When Non-Endorsement Is Not Enough: CAIGA, ICP-2, and the Risk of “Legitimacy Transfer”

Part One argued that the ICANN—Smart Africa—CAIGA episode should not be judged only through the narrow lens of authorship, funding, or endorsement, and here is a track record between ICANN82 and ICANN84. ICANN’s later clarification may define the limits of its intent, involvement, review, and endorsement, but it does not resolve the deeper institutional question: when an ICANN-supported regional process began touching the governance of an independent Regional Internet Registry, did ICANN warn, correct, and record the boundary early enough?

The issue is not whether ICANN should engage Africa, it should. The issue is whether ICANN’s engagement was bounded before its presence could be read as “implied legitimacy” for a process affecting AFRINIC governance.

This second part turns to the next layer of the problem: non-endorsement. The absence of formal endorsement does not necessarily mean the absence of institutional effect, especially when the project is promoted as a “partnership,” and the logos of entities are visible. A proposal can gain authority through proximity, funding, public presentation, expert presence, and repeated association before any institution signs its name to it. That is why CAIGA must be examined not only as a text, but as a process that moved through an ICANN-supported environment (ICANN meetings and Sidelines) while ICANN positioned itself as a future evaluator rather than a contemporaneous boundary-setter.

Why does the absence of formal endorsement not erase institutional effect

ICANN’s reliance on non-endorsement is institutionally important, but it is not dispositive.

In its response to the Non-Commercial Stakeholder Group, ICANN made several clarifications: it did not enter the Smart Africa MoU or Project Agreement intending the Internet Governance Blueprint to propose a new governance model for AFRINIC; it did not request Smart Africa to deliver AFRINIC governance proposals; it did not pre-evaluate CAIGA; and it had not endorsed CAIGA or any other model for the governance of an RIR.¹

Those clarifications matter as they narrow the authorship question, establish institutional distance from the final content, and they tell the community that ICANN does not regard ICANN funding as automatic endorsement.

But they do not settle the deeper issue, endorsement is not the only way legitimacy travels.

In political governance, authority often travels through channels softer than command: visibility, convening power, funding, procedural access, repeated attendance, administrative support, expert presence, diplomatic proximity, and the use of an institution’s name beside a proposed reform. Political science would recognize this as agenda-setting, legitimacy transfer, and institutional signaling. Political economy would add that funding and access shape the field of possible choices long before any formal decision is made. These forms of influence do not announce themselves as control, they appear as association. Yet association is often enough to alter the political meaning of a process.

Facilitation Without Endorsement.

Facilitation without endorsement occurs when an institution does not formally approve, adopt, or sign a proposal, but still enables the environment through which that proposal acquires credibility, circulation, and seriousness. It is the governance equivalent of saying: we did not bless the outcome, but we funded the pathway, sat near the process, and allowed the proposal to mature under the shade of our institutional presence.

That distinction is especially important in Internet Governance because legitimacy is often reputational before it is legal. A proposal does not need formal approval to become powerful. It may become powerful because stakeholders believe it is already acceptable to institutions whose names carry global technical authority, regional development legitimacy, or state-backed political weight.

Smart Africa’s role matters in this respect because it is not an ordinary stakeholder association; it is a heads-of-state and ministerial platform capable of giving a proposal the appearance of continental political direction. When that political weight is placed beside ICANN’s global technical authority and GIZ’s development legitimacy, the result is not ordinary association. It is a compound legitimacy effect, one that can make a proposal appear institutionally mature before the affected technical community has actually authorized it.

This is why (non-endorsement) is a weak shield when the issue is boundary maintenance. ICANN can correctly state that it did not pre-endorse CAIGA. However, the right question remains: did ICANN’s funding, participation, and meeting-space association make CAIGA appear more legitimate than it had yet become through AFRINIC community consent?

In any governance system, perceived legitimacy can begin changing behavior long before formal approval exists. Once a proposal appears to have institutional backing, governments may treat it as diplomatically alive, consultants may refine it as a viable governance option, and regional bodies may present it as emerging consensus. Operators, seeing global institutions nearby, may begin to calculate whether resistance is already too late. In the Internet number resource system, the effect is even more sensitive because registry governance is not symbolic; it affects resource-holder rights, operational continuity, routing trust, registry data integrity, and the predictability of global coordination. By the time affected communities are invited into consultation, they may no longer be shaping an idea at birth; they may be defending against an architecture that has already acquired prestige.

