Summary
- The Geneva Declaration joined two positions without choosing a superior institution. It recognized the sovereign policy authority of states while preserving important technical and economic roles for the private sector and continuing roles for civil society and international bodies.
- The Declaration's requirements that international Internet management be multilateral, transparent, democratic and fully inclusive described qualities by which arrangements could be judged. They did not identify a final decision maker, voting rule, appeal path or operating mandate.
- Paragraph 50 of the Declaration and paragraph 13 of the Plan of Action sent the unresolved questions to a new Working Group on Internet Governance. Its assignments began with definition, issue identification and a common understanding of roles, showing that Geneva had not settled those matters.
- The Plan contained several specific national and developmental actions, including country-code domain responsibilities, Internet exchange centres and regional root-server promotion. Their specificity makes the missing global transfer more visible; none authorized WSIS or the United Nations to run Internet infrastructure.
- Geneva materially changed the politics of Internet governance by opening a United Nations-linked, multi-actor process and raising the standing of developing-country concerns. Its legacy should be described as agenda authority and procedural inclusion, not retrospectively inflated into operational sovereignty.
The settlement was designed to carry disagreement, not eliminate it
Diplomatic compromises often work by finding a proposition every party can endorse for a different reason. The result may be durable because it permits several institutional futures, not because it chooses one. Geneva's Internet-governance text belongs to that category.
One coalition wanted governments, especially governments outside the small circle historically closest to the Internet's coordinating institutions, to have a recognized place in international policy. Another defended private-sector leadership, distributed technical administration and institutions that had grown outside classical intergovernmental control. Civil-society entities sought meaningful participation, rights protections and public accountability without simply transferring power from private organizations to governments.
Technical operators were concerned that diplomatic redesign could destabilize working systems or place operational judgment in institutions without the necessary competence.
The final Geneva Declaration of Principles gave each camp language it could use. Internet governance became a core issue. International management should meet public qualities. States possessed policy authority in a defined field. The private sector's technical and economic contribution should continue. Civil society should continue its role. Intergovernmental and international organizations retained facilitating, standards and policy functions. A new inquiry would continue the debate.
What the settlement withheld was equally important. It did not say which institution could overrule another. It did not define the jurisdictional edge between public policy and technical administration. It did not establish whether stakeholder involvement meant consultation, equal participation, concurrence or merely an opportunity to speak. It did not direct a transfer of any root-zone, naming, numbering or standards function. The agreement became possible because recognition was distributed while supremacy was deferred.
WSIS was a development summit before it became an authority contest
The authority question can be misunderstood if Geneva is treated as a conference convened solely to redesign Internet coordination. The United Nations General Assembly had endorsed a much broader summit about an inclusive information society, digital development and access. The Declaration addressed human rights, infrastructure, education, security, cultural and linguistic diversity, media, financing and international cooperation. Internet governance occupied a small but unusually contested part of a document with global development ambitions.
That breadth affected the bargain. Many participating states approached the Internet not only as a technical system but as a condition for development, commerce, public administration and cultural participation. International connectivity costs, concentration of technical expertise, unequal access to policy forums and the location of key coordinating functions were therefore linked to a wider concern: countries expected to rely on a global facility without having equal capacity to shape its rules.
At the same time, the Internet was not waiting for a summit to make it function. Networks exchanged routes, registries issued number resources, operators ran name servers, standards communities developed protocols and businesses supplied connectivity. Existing institutions had procedures, contracts, staff and technical dependencies. A political text could criticize their legitimacy or call for reform, but operational authority could not be created merely by naming the Internet a global facility.
Geneva had to connect two institutional worlds. The summit's intergovernmental process could state global aspirations and invite states to act. Internet operations were dispersed across organizations and networks with different legal bases and constituencies. The Declaration did not fuse them. It built a vocabulary through which they could be debated together.
This is why the text should be read as a summit outcome rather than as an Internet charter. Its public importance is substantial. Its legal and operational precision is deliberately limited.
The conflict was not simply the United Nations against the Internet
Retelling Geneva as a contest between a centralized United Nations and a unified technical community erases the real coalitions. Governments did not hold one position. Private companies differed according to market, function and national setting. Civil-society groups did not uniformly prefer either state authority or the existing order. Technical expertise was distributed across companies, universities, public agencies, nonprofit bodies and individual operators.
