Summary
- The Court of Civil Appeal set AFRINIC’s 28 September 2023 appeal aside after holding that the person who purported to act for the company lacked the required corporate authority and that the lawyers who lodged and appeared in the appeal lacked locus.
- The respondent had also argued that permission from the Official Receiver or court sanction was missing, and that leave under section 3(2)(c) was required because, on the respondent’s case, the order was interlocutory.
- The Court expressly found it unnecessary to decide whether Official Receiver permission and first-instance leave were required. The judgment therefore supplies no general answer in either direction.
- Restoring the 12 September 2023 order and substituting a two-month election-completion period were parts of the disposition. They did not silently convert the reserved objections into holdings.
- The responsible institutional response is an appellate-authority matrix that separates proposed authority, permission, sanction, leave, order classification, filing requirements, holdings and unresolved issues instead of treating one failed threshold as proof about the others.
The appeal that never reached the next two doors
There are judgments in which every disputed issue seems to arrive at the same destination. This was not one of them. In African Network Information Centre (AFRINIC) Ltd v Cloud Innovation Ltd & Another, 2024 SCJ 473, the Court of Civil Appeal confronted several objections to an appeal lodged in AFRINIC’s name. The appeal had been filed on 28 September 2023, after a sequence of corporate and receivership events that made the identity of the company’s lawful decision-maker central. Four preliminary objections were advanced.
But a court does not have to decide every objection merely because counsel has briefed and argued it.
The Court first resolved an anterior question: was there a valid corporate actor behind this appeal, and did the lawyers before the Court have standing to lodge and conduct it in AFRINIC’s name? Its answer was dispositive. It held that Benjamin Eshun’s three-year directorship had expired by effluxion of time on 18 September 2023. It also held that the litigation instruction contained in a 2021 resolution operated through the then-chief executive, Eddy Kayihura, who had ceased to hold office in November 2022.
On the Court’s analysis, Eshun lacked authority to institute and prosecute the appeal, and Messrs Moollan and Mardemootoo lacked locus to lodge it and appear for AFRINIC.
That was enough. Once the Court concluded that no valid appeal was before it, two further gates no longer needed to be tested. One concerned whether permission of the Official Receiver was required before filing. The other concerned whether leave of the first-instance Judge was required under the statutory route invoked by the respondent. The Court did not choose between the parties’ positions on either question. It said it was unnecessary to deal with them in light of the decision already reached.
This sequence matters beyond the names in the caption. Any organisation attempting to challenge a receivership order can encounter several procedural controls at once: the identity of the corporate actor, a live appointment or delegation, the position of a receiver, possible judicial sanction, the classification of the order, a statutory leave route, the filing deadline, security and the effect of an appeal on the order below. Those controls are not interchangeable. Failure at the first can stop the proceeding before a court tests the second or third.
A stopped case may therefore produce a firm result for the parties while leaving a general procedural question open.
That is exactly the discipline this judgment requires. The appeal was set aside. The order below was restored. Yet the two further questions remained unanswered. The disposition was real; the missing answers were real too.
Four columns, not one blended story
The safest way to read 2024 SCJ 473 is to divide the record into four columns: submissions, holdings, non-holdings and disposition. Most confusion begins when a statement is moved from one column into another.
The first column contains what the respondent argued. Cloud Innovation objected that receiver consent or court sanction had not been obtained before AFRINIC’s appeal was lodged. It also contended that leave was required under section 3(2)(c) of the Court of Civil Appeal Act 1963 because the appealed order was interlocutory. Those propositions identify live objections. They do not become law merely because they appeared in a judgment’s account of a party’s case. Nor does the record available here justify an independent conclusion that the order was interlocutory.
That characterization belonged to the respondent’s argument, and the Court did not decide it in resolving the appeal.
