Summary

  • AFRINIC launched 6Mandela in Cairo on 13 December 2005 as a regional IPv6-awareness programme built around four modes of activity: meetings, training, operational research and transition test beds. Contemporary and follow-on records show workshops, instruction and continued use of the name, but they do not disclose a project charter, budget, accountable owner, partner obligations, quantified target, causal evaluation or closure report.
  • The proper test is neither whether IPv6 is technically functional nor whether AFRINIC was allowed to educate. It is whether a voluntary registry-led programme produced transparent value for operators without turning institutional recognition, meeting attendance or resource administration into a claim of authority over network architecture. AFRINIC was a number-resource ledger and coordinator, not a sovereign, regulator, police force, prosecutor, judge or punishment authority.

A launch whose ambition outran its public architecture

On 13 December 2005, in the Mena House Oberoi conference hall in Giza, Cairo, AFRINIC chief executive Adiel Akplogan presented and launched a project called 6Mandela. The setting was the first African IPv6 meeting, held alongside AFRINIC-3. Baher Esmat, chair of the Egyptian IPv6 Task Force, chaired the session, and Egyptian communications minister Tarek Kamel opened the event in the presence of participants drawn from the regional and international Internet community. The next day, AFRINIC-3 moved into its Public Policy Meeting.

That sequence is important: the programme launch and the policy meeting were adjacent, but they were institutionally distinct.

AFRINIC’s contemporary account described 6Mandela as an IPv6-awareness programme for its service region. It identified four ways of working: meetings, training, operational research and test beds for the transition process. It traced the idea to discussion at AFRINIC-2 in Maputo and to an IPv6 task-force meeting held in Cairo in May 2005. AFRINIC said it would work with interested parties during 2006 to finalise the project and directed participants to the afripv6-discuss mailing list, which offered an open channel for participation and general discussion of IPv6 deployment.

Those facts establish more than a slogan. A named institution, a named presenter, a precise date and place, a stated purpose, four modes of work and a proposed participation channel all appear in the record. Follow-on materials also show that the label did not vanish immediately after the conference. AFRINIC reported training and workshop activity in 2006, referred to the project in updates to peer registry meetings, and continued to present the name in early 2007. A fair assessment must begin by accepting that an institutional act occurred and that visible activity followed it.

But the same record stops short of describing a fully constituted programme. The launch account does not identify an approval resolution, a member vote, a Board decision, a signed charter or a defined line of responsibility below the chief executive. It does not state the programme’s budget, the source of its money, the value of in-kind support, the duties accepted by partners, a baseline for African IPv6 awareness or deployment, a target, a country list, a milestone calendar, an evaluation method or an end date.

Saying that AFRINIC would work with interested parties to finalise the project during 2006 makes the launch sound both real and incomplete: the public announcement preceded the public evidence of final institutional form.

This is not a semantic objection. A programme can start before every governance document is published, and a young regional registry in 2005 need not have presented itself like a modern grant-making body. Yet a launch, an activity stream and an auditable programme are not identical things. The available evidence supports the first two. It does not give the public enough information to reconstruct the third.

What AFRINIC did—and what it did not do

6Mandela was a technology-promotion act. It was not an IPv6 standard, a number-resource allocation decision, a policy proposal adopted by consensus at the 14 December Public Policy Meeting, a state regulation, a compulsory migration order or a sanction. No checked record says that an operator could be fined, denied service, lose a resource, have a licence revoked or suffer another penalty for declining to participate or deploy IPv6. Nothing in the launch transferred control of carrier or ISP architecture to AFRINIC.

That boundary follows from the institution’s actual function. AFRINIC coordinates and records unique Internet number resources and processes requests in its region. The ledger is operationally important: uniqueness and accurate records support networks that need to identify and use number resources without collision. But recording resources is not ownership of the networks that use them. It does not give the registry title to an operator’s routers, procurement budget, security systems, customer contracts or engineering choices.

The operator remains responsible for deciding what it deploys, paying for it, running it and absorbing the consequences when it fails.