Entity histories nonetheless identify a recognizable divide. Markus Kummer, who led the Internet-governance negotiating group and later served in the WGIG secretariat, described two broad schools of thought. One held that the existing, private-sector-led system worked and that problems should be defined before institutions were changed. The other questioned the legitimacy of existing arrangements and wanted a stronger governmental role through a multilateral mechanism linked to the United Nations. Kummer also recalled that the debate remained abstract because "public policy" was not specified and that Internet professionals were excluded from the final governmental negotiations.
Marc Furrer, Switzerland's secretary of state for the summit, offered a compatible but distinct entity account in Last Minute Diplomacy. He wrote that some delegations wanted all Internet-related issues addressed while others resisted discussion or change, particularly around critical resources. He said negotiators knew they could not settle all outstanding questions or ICANN's future role in Geneva. The late compromise was to create a working group that would analyze the subject and report to Tunis.
These accounts are valuable because their authors occupied the negotiation. They are not neutral transcripts. Each author had an institutional role and later interest in explaining the resulting process. Their strongest use is to illuminate the structure also visible in the adopted text: rival positions remained, and process replaced a final allocation of authority.
Paragraph 48 set performance criteria without naming the performer
Paragraph 48 of the Declaration contains the sentence most often treated as Geneva's constitutional core. It calls the Internet a global facility available to the public, places its governance on the information-society agenda and says its international management should be multilateral, transparent and democratic with the full involvement of governments, the private sector, civil society and international organizations. It adds goals of equitable resource distribution, access, stability, security and multilingualism.
The paragraph is politically ambitious. It rejects the proposition that international Internet management is the concern of a closed technical or commercial circle. It gives developing-country demands for participation a globally negotiated reference. It identifies stability and equitable access as public concerns rather than private side effects.
But the grammar does not establish an office. "Should be" describes how management ought to be conducted; it does not say who may conduct it. The listed qualities can be applied to existing institutions, a reformed arrangement or a future body. "Full involvement" names entities but not the consequence of their participation. "Multilateral" can describe cooperation among states, while the same sentence extends involvement beyond states. "Democratic" is not accompanied by an electorate, representation formula or decision threshold.
The paragraph also speaks of "international management" without specifying whether that means root-zone authorization, standards, address policy, cybersecurity cooperation, commercial interconnection, content rules or all of them. Its goals cross several institutions and legal domains. No single performer could satisfy them through one command.
Paragraph 48 therefore creates normative leverage, not a deed of transfer. A government, company or organization could be challenged to explain whether its conduct was transparent or inclusive. None could point to the paragraph alone as proof that WSIS had placed an operational function under its control.
"International management" concealed several unlike powers
The phrase "management of the Internet" sounds singular. The underlying powers are not. Domain-name policy, root-zone changes, address distribution, route selection, protocol parameter assignment, standards development, interconnection contracts, cybersecurity enforcement and content regulation involve different actors, instruments and failure modes.
A sovereign state can legislate for providers and users within its jurisdiction. It can pursue crime, protect consumers, regulate communications markets and set public-sector procurement. Those powers do not give it the ability to compel every network to accept a route or every resolver to use a particular root. A registry can maintain authoritative records within its delegated scope without having authority over speech or competition. A standards body can publish an interoperable protocol without compelling deployment. A network operator can choose routes and peers without legislating for the public.
Geneva did not inventory these powers. It joined them under a broad term at the moment when the lack of a common definition was itself part of the dispute. This compression was useful for diplomacy. Parties could agree that "management" should be legitimate without agreeing which decisions belonged inside the term.
Compression becomes hazardous when later readers infer a unified sovereign function. If management is imagined as one entity, a declaration about its qualities can look like authorization for a single manager. The text supports no such conclusion. Its next paragraph immediately distributes roles, and its following paragraph sends foundational questions to a working group.
A sound authority reading must therefore disaggregate the system. For every claimed Geneva mandate, ask: Which act? Which resource? Which institution previously performed it? What instrument moved the power? What limits and review accompanied the move? The 2003 documents rarely answer those questions because their central task was to make continued discussion possible.