The second column contains the holdings that did the work. The Court held that Eshun lacked the authority needed to institute and prosecute the appeal. It held that the lawyers lacked locus to lodge and appear in it in AFRINIC’s name. Those findings meant that there was no valid appeal before the Court. The authority defect came first in the logical sequence, even though other objections could have presented substantial questions in another properly constituted proceeding.
The third column is the express non-holding. Having reached its anterior decision, the Court found it unnecessary to determine whether permission from the Official Receiver and leave from the first-instance Judge were required before the appeal was lodged. “Unnecessary” is not a coy synonym for “yes.” It is not a coded “no.” It marks the point where the Court stopped deciding.
The fourth column is the disposition. The appeal was set aside. The Court made no costs order in the particular circumstances. It restored the 12 September 2023 order and substituted a two-month period for completion of the directed election work. Those orders explain what changed procedurally after judgment. They do not migrate the two untouched questions into the holdings column.
This four-column method resists a common narrative shortcut. A commentator sees that an appeal failed, notices that receiver permission and statutory leave had been disputed, and writes as if the failure proved the absence of permission or leave was fatal. The reverse shortcut is equally unsound: because the Court did not rely on those objections, a commentator writes that neither permission nor leave mattered. Both versions erase the judgment’s architecture. The appeal failed on the authority and locus route. The Court did not test the other routes.
An accurate account can be concise without becoming vague: the respondent raised the permission and leave objections; the Court did not decide them. Any longer analysis should add context, not change that status.
Why a non-holding is more than silence
An express non-holding differs from an accidental omission. The Court did not simply forget that other objections existed. It identified them and explained that the holdings already reached made it unnecessary to decide them. The open status is therefore part of the judgment’s deliberate structure.
That structure places a boundary around legitimate inference. It tells future readers where the Court’s public authority ends. Courts routinely decide cases on grounds that make other issues unnecessary, premature or unsuitable for determination. This can serve several institutional purposes. It avoids pronouncing on questions without need. It confines judicial reasoning to a dispute capable of affecting the result. It preserves questions for a record in which the correct parties, facts and procedural posture bring them squarely forward.
Whatever the broader theory, the practical reading instruction is simple: an issue expressly not decided remains undecided by that judgment.
The boundary works symmetrically. A party who favors strict receiver control cannot cite 2024 SCJ 473 as a holding that permission was required. A party who favors an independent corporate path of appeal cannot cite it as a holding that permission was unnecessary. A party who describes the underlying order as interlocutory cannot use the judgment’s disposition as judicial confirmation of that description. A party who believes no statutory leave should have been needed cannot transform the Court’s refusal to reach the objection into approval of that view.
This symmetry is not evasiveness. It is legal accuracy. A judgment can be decisive for the proceeding and incomplete as a source of general rules. The appeal ceased to exist in the form presented because the Court found no valid corporate authority behind it. That conclusion did not require a theory of all receiver-era appeals.
Indeed, the facts of a future appeal could be materially different: a current board might act, a live director might hold a clear mandate, the receiver might give an instruction, a court might grant or refuse sanction, the order might be classified in a particular way, or a statutory leave application might be made and ruled upon. None of those possible configurations receives a general answer from this judgment.
The open status should survive repetition. A summary published a year later does not acquire more authority than the source it summarizes. Ten confident paraphrases cannot turn a non-holding into precedent. Institutional memory must preserve the empty cells instead of filling them with whichever proposition becomes convenient.
The receiver-permission question, kept in its proper frame
The first untouched gate concerned the Official Receiver. The respondent’s objection alleged that consent from the receiver or court sanction had not been obtained before the appeal was lodged. The surrounding receivership explains why the objection mattered: the first-instance order had appointed the Official Receiver and directed work concerning AFRINIC’s governance. But context does not supply the missing ruling.
