session. The later retrospective on critical Internet resources describes an Athens moment in which a question about political authority over the root and address space was redirected toward the broader conception of Internet governance. One intervention cannot establish a general pattern, but it illustrates how an omitted topic can appear from the floor and be excluded again through moderation.
This is why the public record should connect session design to mandate. A moderator may properly prevent one dispute from consuming a panel convened for another purpose. The session report should then record the unaddressed question and indicate where it can be pursued. Otherwise, time management becomes an invisible second veto.
Panel selection also determines whose absence is mistaken for consent. People without travel funding, visas, English fluency or institutional permission may not be available in Athens. Their absence should be recorded as an evidence limit, not interpreted as lack of concern. Remote contribution in 2006 was innovative but could not erase differences in connectivity and time zones.
The first programme's legitimacy consequently rests on three linked maps: subjects, formats and people. Publishing only the first map makes agenda formation look more inclusive than it may have been in practice. Publishing all three allows later readers to see whether a broad heading carried the conflict it promised.
The counterfactual was not a four-day negotiation over the root
Criticism of omission can produce an unhelpful binary. Either Athens should have used the programme it did, or the entire meeting should have been consumed by the governance of names and numbers. There were intermediate designs.
The Forum could have held a carefully bounded session on the public accountability of critical-resource institutions rather than a general contest over who should control them. It could have paired institutional incumbents with critics and required each to identify facts, authority and proposed remedy. It could have separated operational continuity from political oversight so that challenging one did not imply disrupting the other. It could have scheduled a listening session whose output was a map of disagreement rather than a recommendation.
Another option was a two-stage design. Athens could acknowledge the issue in a short main session, accept focused workshops on individual components and commission an open synthesis for Rio. This would recognise the mandate while preserving the first meeting's relationship-building objective. A visible deferral can itself be a legitimate institutional act when it specifies what evidence and preparation are needed next.
The existence of alternatives does not prove the Advisory Group chose wrongly. It shows why a decision record matters. Without the alternatives, later observers can only infer that inclusion was impossible or that the subject was overlooked. With them, the public can evaluate whether risk was proportionate and whether the compromise achieved what selectors expected.
Counterfactual design also prevents retrospective certainty. It is easy, after the Forum survived and Rio added the subject, to say Athens should have taken more risk. In May 2006, advisers faced a new body, limited preparation time and constituencies still carrying WSIS conflict. An audit should preserve that context. Accountability is strongest when it tests a choice against options available at the time rather than against knowledge acquired later.
The purpose is not to punish caution. It is to make caution specific. Which feared failure was being prevented? Walkouts, repetitive speeches, withdrawal of participation, inability to moderate, or confusion with operational authority? Different risks support different mitigations. A general fear of controversy gives incumbents too much agenda protection; an identified risk permits design.
Athens succeeded, but success is not an exemption from examination
Contemporary assessments were broadly positive. Mueller reported that most constituencies considered the experiment worthwhile, that focused workshops were energetic and that no group appeared to believe it had been wholly excluded. Around 1,350 entities attended according to a subsequent UN Commission on Science and Technology for Development report. New coalitions formed. Actors who had confronted one another during WSIS found a less rigid setting for exchange.
These achievements explain the caution of the agenda makers. A first meeting can fail by trying to settle the dispute that created it. The four-theme architecture offered breadth, comprehensibility and practical entry points. It proved that equal-status dialogue among stakeholder types could occur under UN convening without converting the meeting into an intergovernmental negotiation.
Success, however, can obscure its own costs. Because Athens worked, later accounts may treat the selected headings as obvious. They were not. They were the product of a compressed period in which a new appointed group translated hundreds of interventions and unresolved geopolitical conflict into a programme. Critical Internet resources did not disappear because they were outside scope. They disappeared from the main frame because agreement on including them could not be reached without risking the Forum's launch.
That distinction changes the lesson. The lesson is not that founding institutions should place every explosive dispute at the centre immediately. It is that compromise should leave a visible expiry date. If an issue is deferred to establish trust, say so. If it is moved to workshops, preserve the connection. If it is excluded because existing institutions entity, disclose that institutional concern in neutral terms. Then test the choice in the next cycle.
Rio's addition of critical Internet resources suggests learning occurred. But later correction does not remove the need to understand the first selection. The path from omission to inclusion is part of the Forum's legitimacy story. It shows that boundaries could move.
The first agenda's deepest legacy is the right to inspect the boundary
Athens gave Internet governance a forum that was useful precisely because it did not require every entity to accept one sovereign centre. Its agenda made that coexistence practical. Openness, security, diversity and access were capacious enough to reveal real disagreements while avoiding an immediate institutional showdown.
The same design also located power in a less visible place. The Forum would not decide who controlled the root or how addresses were allocated, but its appointed advisers and supporting Secretariat decided whether those questions would command the common stage. They did not control the Internet. They controlled the first shared account of what discussing its governance would look like.
That power should not be dramatised as censorship. The record does not support a claim that every omission was malicious or that the consultations were empty. It supports a more exact proposition: open submission and diverse composition did not by themselves make selection reviewable. The public could see the doors into the process more clearly than the gate where proposals changed status.
A mature multistakeholder institution should be able to show its rejected alternatives. Doing so does not weaken consensus. It distinguishes agreement from accommodation, capacity limits and unresolved objection. It lets groups that lost an agenda choice know whether to improve a proposal, seek another venue or return when conditions change.
The first IGF is remembered as a breakthrough in dialogue. It should also be remembered as an early demonstration that dialogue has an architecture. Main themes, session formats, speaker lists and summaries determine which conflicts become common knowledge. The right to participate begins before the meeting, with the ability to propose. It is completed only when entities can discover what happened to the proposal.
Athens needed a workable agenda more than it needed a perfect one. Twenty years later, that pragmatic truth is no reason to leave the selection obscure. The Forum's founding compromise becomes more legitimate when its limits are visible: who advised, who synthesised, who chose, what was omitted, why the omission seemed necessary and how the boundary later moved. The durable principle is not that every topic belongs on the main stage. It is that no founding agenda should become a silent constitution.
That principle also gives future organisers a practical discipline. Before repeating inherited headings, they should ask which conflict each heading made visible, which it domesticated, and which it left outside. A category that once enabled dialogue may later prevent it by treating new authority arrangements as variations of an old theme. The public record should show when organisers preserve a category for continuity and when they preserve it because reopening the taxonomy would disturb settled influence. Institutional memory is useful only when entities can inspect the choices that memory carries forward.