Paragraph 49 allocated descriptions, not a hierarchy
Paragraph 49 recognized that Internet management includes technical and public-policy issues and should involve stakeholders and relevant intergovernmental and international organizations. It then listed five role statements. The list is often read as a settlement of "respective roles." It is more accurately a map of claims that negotiators could jointly tolerate.
The first statement gives states policy authority for Internet-related public-policy issues as a sovereign right and says states have rights and responsibilities in international Internet-related public policy. The second says the private sector has had and should continue an important technical and economic role. The third continues civil society's role, especially at community level. The fourth gives intergovernmental organizations a facilitating role in coordinating public policy. The fifth continues the role of international organizations in technical standards and relevant policies.
No ranking rule explains what happens when the categories overlap. A security standard can have public-policy consequences. A domain-name decision can be technical, economic and expressive at once. Private infrastructure choices can affect access. Government regulation can change technical architecture. International organizations may develop policies that states later implement. Civil-society evidence can alter the legitimacy of every layer.
The verbs also differ. States possess "policy authority" in a stated domain. The private sector, civil society and international organizations have "important" roles. Intergovernmental organizations "facilitate." These distinctions matter, but they do not form a complete chain of command. The text never says that state policy authority includes control over a function another institution operates, nor that an important technical role immunizes its holder from public law.
Paragraph 49 is consequently a non-exclusion arrangement. Each category is present, and no category receives a universal title. It postpones the difficult work of drawing the boundaries among them.
Sovereignty was affirmed over policy, not converted into global custody
The state clause carried obvious political weight. For governments that believed international arrangements underrepresented them, an express sovereign-right formulation was significant. It prevented the new multistakeholder vocabulary from being used to say that governments had no legitimate Internet-policy role.
The entity of the sentence is equally important: "Internet-related public policy issues." It is not "the Internet," the root, all domain names, address space or every technical operation. Sovereignty remains exercised through jurisdiction, international law, treaties, domestic legislation and the powers of particular public institutions. A summit declaration does not erase those forms.
The second sentence extends rights and responsibilities to international Internet-related public policy. It does not state how those rights are coordinated when jurisdictions conflict or when an operational institution is incorporated in one country but serves users worldwide. That unsolved coordination question was one reason the debate continued.
Nor does the clause establish that every technical decision is public policy. Some decisions implement a settled rule or keep a service running. Others embed distributional choices and deserve policy scrutiny. Geneva recognized that management contains both technical and public-policy issues but supplied no test for the border. Calling an action technical cannot automatically remove it from public accountability; calling it policy cannot automatically give a government the credentials or authority to execute it.
The defensible interpretation is therefore positive and bounded. States were not visitors to Internet governance. They possessed genuine policy authority. The clause did not assign any one state, all states collectively or a United Nations body custody of global Internet operations.
The private-sector sentence preserved contribution without granting title
The private-sector provision also carried two meanings. Historically, companies and private nonprofit organizations had built networks, supplied services, developed equipment and software, operated registries and participated in coordination. The sentence acknowledged that technical and economic reality and said the role should continue. It guarded against an interpretation in which intergovernmental recognition automatically displaced existing expertise and investment.
An important role is not ownership of the public consequences. A company may run infrastructure under contract, delegation or market choice. It may develop a standard or technical service on which others depend. That role can be indispensable while remaining subject to competition, transparency, security, contractual and legal controls. The Declaration did not say private operation was beyond public scrutiny.
Nor did it define "private sector" as one constituency capable of receiving a single mandate. A backbone carrier, domain registry, small access provider, software developer and industry association possess different powers. Some compete; others depend on one another. A statement preserving their collective role does not identify which firm may decide what.
The sentence was thus symmetrical with the state clause in one important respect. It protected standing without resolving boundaries. States could not exclude private technical competence. Private actors could not treat historical operation as a declaration of permanent sovereignty. Both could sign because the document did not force either to surrender its central claim.
That ambiguity was not an accidental drafting defect. It was the mechanism of agreement.
Civil society was included without being made a decorative electorate
Geneva said civil society had played and should continue to play an important role, especially at community level. This recognition mattered in a summit where formal adoption remained intergovernmental and where non-state participation varied across preparatory stages. It made it harder to describe Internet governance as a bargain only between governments and companies.