There is an understandable administrative case for requiring a clear record of the receiver’s position in litigation carried on in a company’s name. A receiver charged with preserving and administering a company’s affairs may need to know which proceedings expose the estate to expense, security obligations or conflicting instructions. A documented permission or refusal can make representation transparent. Judicial sanction, where ordinary law makes it relevant, can provide an independent record when the company, office-holder and proposed appellant disagree.
These considerations help explain why lawyers would litigate the gate.
They do not establish that this gate legally applied in the way the respondent claimed. The judgment does not give that conclusion. It also does not determine whether receiver permission, court sanction, both, or neither was required in every possible appeal from a receivership order. Treating the receiver’s control as self-proving would be especially hazardous when the proposed appeal challenges an order connected to the receiver’s own appointment or powers. The law may or may not provide a path around that apparent circle in a given posture; 2024 SCJ 473 did not say.
The distinction between receiver permission and court sanction deserves care as well. They may be discussed together in an objection, but they identify different deciding actors. A receiver’s recorded position is not itself a judicial ruling. Court sanction is not merely a synonym for the receiver’s consent. If a later case reaches the point, the record should identify exactly what was requested, from whom, under what instrument, by which proposed corporate actor, and with what result. Compressing all of that into “receiver approval” would hide the control surface rather than clarify it.
Nor should the receiver’s appointment be treated as transferring sovereignty over Internet number resources. The office-holder operated within a legal mandate concerning a private company. The public authority behind that mandate came from Mauritian law and the Mauritian court. The distinction matters because technical importance can make an administrative office appear larger than its legal source. AFRINIC’s registry function may affect many operators, but the institution remains a private member-based technical coordinator. A receiver administering it does not become an Internet legislature.
The correct statement about this appeal is consequently narrow: the permission issue was raised, and the Court left it open after deciding the anterior authority defect. Anything further must come from another competent legal source, not from inference.
The statutory-leave question, also left untouched
The second untouched gate concerned leave from the first-instance Judge. The respondent invoked section 3(2)(c) of the Court of Civil Appeal Act 1963 and argued that leave was required because, in its submission, the order was interlocutory. That is the verified status of the proposition: it was an objection advanced by a party.
The judgment does not authorize a general account of the provision’s meaning. It does not independently settle the full statutory test. It does not decide the legal classification of the 12 September order for this purpose. It does not tell readers that the first-instance Judge’s leave was required, and it does not tell them that such leave was unnecessary. An article that supplies any of those answers would be writing a different judgment.
Order classification can be consequential. A rule governing appeals from interlocutory orders may screen issues until the court below has completed its work, prevent fragmented proceedings, or channel urgent challenges through a leave application. But labels such as “final” and “interlocutory” cannot responsibly be assigned from intuition about how important an order feels. A structurally significant order can still raise a technical classification question, and the correct answer may depend on the precise instrument, the relief granted, the remaining proceedings and the statute being applied.
Those are matters for competent legal determination on a proper record.
There is also an access concern on the opposite side. Receivership orders can alter who controls a company, who may instruct lawyers and how quickly governance work proceeds. A company or other legally entitled actor needs a lawful route to test an order said to affect it. If that route includes leave, the identity of the applicant, the timing and the available reviewing court must be clear enough to use. If it does not include leave in a particular posture, that conclusion must likewise come from law, not convenience. The high stakes make a real ruling more important, not less.
The statutory-leave question should therefore remain in the same neutral state as the receiver-permission question. It was presented. It was not reached. Its appearance in the judgment makes it visible; the express non-holding prevents its conversion into ratio.
The strongest orderly-administration case
The two reserved objections should not be trivialised merely because they were not decided. Each points toward a plausible concern about orderly administration.
Start with litigation authority. A company under receivership may face competing instructions from former office-holders, remaining directors, members, a receiver or lawyers acting under older mandates. Proceedings filed without a traceable source of authority can consume estate resources, require security, complicate settlement and force counterparties to spend money identifying who actually speaks for the company. A permission record can reduce that ambiguity.