This is how endorsement can be avoided while legitimacy is still transferred. ICANN may later clarify that it did not endorse the proposal; the partner may insist it led the process; the funder may describe its role as support for broader dialogue; and the affected community may be told that consultation remains available. Formally, each explanation may be defensible. Institutionally, however, the damage has already begun: the proposal has been socialized, presented, framed, and surrounded by credibility before the community has fully authored or accepted it.

That is the flaw. Non-endorsement may shield ICANN from the question of formal approval, but it does not erase responsibility for the legitimacy effects of facilitation. In a system where trust is built through institutional signals, the absence of endorsement is not the same as the absence of consequence.

GIZ’s role was not merely invisible background funding. At minimum, GIZ was operationally embedded in the forum infrastructure as data controller for the SmartAfrica Session on the sides of Africa Internet Summit in Ghana 2025, while formally denying authorship, steering, or decision authority over CAIGA.

The Proposal Was Already Moving

ICANN’s response stated that, as currently set out in the Internet Governance Blueprint, it was “not clear” to ICANN that the CAIGA proposal met the ICP-2 test. ICANN added that if any proposal to modify AFRINIC’s governance structure came to ICANN, it would evaluate that proposal neutrally and objectively, and that CAIGA would receive no priority or exemption merely because it emerged from a project that used ICANN funding.²

At first glance, that sounds procedurally safe. ICANN is saying: we have not endorsed CAIGA; if it comes formally, we will evaluate it under ICP-2.

But this answer is too late in the life of the proposal.

CAIGA was not lying dormant in a drawer, waiting to be submitted to ICANN for first review. It had already appeared inside a Smart Africa Blueprint funded through an ICANN-Smart Africa project/mOu agreement. It had already been presented at ICANN84. ICANN itself acknowledged that many in the community first learned of CAIGA and the broader Blueprint during Smart Africa presentations at ICANN’s 84th Public Meeting in Dublin.³

This is where the problem sharpens.

If the community first learned of the proposal at an ICANN public meeting, then ICANN cannot treat the matter only as a future submission. By then, the proposal had already entered public institutional space. It had already acquired visibility through an ICANN-adjacent platform. It had already been placed before the community not as a hypothetical but as an emerging architecture.

A proposal does not become consequential only when it is formally deposited at ICANN’s door. In governance, consequence begins earlier. It begins when a proposal is drafted under a funded project, placed in a regional blueprint, presented in an international meeting, described as the product of joint work, and linked to a reform process affecting a weakened institution.

That is why the right question is not purely: Will ICANN evaluate CAIGA if it arrives?

The question is: Why did ICANN wait for arrival when the proposal was already traveling under the shadow of ICANN’s association?

This is the weakness in the “future evaluation” posture. It imagines governance as a clean filing process: someone prepares a proposal, submits it, ICANN reviews it. But real institutional legitimacy does not work so neatly. By the time a proposal is formally submitted, the political and procedural terrain may already have been shaped. Expectations may already have formed. Stakeholders may already have aligned. Opponents may already be framed as obstructionists. The room may have already decided what the file has not yet declared.

In such circumstances, future evaluation is not enough. What was needed was the contemporaneous boundary correction.

Referral to ICP-2 Is Not the Same as Warning

ICANN’s response states that its coordination with Smart Africa was focused more broadly on Internet governance in Africa, not on developing CAIGA. It says Smart Africa and its consultant included CAIGA as part of the broader Blueprint produced through the ICANN-Smart Africa Project Agreement. ICANN further says that when it was approached to provide substantive input to guide CAIGA, it declined and referred Smart Africa only to publicly available documentation, including ICP-2.⁴

This passage is important because it confirms three things at once.

First, CAIGA was embedded in the Blueprint.

Second, the Blueprint was produced through the ICANN-Smart Africa project framework.

Third, ICANN was approached for substantive input on CAIGA and understood that CAIGA raised RIR-governance questions.

That should have been the boundary moment.

It is not enough to say, “we declined to provide substantive input.” Declining to draft is not the same as warning. Referring to ICP-2 is not the same as telling the community that a process has entered sensitive RIR-governance territory. A constitutional boundary is not protected by quietly pointing to a document. It is protected by placing the warning on the record.

If ICANN knew that CAIGA had entered the Blueprint, and if ICANN knew CAIGA concerned RIR governance, the duty was not to wait. The duty was to say, publicly and clearly:

This process cannot define AFRINIC governance. Any proposal affecting AFRINIC must come from, be tested by, and be supported through the AFRINIC community under ICP-2. Governmental or intergovernmental coordination cannot substitute for member authority or bottom-up legitimacy.

That statement should not have waited for controversy, it should have accompanied the moment of drift.