Yet the community-level qualifier could narrow as well as recognize. Civil-society organizations addressed global human rights, access, media, gender, development, knowledge and institutional accountability. Confining their legitimacy to local implementation would miss the transnational consequences they were contesting. The phrase records a compromise, not a complete theory of civil-society authority.
Civil society also lacks a single authorization chain. An advocacy organization may present expertise, affected-person testimony or a membership position. It does not represent every user. Open participation can improve evidence and contestation without converting attendees into a global electorate. Geneva did not specify selection, balance, disclosure or accountability rules that would make "full involvement" representative in a formal sense.
This limit does not reduce participation to ceremony. Civil society can expose rights effects, organize communities, develop proposals and hold both public and private institutions to account. Its authority often lies in evidence, association, public reasoning and the ability to make excluded consequences visible. Those forms differ from operating custody and state coercion.
The text's lasting value is that it requires a seat for such contribution. Its silence is that it does not say when that contribution must change a decision. That question cannot be answered by counting stakeholder labels in a room.
Intergovernmental and international organizations were not merged into one sovereign
Paragraph 49 separately described intergovernmental organizations and international organizations. The former had a facilitating role in coordinating Internet-related public-policy issues. The latter had an important role in technical standards and relevant policies. The distinction resisted the idea that every body with an international name performed the same function.
Facilitation is not command. An intergovernmental organization can convene states, provide a treaty venue, collect evidence, support capacity and coordinate policy. Its powers depend on its constituent instrument and decisions of its members. The Geneva text did not enlarge every organization's competence to include Internet operations.
Standards and policy work also varies. An international standards organization can publish technical specifications through its own procedures. Adoption may be voluntary, contractual or incorporated into law. The specification's global relevance does not make the organization a legislature for every use of the technology.
Most importantly, the United Nations Secretary-General received a specific procedural assignment in paragraph 50: establish a working group. The Secretary-General did not receive a general instruction to manage the Internet. The precision of that assignment should control interpretation. Where negotiators intended an institutional act, they named it.
Later advocacy sometimes treats "the UN" as though Geneva transferred a unified bundle of powers to a single institution. The Declaration supports a much narrower account. It recognized several kinds of international contribution and used the Secretary-General to convene further inquiry. It did not make the United Nations the operator, standards authority, registry or appellate court of the Internet.
Paragraph 50 openly admitted that the conceptual work remained undone
After the role list, paragraph 50 asked the United Nations Secretary-General to establish a working group through an open and inclusive process with full and active participation from governments, the private sector and civil society in developed and developing countries, involving relevant intergovernmental and international organizations and forums. The group was to investigate and make proposals for action, as appropriate, by 2005.
The Geneva Plan of Action made the assignment more revealing. The group was to develop a working definition of Internet governance, identify relevant public-policy issues, develop a common understanding of the roles and responsibilities of governments, existing organizations, other forums, the private sector and civil society, and prepare a report for consideration and appropriate action in Tunis.
These are preliminary tasks. A body asked to define the field, identify its issues and build a common understanding of roles is not implementing a settled allocation of authority. It is producing the knowledge on which a later negotiation might act.
The mandate also contained layers of institutional caution. The group would "investigate" and "make proposals." Its report would be "presented for consideration." Action would occur "as appropriate." None of those verbs empowered the group to replace an operator or bind a government, company or technical organization.
This sequence is the clearest answer to claims that Geneva had already decided who governed. If it had, a working definition and common understanding of responsibilities would not have been the first assignments. Paragraph 50 is not a footnote to a completed settlement. It is the settlement's mechanism for carrying the unresolved authority question into phase two.
The working group was an authorization to inquire, not to rule
Creating the Working Group on Internet Governance changed the institutional landscape. A disputed subject that some delegations had resisted placing on the agenda now had a named process, a deadline and an expectation of multi-actor participation. That was real agenda power.
Agenda power should not be confused with decisional power. The working group could define, map and recommend. It could give developing-country concerns and non-state expertise a venue. It could compare existing arrangements and propose institutional models. The summit reserved later consideration to the Tunis phase.