If judicial sanction is the appropriate mechanism under the governing law, a public order can identify the actor entitled to proceed and the scope of the authority granted.
Then consider statutory leave. Appellate systems often distinguish among orders and procedural stages for reasons connected to timing and case management. A leave route may allow the first-instance Judge to address whether an immediate appeal should proceed before the underlying matter is complete. It may also clarify whether the order continues to operate, whether a stay has been requested, what security is due and which issues are actually ready for appellate review. A documented leave application and ruling can create a clean bridge between the court below and the appellate court.
In combination, these controls may protect a company from unauthorised litigation while also protecting the courts from an appeal that arrived through the wrong door. In the AFRINIC setting, where governance decisions can affect a technical registry’s relationships with members and network operators, clear authority records can support continuity planning. Operators may not be parties to the corporate dispute, but they still need to assess whether instructions are stable, whether a public order is in force and whether an announced challenge has legal traction.
This is the strongest benign case for taking the two questions seriously in a future proper proceeding. It explains why permission, sanction, classification and leave should be documented rather than waved away. It does not prove that either disputed requirement applied to the 28 September 2023 appeal. Good institutional design and positive law are related, but they are not the same thing. A sensible control cannot be assumed to be a legal precondition unless a competent source makes it one.
The strongest concern about a circular barrier
The opposite concern is equally serious. A company subject to a receivership order must have some lawful avenue through which an eligible actor can challenge that order. If permission must always be obtained from the office-holder whose appointment or powers are being challenged, the remedy could become circular. The gatekeeper would appear to control access to review of the gatekeeper’s own position.
That concern does not itself prove that receiver permission was unnecessary here. Ordinary law might supply a judicial-sanction route, define an independent applicant, distinguish between kinds of proceedings, or provide some other answer. The important point is that the apparent circle makes the architecture of the remedy a question worthy of careful adjudication. It cannot safely be resolved by institutional preference. A receiver’s desire for orderly administration does not create the rule; a company’s desire for unrestricted review does not erase a rule that law actually provides.
The same tension applies to statutory leave. A leave requirement can prevent fragmented appeals, but it can also delay review of an order with immediate control effects. Whether the order is classified in the relevant statutory category, who may seek leave and which judge decides are not rhetorical questions. A sound system needs a usable route, a known deadline and a record of the decision. It also needs a way to handle refusal, urgency, security and any application for a stay.
The strongest challenge to the respondent’s proposed gates, then, is not a declaration that the gates did not exist. It is a demand for a non-circular, lawfully specified remedy. The strongest response is not a declaration that the gates always exist. It is a demand for orderly, traceable authority. Both positions describe institutional needs. Neither position can claim victory from this judgment’s non-holding.
That distinction is valuable because public debate often treats a plausible policy concern as if it were already law. The receiver side can point to preservation and coherent administration. The challenging side can point to meaningful review and avoidance of circular control. A court applying the right instruments to a properly presented case must convert those competing concerns into a legal answer. Until then, the honest ledger has two unresolved entries.
Restoration was not an answer by implication
One tempting shortcut remains. Because the Court set the appeal aside and restored the 12 September 2023 order, a reader may assume that the restoration validated every procedural objection connected to the receiver. It did not.
A disposition tells the parties what happens as a consequence of the holdings the Court actually made. Here, the appeal was not validly before the Court because of the authority and locus findings. Setting it aside removed the appellate proceeding in that form. Restoring the earlier order followed from that result. The Court also substituted a two-month period for completion of the election-related direction and made no costs order in the particular circumstances.
None of those steps required the Court to decide whether a differently authorised appeal would have needed receiver permission or first-instance leave.
An implicit answer would be particularly difficult to defend when the judgment expressly says the questions were unnecessary to decide. Express language beats narrative inference. A reader cannot use the remedy section to overwrite the Court’s statement about the limits of its own reasoning.