This is the point many observers missed. The problem was never only whether ICANN had co-authored CAIGA. The problem was whether ICANN corrected the misunderstanding while correction still had preventive value.

The question most observers failed to ask, including me, is not whether ICANN later clarified its position. It is not whether ICANN denied authorship. It is not even whether ICANN said CAIGA would still need to satisfy ICP-2. Those answers came after the architecture had already surfaced.

The missing question is more precise:

When did ICANN first realize that the Smart Africa process was touching AFRINIC-specific governance, and what did ICANN say in the room at that moment?

That is the threshold question.

If ICANN was involved in the Blueprint process only as a general Internet governance partner, then the moment CAIGA entered the text should have triggered a recorded boundary warning. If ICANN was asked for substantive input on CAIGA and declined, then that request itself confirms that ICANN knew the process had crossed from broad policy dialogue into RIR-governance territory. If ICANN referred Smart Africa to ICP-2, the next question is whether that reference was merely documentary or whether it was accompanied by a clear institutional warning that RIR governance could not be redesigned through governmental or intergovernmental architecture without prior community authorship.

This matters because governance failures often do not begin with a formal decision. They begin with an unrecorded moment of institutional hesitation.

Governance failures often begin long before a formal decision is taken. A participant sees the boundary shift; a funder notices the mandate expand; a facilitator recognizes that a capacity-building exercise is becoming constitutional architecture; a technical institution understands that political language is entering technical governance. Yet instead of recording the boundary at the moment of drift, the actors wait. They wait for a final draft, for a formal submission, for the community to object, or for controversy to mature.

By then, the process had already accumulated diplomatic weight. The proposal is no longer simply being discussed; it is being prepared for validation and is currently being validated, as per a copy email I got through a friend, which was scheduled for 14 April 2026 at 14:00–16:00 GMT.

That is precisely the stage now surrounding CAIGA and the Internet Governance Blueprint: not an abstract idea waiting to be born, but an architecture moving through consultation, presentation, and attempted normalization. Once a process reaches that phase, warning becomes harder, correction becomes more costly, and the affected community is no longer debating a blank page. It is being asked to react to a structure that has already learned how to stand.

This is why the warning question is more important than the authorship question. An author can be identified after the text is written. A warning must exist before the text becomes architecture.

ICANN has now placed a boundary position on the public record. Its 18 November blog and 24 November response made clear several points, including that ICANN’s support for the Smart Africa Blueprint should not be read as endorsement of CAIGA, and that any proposal affecting AFRINIC governance must satisfy ICP-2 and obtain support from the affected community. That clarification matters. But it came after CAIGA had already appeared inside the Blueprint, after Smart Africa presentations at ICANN83-84, and after community concern had already matured.

The missing record is therefore not a post-controversy clarification. The missing record is contemporaneous: when ICANN first realized that the Smart Africa process had crossed from general Internet governance dialogue into AFRINIC-governance territory, what warning did it issue then, to whom, and in what form?

If ICANN has a memorandum, email, meeting note, public reservation, or formal statement from that earlier moment saying, in substance, that ICANN support for the Smart Africa Blueprint did not extend to any proposal affecting AFRINIC governance unless developed through the AFRINIC community, tested under ICP-2, and subjected to open, bottom-up consultation before institutional or political advancement, it should publish that record.

If such a contemporaneous record does not exist, then the issue is no longer only CAIGA. It becomes a stress test of ICANN’s regional engagement model itself: whether that model contained, at the relevant time, a mandatory warning mechanism for ICANN-supported projects that drift into RIR authority.

This is the institutional failure hidden beneath the episode: not authorship, not endorsement, not intention, but warning.

Consultation After Architecture

ICANN’s response also encouraged those with questions about Smart Africa’s engagement with AFRINIC and its members to reach out to Smart Africa. It noted that Smart Africa was considering broader public consultation on the draft Blueprint, and ICANN said it would encourage that effort. ICANN also emphasized that sufficient time for engagement and meaningful treatment of contributions was essential for a strong multistakeholder-supported outcome.⁵

This is better than silence, but it is still insufficient.

There is an uncomfortable circularity here. ICANN funded the process. ICANN acknowledged that the process produced a Blueprint containing CAIGA. ICANN recognized that CAIGA raised questions under ICP-2. Yet affected community members were effectively directed back to Smart Africa for consultation, yes Smart Africa!.

That is facilitation without responsibility.