Kummer's entity account explicitly described the group as non-negotiating and as preparation for future negotiations. That description aligns with the adopted verbs. It also explains why the compromise attracted both camps. Supporters of greater governmental involvement obtained a United Nations-linked route to reconsider international arrangements. Defenders of distributed and private-sector-led mechanisms obtained an inclusive inquiry that would define problems before imposing solutions.
The distinction protects the credibility of the process. A study group can be broad, exploratory and candid because its proposals do not themselves seize a function. A decision body needs rules for authority, conflicts, review, implementation and responsibility for failure. Geneva supplied the former, not the latter.
Calling WGIG's later report evidence of what Geneva "authorized" therefore reverses time. Its definition and models were outputs produced under the inquiry. They can inform the meaning of the subsequent Tunis settlement. They cannot be inserted backward into the 2003 text as powers already granted.
The Plan's specific actions reveal the absence of a global transfer
The Plan of Action did more than create WGIG. It invited governments to facilitate national and regional Internet exchange centres, to manage or supervise their respective country-code top-level domains as appropriate, and to promote Internet awareness. In cooperation with stakeholders, it encouraged regional root servers and internationalized domain names. It called for participation by developing countries in international ICT forums and for open, interoperable and non-discriminatory standards.
These clauses matter because they show the drafters could state concrete actions. They named governments, identified national or regional scope and used operationally adjacent subjects. The country-code clause did not say every government must operate a registry; "as appropriate" recognized different domestic arrangements. It nevertheless addressed a specific national interest.
The exchange-centre clause encouraged infrastructure that could lower cost and improve local traffic exchange. It did not authorize a summit body to direct commercial peering. The regional-root-server clause concerned reach and resilience through cooperation. A root-server instance does not confer authority to change root-zone contents. The internationalized-name clause promoted access and linguistic inclusion; it did not assign approval power over the global naming system.
The contrast is instructive. Where Geneva intended governments to undertake bounded actions, it said so. Nowhere did it direct the United Nations, ITU, all states collectively or a new body to assume the IANA functions, authorize root changes, distribute global address blocks or approve protocol standards.
One should not infer a major transfer from the broad Declaration when the more concrete Plan stopped short of it.
Country-code language recognized national interest without settling the delegation chain
The invitation that governments manage or supervise their respective country-code top-level domains is the closest Geneva came to an explicit institutional role in naming operations. It deserves precision rather than either exaggeration or dismissal.
Country-code domains are associated with territories, and governments can have legitimate public interests in their continuity, security and administration. Actual arrangements differ. Some registries are governmental, some academic, some nonprofit and some private. Delegation histories, local communities, contracts, technical competence and applicable law all matter.
"Manage or supervise, as appropriate" accommodates that variation. It does not prescribe one organizational form or erase existing delegation relationships. It also says "their respective" country-code domain, not the root as a whole or another country's domain. The scope is deliberately national and bounded.
This clause therefore cannot support a claim that Geneva authorized collective governmental control of all domain-name operations. Nor should it be read as proof that governments have no role. It is an example of differentiated authority: legitimate national interests, local institutional arrangements and continued need for global coordination.
The authority question remained at the interfaces. Who determines when a delegation serves the relevant community? What happens when a government and incumbent registry disagree? Which process evaluates technical competence and continuity? How does global coordination respect national interests without making cross-border naming hostage to diplomatic bargaining? Geneva did not establish the appeal or enforcement structure needed to answer those questions.
The clause marks a domain of responsibility. It does not complete the chain.
Promoting root servers was not authority to alter the root
The Plan's reference to regional root servers is especially easy to misuse because "root" can signify both service distribution and policy authority. Root-server instances answer queries using the authoritative root-zone data. Adding geographically distributed instances can improve latency, resilience and local reach. It does not, by itself, change who authorizes additions or modifications to the zone.
Geneva encouraged promotion of regional servers in cooperation with relevant stakeholders and linked the action to overcoming access barriers. The language concerns deployment and availability. It contains no procedure for changing top-level-domain entries, no transfer of signing authority and no power to instruct root operators to serve different contents.
This technical distinction carries a governance lesson. Physical or logical participation in service delivery does not necessarily include policy control over the service's authoritative content. A country can host an instance without acquiring a vote over every root change. Conversely, a central authorization function can affect users in regions where the authorizing institution has little representation.