The point can be tested with a simple counterfactual. Imagine that the appeal had been lodged by a corporate actor whose appointment and delegation were accepted as valid. The anterior defect would disappear. The Court might then have needed to examine the receiver’s position, possible sanction, the order’s classification and the statutory leave objection. The result of that examination cannot be known from 2024 SCJ 473. The counterfactual does not predict a win for either side. It only shows why the questions were logically downstream of the defect that ended the actual appeal.
Restoration also does not make the receiver a source of public law. The Court’s order had executable force because it was the act of a state court. The Official Receiver’s mandate derived from the lawful instruments governing the appointment. AFRINIC’s operational role did not elevate its internal preferences into legal commands. Each actor stayed within a distinct layer: registry operations, corporate authority, receivership administration and judicial decision.
Keeping those layers separate is the difference between reporting a judgment and building a myth around it.
Public authority and the private bookkeeper
AFRINIC’s regional function can make a corporate dispute sound constitutional. The organisation helps maintain number-resource records and coordinates services important to networks across Africa. NRS describes the member-registry context in which that work occurs. Yet scale and dependence do not turn a registry into a sovereign.
AFRINIC is a private member-based technical registry, bookkeeper, service provider and coordinator. It has no legislative, police, prosecutorial, punitive, confiscatory or public-law adjudicatory power. It cannot create Mauritian appellate law through a policy statement, a community process or the perceived indispensability of its database. Nor can a constituency around it fill a judicial silence by consensus.
Heng Lu’s distinction between reality layers is useful here because it identifies what can actually bind whom. Narratives about regional stewardship, community legitimacy or technical necessity may influence public debate. Executable legal authority, however, comes from courts, statutes, contracts and valid corporate instruments. In this event, the Mauritian court was the sovereign legal authority. It applied ordinary law to the proceeding of a private company. It did not sit as an Internet-policy council, and AFRINIC did not appear as a government.
The bookkeeper principle is not an insult to registry work. Bookkeeping can be operationally critical. Reliable records, neutral service and continuity are precisely what network operators need during institutional conflict. But the importance of the ledger makes legal accuracy more important: a registry must not use technical leverage as a substitute for authority, and commentators must not give it powers it does not possess. Protect the ledger; do not mythologise the keeper.
The same restraint applies to the Official Receiver. The role may carry substantial control within the scope of a court order and ordinary law. It does not confer global Internet sovereignty. Any answer about permission to appeal must be traced to the relevant legal source and ruling. The title of the office cannot answer the question by itself.
BTW’s broader reporting on AFRINIC’s litigation, Mauritius company law and the receiver supplies context for why these lines became contested. Those accounts help readers see the collision between a technical institution’s public significance and its private legal form. But context remains bounded. A general account of the Companies Act does not decide these two objections. A study of the Official Receiver’s operational role does not prove a permission requirement. A litigation overview does not transform an unresolved issue in one judgment into a settled rule.
The result is a clear hierarchy of sources. NRS helps identify what AFRINIC is. Heng Lu helps identify what it is not and where executable authority resides. LARUS helps explain why uncertain governance instructions can matter to operators. BTW supplies the wider corporate and receivership setting. The judgment alone supplies the holdings, non-holdings and disposition in this appeal. None of the contextual sources may be used to complete a sentence the Court deliberately left incomplete.
The economic cost is uncertainty, not an invented outage
Legal uncertainty can be costly without causing a technical incident. That distinction is important in a registry dispute, where dramatic claims about Internet stability can outrun the evidence.
The record does not prove that the non-holding caused an outage, a route change, an RPKI change, a registry-record alteration, customer loss or quantified economic damage. It would be irresponsible to attach any such event to the judgment without evidence. There is no need to do so. The credible economic mechanism is the cost of planning around an unresolved appellate pathway.