The IANA recognition report published in April 2005 recorded AFRINIC’s recognition as a Regional Internet Registry and described objectives that AFRINIC had submitted, including Internet growth, efficient assignment procedures, education in technical and policy areas and public-policy development within areas of competence. That document supports the proposition that voluntary education and development work could sit within AFRINIC’s corporate purposes. It proves what IANA reviewed, recorded and recognised.

It does not grant sovereignty, legislative jurisdiction or punishment power, and it cannot convert an educational objective into authority to order a carrier’s protocol choices.

Likewise, a minister’s presence at the Cairo meeting demonstrated host-country support; it was not an African treaty or a continent-wide delegation. Discussion by members or conference participants showed interest; it was not legislation. A mailing list enabled participation; it was not a parliament. Support from an international technical forum, vendors, carriers or training organisations contributed expertise and visibility; it did not bind operators that were absent, silent or unconvinced. Official AFRINIC, IANA and peer-registry records prove what those institutions said and did.

Their official character does not make every institutional claim a legal fact about wider authority.

That distinction protects useful coordination rather than diminishing it. AFRINIC did not need sovereign authority to convene engineers, share deployment experience, offer training, operate a voluntary test environment or publish research. It needed corporate capacity, resources, willing participants and honest boundaries. Calling those activities voluntary services sets a defensible foundation. Calling them the expression of a regional power to direct networks would not.

From a ledger service to an advocacy role

The institutional interest of 6Mandela lies in AFRINIC’s movement beyond receiving, processing and recording number-resource requests. A registry waiting for applications performs a bounded administrative function. A registry that promotes one technology transition becomes an active participant in shaping demand, expectations and investment. The second role may be legitimate and beneficial, but it raises questions that the first role does not.

An applicant that requests IPv6 asks AFRINIC to carry out a registry service. An operator attending a 6Mandela workshop may instead be evaluating whether to make a technical and capital commitment. In the first exchange, the registry handles a resource record. In the second, it frames a technology problem, selects instructors and partners, presents transition experience and may influence what engineers, executives or governments regard as the expected direction of travel. The institution therefore moves from administering a ledger to helping construct the case for a deployment choice.

That movement creates a duty of clarity. AFRINIC had to separate its authority to perform registry tasks from its capacity to persuade. Resource holders should have been able to tell whether programme costs came from mandatory registry-service charges, voluntary sponsorship, grants, partner contributions or some combination. Participants should have known whether advice was independent, vendor-supported or promoter-led. Operators should have been assured that eligibility, standing and service continuity did not depend on accepting the promoted architecture.

The public material located for 6Mandela does not provide that complete separation in documentary form.

Absence from the checked public record is not evidence that no internal record ever existed. A budget may have been prepared but not preserved online. Responsibilities may have been agreed in correspondence. A working group may have convened without an indexed archive. Procurement or sponsorship may have been documented privately. The responsible claim is narrower: the public records examined through the evidence cutoff do not disclose those elements, so outsiders cannot audit them and later institutional histories cannot safely assume them.

This matters especially because promotion can produce institutional benefits even before it produces operator outcomes. A registry can gain visibility, relevance and convening power. A vendor can gain exposure to potential customers. A specialist trainer can gain work and standing. A ministry can associate itself with modernisation. An international forum can extend its campaign. Those benefits are not evidence of misconduct. They are ordinary incentives, and precisely for that reason they should be named and separated from the return experienced by the operators who pay to deploy and maintain networks.

Four work modes, four different tests

The launch’s most concrete contribution was its division of the programme into meetings, training, operational research and transition test beds. The four categories describe plausible ways to lower information and experimentation costs. They do not, by themselves, specify deliverables or success.

Meetings can connect engineers who would otherwise solve similar problems alone. They can expose local operators to implementation experience and create professional relationships that persist after the event. Their immediate outputs can be counted: sessions held, people attending, countries represented, presentations delivered. But attendance cannot show that participants learned, that their organisations adopted a production service, that customers benefited or that the event caused any later decision. Nor does a crowd constitute regional consent.

To evaluate meetings, an operator-focused programme would need to know who attended, which operational questions they brought, what they took away, what follow-up occurred and whether their decisions differed from those of comparable non-participants.