The summit's inclusive principles created a reason to scrutinize that mismatch. They did not resolve it. Any later proposal for root-zone reform had to identify the actual authorization function, the legal and technical instruments governing it, continuity safeguards and the institution that would assume responsibility. Geneva's server-promotion clause cannot do that work.
Treating distribution as sovereignty would make the text appear more decisive than it was and obscure the real locus of DNS delegation power.
Silence preserved the practical status quo without blessing it forever
When a negotiation declines to transfer an existing function, the current operator normally continues. That is a practical consequence of institutional continuity. It is not the same as a permanent endorsement.
After Geneva, networks still needed names, numbers, routes and protocols to work. Existing institutions continued because no replacement had been authorized and because disruption would have imposed immediate costs. The Declaration's private-sector language and concern for stable, secure functioning reinforced caution toward abrupt change. At the same time, the decision to investigate governance and consider proposals in Tunis made clear that arrangements were open to review.
This dual effect is central to the compromise. Defenders of existing systems avoided an immediate transfer. Critics gained a recognized path for challenge. The status quo survived as an operation, not as an untouchable constitutional order.
Later histories sometimes describe Geneva as either a defeat for governmental reform or a founding victory for multistakeholder governance. Both compress a two-stage bargain. No government-led replacement was created. No final rule insulated private or technical institutions from later international scrutiny. The parties agreed to defer the choice while preserving service.
The distinction also clarifies burden of proof. An institution that continued an existing function could point to prior authority and continuity needs, not to Geneva as a new grant. A reform advocate had to propose a credible transfer, not merely cite dissatisfaction with representation. Stability and legitimacy remained separate tests.
"Multilateral" and "multistakeholder" were not interchangeable
The Geneva text combined terms from different traditions. "Multilateral" usually emphasizes relations among states. The role list and participation clauses brought in private business, civil society and international bodies. Later discourse increasingly used "multistakeholder" to describe that wider arrangement, but the 2003 compromise did not supply a single institutional definition.
This ambiguity was productive. Governments could read multilateralism as recognition that international public policy could not be left to one state or private organization. Non-state actors could read full involvement as protection against an intergovernmental monopoly. Both readings opposed unilateral control while disagreeing about the decision structure that should replace it.
The terms should therefore not be collapsed into proof of equal voting power. States negotiated and adopted the summit text. Non-state actors contributed through consultative and participatory channels whose openness varied. The working-group mandate promised full and active participation, but it did not state that every stakeholder category could veto the report or the Tunis outcome.
Nor does multilateralism alone establish democratic legitimacy. A room containing many governments may still underrepresent affected communities, and equal state votes do not equal equal population representation. A multistakeholder meeting may broaden expertise while leaving selection, resources and accountability unclear.
Geneva named the aspiration before designing the mechanism. That was enough to constrain later argument: exclusion needed justification, and no single constituency could plausibly describe itself as the whole Internet. It was not enough to determine final authority.
Transparency and democracy were standards of criticism, not operating instructions
The words "transparent" and "democratic" gave reform advocates a durable benchmark. Closed decisions, unreasoned delegations and institutions inaccessible to developing countries could be measured against a globally agreed expectation. The words also signaled that technical effectiveness alone was not a complete legitimacy claim.
But neither word implements itself. Transparency requires an entity: published proposals, reasons, decision makers, affiliations, records, contracts, performance or appeal outcomes. Democracy requires a constituency and a method of authorization. A standards community, company, state agency and intergovernmental assembly will answer those questions differently.
The Geneva Declaration did not choose among direct election, state representation, membership governance, open consensus, corporate accountability or judicial review. It did not say which decisions should be public and which operational or security details could remain restricted. It did not specify remedies when an institution failed the benchmark.
That incompleteness should not make the principles empty. They can discipline institutional design when translated into concrete controls. A registry can publish criteria and reasons. A government can consult affected operators and provide review. A standards body can keep participation and records open. An international process can support developing-country participation and disclose how input altered text.
What the principles cannot do is substitute for the translation. An actor cannot claim authority merely by calling its own process transparent and democratic. Geneva made those qualities relevant; it did not certify any institution as their exclusive embodiment.