ICANN cannot fund the room, acknowledge that the room produced an RIR-governance proposal, then tell the affected community to ask the room’s convener for clarity. At minimum, ICANN owed the community its own boundary statement: what did ICANN fund, what did it not fund, when did it learn of CAIGA, what did it say when CAIGA appeared, and what safeguards did it require once RIR governance entered the project’s orbit?

The issue is not whether Smart Africa may consult. Of course, it may, selectively though. The issue is whether consultation after presentation cures the absence of community authorship before presentation.

Consultation is not authorship, review is not consent, and a meeting is not consensus. Feedback is not bottom-up governance if the architecture has already been drafted elsewhere.

The ICP-2 Trap

The most dangerous passage in ICANN’s response is the one concerning direct governmental involvement in RIR governance.

ICANN stated that if RIR communities wish to have the ability for an RIR to be governed in the way CAIGA proposes for AFRINIC, including direct governmental involvement in RIR governance separate from governmental participation as RIR members, then now is the time for such proposals to be raised and evaluated in the ICP-2 conversation.⁶

This sentence requires careful scrutiny.

On one level, ICANN may have intended this as procedural neutrality. Anyone can raise proposals in a community process. The ICP-2 revision is open. The community can debate what it wants. ICANN need not prejudge every idea before the process receives it.

On a 2^nd level, there is another reading, and it is dangerous.

This formulation risks treating direct governmental involvement in RIR governance as an ordinary design option for the ICP-2 process, rather than as a fundamental challenge to the very model ICP-2 exists to protect.

There is a difference between governmental participation and governmental structural involvement.

Governmental participation is part of multistakeholder governance. Governments can attend, speak, advise, object, contribute, and participate. On the RIR level, if they are members, they can vote. They are affected stakeholders, and on a wider scope, they have legitimate public-policy interests.

Governmental structural involvement is different. That is where governments, intergovernmental platforms, ministerial mechanisms, or heads-of-state processes acquire formal or quasi-formal authority over the governance structure, recommendation pathway, reform approval, or institutional direction of an RIR.

Smart Africa’s structure also matters because it is not simply an open technical community forum. Its public membership page lists private-sector membership categories with annual fees ranging from USD 5,000 for startups to USD 200,000 for Platinum membership. The same page identifies differentiated access benefits, including eligibility for Platinum members to sit at board meetings with Heads of State, eligibility for Platinum and Gold members to attend steering committee meetings with ICT ministers and policymakers, and access to Smart Africa members. None of this is inherently improper for a public-private development alliance. But it reinforces the point: a paid-access regional platform, even one with legitimate development objectives, cannot be treated as equivalent to the open, bottom-up, affected-community legitimacy required for RIR governance.

One can strengthen multistakeholder model, while the other can invert it.

ICP-2 should not become the doorway through which political supervision enters the RIR system. It should remain the test by which such supervision is examined and, where inconsistent with independence, openness, bottom-up legitimacy, and community support, rejected.

Is this African Specific?

This is not an African issue alone. If this logic is normalized in Africa, it can travel elsewhere. A European regulatory coalition could propose a supervisory framework around RIPE NCC. A regional governmental bloc in Asia could propose an oversight layer around APNIC. A state-security coalition could attempt to frame number resource governance as sovereignty infrastructure in Latin America or elsewhere. In every case, the language would sound familiar: coordination, stability, capacity building, regional voice, resilience, reform.

The danger is not the continent, but the method.

The method is simple: begin with broad Internet governance cooperation; embed institutional redesign inside a regional blueprint; present it through recognized platforms; rely on post-facto consultation; and when challenged, say that the affected community can discuss it later. That is not bottom-up legitimacy, nor how it is protected.

A Global Design Flow

The ICANN—Smart Africa—AFRINIC episode therefore, exposes a global design flaw: ICANN lacks a clear, public boundary protocol for regional partnerships that drift toward RIR governance.

The answer is not for ICANN to disengage from Africa or from any region; that would be a mistake. ICANN should engage, it should support capacity building, help communities participate in global Internet governance, support DNS security, technical training, policy literacy, and multistakeholder development. However,; ICANN needs cleaner engagement.

It needs a rule that when an ICANN-supported process begins touching the governance of an RIR, ICP-2 enters the room before politics writes the agenda.

The concern is not that ICANN must prejudge all proposals. The concern is that ICANN should not normalize direct governmental structural involvement as merely another option without emphasizing that ICP-2 is a protective test, not a neutral marketplace for redesigning RIR independence.

The issue is not whether communities may debate hard questions; they may. The issue is whether a proposal that structurally alters RIR independence should be framed first as an option or first as a risk to be tested against the architecture that made RIR recognition legitimate in the first place.