Training can transfer practical capability. Hands-on instruction is more demanding than a promotional speech and may be particularly valuable for small networks that cannot maintain an internal laboratory or specialist curriculum. The 2006 record shows actual IPv6 training, including instruction delivered around AFRINIC-5. Yet a participant count remains an output. Evaluation would require measures of competence retained, production changes attempted, failures avoided, support demanded later and costs incurred.

A course can be well taught and still fail to justify a wider capital programme; it can also create value even when a participant reasonably decides not to deploy.

Operational research should produce questions, methods, evidence and findings that can be inspected. The launch named research but the checked public programme record does not reveal a 6Mandela research agenda, dataset, publication ledger or finding that can be tied to the project. Without those elements, the term signals intent more than measurable production. A rigorous research stream would distinguish facts collected by the registry from claims made by promoters, define the relevant operator population, disclose limitations and publish negative or null findings rather than only success stories.

Test beds can reduce the risk of experimenting on live customer networks. They can help engineers test routing, addressing, security, monitoring and compatibility before procurement or deployment. AFRINIC’s later institutional history says an IPv6 testbed was launched in November 2008, which is consistent with one of the four modes named in 2005. The retrospective does not, however, identify that testbed as a 6Mandela deliverable. It does not disclose a project funding chain, test design, user list, research questions, results or usage data.

The correct description is therefore a later activity compatible with the original idea, not proof that the original project delivered and evaluated its test-bed promise.

The four-mode structure was sensible as a programme outline. Its weakness was not that the activities were inherently empty. It was that the launch record did not attach a responsible owner, budget, dates, geographic scope, outputs, outcomes or completion criteria to each mode. A list of methods can direct activity; it cannot on its own allow members to decide whether the programme was worth its cost.

The antecedents show interest, not delegated rule

The April 2005 AFRINIC-2 meeting in Maputo included presentations about IPv6 deployment. Jordi Martinez Palet presented IPv6 as an opportunity for innovation and demonstrated remote IPv6 use. The second day’s agenda recorded the possibility of an IPv6 task force in the AFRINIC community. The report does not name 6Mandela, publish a charter or record a resolution adopting such a programme.

AFRINIC’s later launch account also cited an IPv6 task-force meeting in Cairo in May 2005. The checked sources do not give that meeting’s exact date, minutes, participant list, decision text or instrument of authority. These two antecedents reasonably explain why AFRINIC perceived an information and coordination need. They do not prove that a defined electorate approved a particular programme, voted a budget or authorised compulsory action.

In March 2006, AFRINIC reported to an APNIC member meeting that it was working on an extensive IPv6-awareness programme called 6Mandela. Akplogan’s presentation connected the work to an April 2005 member recommendation, noted that the Cairo event had taken place and said IPv6 modules had been added to the 2006 training plan. That plan contemplated bilingual delivery of as many as twelve training events a year. The number is informative but must remain in its proper category: it was a maximum in AFRINIC’s organisation-wide training plan, not a guarantee that twelve 6Mandela events occurred.

The distinction between recommendation and delegation is essential. Members can ask for information. Meeting participants can encourage cooperation. A task force can advise and organise. None of these actions makes the registry sovereign over the carriers, service providers, enterprises and public networks that operate independently across multiple jurisdictions. Even unanimous enthusiasm among those present could not lawfully substitute for the consent of every absent operator, and the public evidence does not provide a denominator or an approval vote in any case.

Partners, supporters and the missing obligations ledger

The Cairo event brought an extensive cast around the programme. Baher Esmat and the Egyptian IPv6 Task Force helped anchor the meeting. Tarek Kamel represented host-government support. Latif Ladid, chair of the Global IPv6 Forum, presented the related Khawarizmi-v6 initiative for Arab countries and offered support for IPv6 initiatives in Africa. Representatives associated with France Telecom, Teleglobe and 6DISS described experience and willingness to support 6Mandela and Khawarizmi-v6. African operational experience also appeared through participants including Alain Aina of TRSTech Togo and actors associated with TENET.

This network could produce genuine value. Regional operators might learn from deployments already attempted. International programmes could provide specialised knowledge. Vendors could contribute equipment, instructors or practical experience. Host institutions could reduce event costs and convene participants. None of that should be dismissed simply because contributors had their own interests.