Consider a company, receiver, member, counterparty or service provider evaluating a future challenge. If the source of corporate authority is unclear, lawyers must trace appointments, board terms, delegations and old resolutions. If the receiver’s permission may be relevant but no decision states when or why, parties may prepare parallel requests to the receiver and the court. If order classification is unsettled, they may brief both final-order and interlocutory-order possibilities. If statutory leave might apply, they must assess timing, applications and the consequence of refusal.
If the effect of filing on the order below is uncertain, they may also seek a stay and arrange security.
Each layer creates work. Lawyers duplicate analysis. Decision-makers reserve time for applications that may prove unnecessary. Counterparties ask for opinions, board records, receiver correspondence and court orders. Insurers, lenders or suppliers may want assurance that the person giving instructions is authorised. Technical teams develop continuity plans for more than one governance outcome. None of that proves a collapse of service. It describes diligence cost.
Delay can amplify the cost. An appeal filed through the wrong corporate actor may consume time before the court reaches any substantive challenge. A later, properly authorised proceeding may have to begin again, subject to whatever deadlines and procedural gates the law actually imposes. Security deposits or costs exposure may need funding. Uncertainty about a stay can force parties to plan for the order’s continued operation while also preparing for the possibility of review.
Public statements that misdescribe a non-holding can make matters worse by causing stakeholders to rely on a rule that no court announced.
LARUS’s operator-facing perspective is relevant at this point, but only within its evidentiary boundary. Governance uncertainty at an RIR can increase diligence and continuity-planning burdens for organisations that depend on stable registry administration. That context explains why a clean authority record is not merely a lawyers’ preference. It does not show that 2024 SCJ 473 changed a route, broke infrastructure or caused a particular loss.
The proportional response is therefore documentation, not alarm. Record which actor proposed to appeal. Record the current corporate authority. Record the receiver’s actual position. Record any sanction or leave application and the deciding court. Record the status of the order, filing, security and stay. Most importantly, record which issues were decided and which remained open. Better information reduces the uncertainty premium without pretending that uncertainty has disappeared.
Why future cases must begin from their own authority record
A later receiver-era appeal should not be treated as a replay whose outcome can be copied from this case. The foundational facts may differ. A board could have been elected. A director’s term could be current. A delegation could identify the litigation and the lawyers. The receiver could request, approve, oppose or decline to act. A court could grant sanction. A judge could classify the order and rule on leave. The filing could fall inside or outside a deadline. Security and stay questions could be separately resolved.
These are not predictions about what will happen. They are variables. Their presence shows why the judgment’s narrowness must be preserved. The Court’s authority holding was tied to the corporate route actually used for the 28 September appeal. Its non-holding means a future court must examine any properly presented receiver-permission or leave objection on the record and law before it. The prior set-aside does not automate the answer.
Future reporting should also distinguish the person who makes a legal claim from the actor who decides it. A company may assert authority. A receiver may state a position. A respondent may object. A first-instance Judge may grant or refuse leave. An appellate court may classify an order or review a ruling. A journalist may explain the sequence. Only the competent legal decision-maker supplies the binding determination within the relevant proceeding. Attribution is not pedantry; it is the map of authority.
That map becomes essential when public communications compress several steps into a single phrase such as “the court rejected the appeal for lack of receiver permission.” That sentence would be wrong here. The Court rejected the validity of the appeal on corporate-authority and locus grounds, then did not decide the receiver-permission question. A similarly inaccurate phrase—“the court allowed appeals without receiver permission”—would be wrong for the opposite reason.
The correct vocabulary is available: alleged, submitted, held, declined to decide, set aside, restored. Those verbs keep the actors and effects aligned.
Member Briefing
Deeper Profile Context
Sign in with the right membership level to unlock the full briefing and source notes.
Only for Strategic Circle
Strategic Circle
Open to all readers. Unlock profile briefings after joining and signing in.
Join Strategic CircleOnly for Leadership Alliance
Leadership Alliance
For qualified IP-asset owners and management; sign in to unlock alliance briefings.
Join Leadership Alliance