Full involvement was an access promise without a decision rule
"Full involvement" sounds stronger than consultation. It should be taken seriously. A process cannot claim Geneva's inclusive legitimacy if key affected groups receive documents after decisions are fixed, cannot submit proposals, lack access to deliberation or are invited only to endorse a prepared result.
Yet involvement still has stages. A entity can receive notice, submit evidence, propose text, deliberate, entity, vote, appeal, implement or review. Institutions often provide the early stages while reserving the later ones. Without a declared decision rule, observers may mistake access for shared authority.
Geneva did not specify the minimum consequence of involvement. The new working group was to devise an open and inclusive process, but the Secretary-General retained responsibility for establishing it and states retained control of the summit outcome. Companies and civil-society organizations brought expertise and positions without becoming treaty parties.
The practical legitimacy test should therefore ask what participation could change. Were objections answered? Did affected actors see revisions? Could less-resourced entities remain engaged? Were conflicts disclosed? Who decided when consensus existed? Who could correct a factual error or challenge an unauthorized act?
These questions follow from the principle but are not answered by it. They also show why inclusive vocabulary could reconcile rival camps. The text promised presence while leaving the conversion of presence into power for later institutional design.
No Geneva clause authorized a new Internet operator
An operational mandate needs more than a principle. It needs an identified function, an authorized performer, scope, effective date, assets or access, service obligations, security controls, funding, liability, review and continuity arrangements. A global technical service also needs cooperation from the networks and institutions that depend on its outputs.
The Geneva documents contain none of that for a new global operator. They do not establish a legal entity to run naming or numbering. They do not direct incumbents to hand over credentials, records, contracts or facilities. They do not require resolvers, registries, address holders or network operators to recognize a new authority. They do not define service levels or emergency powers.
The absence is not surprising. A summit declaration and action plan were not designed as transfer instruments. The problem arises only when later rhetoric treats them as such.
Even the strongest institutional language is bounded. The Secretary-General was asked to establish a working group. Governments were invited to take certain national actions. ITU's spectrum and development competencies were recognized elsewhere in the Plan. None of those clauses silently expanded into general custody of Internet operations.
This is also why the stability objective cannot be read as a blank cheque. Many actors contribute to stability. A declaration that stability should be ensured does not authorize whichever institution invokes the word to overrule the others. Authority must come from the applicable legal, contractual, technical or membership instrument.
Geneva set the question. It did not appoint the final operator.
The compromise did not turn existing operators into constitutional sovereigns
Refusing to infer a new United Nations mandate does not require treating existing technical organizations as natural rulers. Their authority also rests on specific instruments, communities and dependencies. A registry's delegation can be reviewed. A company's contract can expire. A standards body's influence depends on implementation. A network's routing choice affects only the paths others accept.
The Geneva principles placed those arrangements under public scrutiny because the Internet had become a global facility. Transparency, equitable participation, stability, security and multilingual access became legitimate questions. Existing institutions could respond through reform, internationalization, public reporting and broader participation.
But reform is not the same as retroactive authorization. If an institution improved after Geneva, the improvement may answer a principle without proving the summit had granted it exclusive authority. If another institution claimed to embody multilateral legitimacy, that claim still required competence and a valid mandate for the function.
This symmetrical discipline prevents the debate from becoming a choice between two myths: inherent private sovereignty and inherent intergovernmental sovereignty. Internet authority is fragmented and contingent. Each function must be justified by the instrument and community that sustain it, with public consequences tested against law and accountable process.
Geneva's role was to make that justification globally contestable. Its failure to choose a sovereign can be understood as institutional realism rather than evasion alone. No proposed single authority matched the Internet's distributed operation or the political diversity of its users.
Later definitions cannot be inserted into the Geneva text
In 2005, the WGIG report proposed a working definition covering the development and application of shared principles, norms, rules, decision-making procedures and programmes by governments, the private sector and civil society in their respective roles. The Tunis Agenda later adopted that definition and created the Internet Governance Forum while separately addressing enhanced cooperation.
Those developments clarify the path after Geneva. They do not change what was agreed in 2003. The Geneva Plan explicitly requested a working definition because one did not yet exist. Reading the later definition backward would erase the very task the first phase created.
The same applies to later institutional language. Tunis provisions, subsequent General Assembly renewals, the Global Digital Compact and the 2025 decision to make the IGF permanent have their own dates, entities and authority. They may extend or refine institutions. They cannot prove that Geneva had already granted those powers.
Historical accuracy requires versioned interpretation. Ask what the text said at adoption, what institutional arrangements then existed and which later instrument changed them. Political narratives often prefer an unbroken founding story. Authority analysis requires a dated chain.
Geneva's chain ends with principles, role recognition and a request for inquiry. That is enough to explain its influence. It is not enough to support a claim of completed constitutional settlement.
A paragraph-by-paragraph authority audit produces a narrow but durable result
An authority audit of the core text can be stated compactly.
Paragraph 20 recognizes joint responsibility and stakeholder roles across the information society, with participation in decisions where appropriate. It creates no global Internet office. Paragraph 48 establishes public qualities and goals for international management. It names no final decision maker. Paragraph 49 recognizes technical and policy dimensions and allocates broad roles. It supplies no hierarchy for overlap. Paragraph 50 instructs the Secretary-General to establish an inclusive working group to investigate and propose action. It authorizes inquiry, not operation.
The Plan's paragraph 13 gives the working group four preparatory assignments and gives governments several specific national and developmental invitations. It does not direct a global transfer. Other action lines address standards, security, development and international cooperation within their own scopes.
The result is neither trivial nor sovereign. Geneva created an agreed vocabulary, a legitimacy test and a process. It preserved continuity while making future reform discussable. It expanded the standing of states and non-state actors without defining equal power. It did not decide who could issue a binding global policy, supervise existing institutions or run daily technical functions.
This narrow reading is durable because it follows the verbs rather than the mythology. It allows Geneva to matter without making the text perform work it never did.
The authority question should always be restated as a function
"Who governs the Internet?" invites a misleading singular answer. Geneva showed why the better question is functional.
Who can enact a rule binding a domestic provider? A competent public authority under applicable law. Who can publish a protocol specification? A standards process under its own procedures. Who can modify an authoritative naming record? The actors authorized within that delegation and service chain. Who can choose a route? Networks and their counterparties, subject to technical, contractual and legal constraints. Who can investigate crime? Public authorities with jurisdiction and cooperation arrangements. Who can persuade institutions to change? Any entity with evidence and access, though persuasion is not command.
Once the functions are separated, the Geneva compromise looks less mysterious. States obtained recognition where public policy was genuinely at issue. Private and technical actors retained roles where operation and development depended on them. Civil society and international bodies remained necessary to evidence, coordination, standards and legitimacy. No universal hierarchy was available because the functions did not share one source of authority.
The useful legacy is therefore a requirement of institutional specificity. An actor citing Geneva should name the act it seeks to perform, the paragraph that supports its participation, and the separate instrument that supplies actual power. If it cannot, it has a policy argument, not an authorization.
The compromise succeeded because it refused a false finality
Geneva did not answer who had final power because there was no single power to award and no coalition able to define the boundaries of the powers in dispute. The summit instead recognized multiple roles, declared public principles and created a route toward further negotiation.
The bargain was vulnerable to strategic interpretation. Governments could emphasize sovereign policy authority. Companies and technical institutions could emphasize continuity of the private sector's role. Civil society could emphasize full involvement. International bodies could emphasize multilateral coordination. Each reading was grounded in text; none exhausted it.
That openness created the danger of consensus capture, in which one camp presents a shared phrase as endorsement of its preferred institutional design. The remedy is not to dismiss the phrase. It is to read it with the neighboring clauses, the verbs, the omissions and the next procedural step.
Geneva's principles did not authorize Internet operations. They made those operations a legitimate subject of global public inquiry. They did not install an intergovernmental sovereign. They denied any existing actor an easy claim that the international public had no standing. They did not settle stakeholder equality. They made exclusion harder to justify.
The compromise's importance lies in that exact balance. It opened the authority question without pretending to close it. Any institution claiming more must identify the later, specific and valid act that supplied the missing power. The first phase of WSIS should be remembered not as the moment the world chose an Internet governor, but as the moment it agreed that the location and legitimacy of Internet authority could no longer remain outside public argument.