But the evidentiary vocabulary must remain exact. An offer of support is not a contract. Event sponsorship is not a project budget. Delivering a course is not accepting long-term responsibility for outcomes. Presenting operational experience is not a commitment by other operators to follow it. Ministerial encouragement is not regulatory delegation. Appearance in a meeting report is not membership in a governing board.

The IPv6 Forum’s roadmap, dated May 2006, offers the most specific numerical rollout concept located in connection with the broader initiative. It described the Khawarizmi concept as extending toward Africa through 6Mandela and proposed agreement among carriers and ISPs, support from national ministries, a budget for tunnel brokers where required, and an initial core of two or preferably three countries. Those were the Forum’s programme-design claims. No located AFRINIC instrument establishes that AFRINIC formally adopted, financed or delivered each element, and the countries were not identified in the checked record.

The roadmap cannot be silently converted into AFRINIC’s charter.

A proper partner ledger would say who promised what, over what period, with what cash or in-kind value, subject to what conflicts policy and reporting to whom. It would distinguish a trainer from a sponsor, an adviser from a decision-maker, a host from a funder and a vendor from an evaluator. The public 6Mandela record names many useful relationships without supplying that institutional map.

Activity after launch, and the working group still to come

The programme’s follow-on record rebuts any claim that nothing happened. At AFRINIC-4 in Nairobi in May 2006, the chief executive’s activity report listed the launch of the IPv6 Mandela project, IPv6 training and participation in 6DISS. In November and December, AFRINIC-5 in Mauritius combined two days of IPv6 training, an IPv6 conference and the Public Policy Meeting. The meeting report placed AFRINIC, the Egyptian IPv6 Task Force, TENET, Jordi Martinez Palet and the Global IPv6 Forum around a 6Mandela workshop and described the aim as increasing IPv6 awareness and usage in Africa.

The accompanying press release reported that fifty delegates received hands-on IPv6 training from Cisco Systems and Consulintel. It said the full meeting drew 120 delegates from 35 countries. These are meaningful operational counts. They show that people travelled, instruction was delivered and a regional meeting assembled participants from a substantial number of countries. They do not show how many workshop attendees represented network operators, how many completed the training, what they learned, whether they later deployed IPv6, what deployment cost or whether similar changes would have happened without 6Mandela.

The same December 2006 release said a working group on the 6Mandela project would be set up to raise awareness. The announcement came at least 354 days after the launch. Future tense nearly a year later is evidence that formalisation remained in progress, but it is not proof that the preceding activity was imaginary or that no informal group existed. It creates a precise documentary question: was the group subsequently constituted, with a charter, roster, minutes and reporting duties? Those records were not located in the checked public corpus.

In March 2007, an AFRINIC update at APNIC 23 referred to the previous meeting as having shown the 6Mandela project. That establishes continued presentation of the brand. It does not provide a budget, project governance, key performance indicators or a causal result. The name remained visible; its institutional destination remains uncertain.

Later AFRINIC materials widen the time horizon but not the attribution. A ten-year history records more than 100 attendees at AFRINIC-3 overall, the first African IPv6 event hosted by the Egyptian IPv6 Forum during the meeting and the Afri-IPv6 mailing list. It also records the November 2008 testbed. A current AFRINIC resource page says the organisation has promoted and supported IPv6 deployment since 2005 through deployathons, e-courses, onsite training, webinars and a testbed. These are later institutional accounts of continuity.

They do not demonstrate that every subsequent activity belonged to 6Mandela, drew money from it or resulted from the 2005 launch.

Numbers that say less than they first appear to say

The AFRINIC-5 report stated that IPv6 allocations had increased from five to twenty-three. The difference was eighteen allocations; the later count was 4.6 times the starting count, equivalent to a 360 per cent increase over the base. The arithmetic is straightforward. Its interpretation is not.

An allocation is a registry action. It records that a resource was issued under the applicable process. It is not the same as a route announcement, a reachable network, customer traffic, service availability, an application working end to end or a reduction in IPv4 dependence. Even if every allocation led to routing, the report does not establish that exposure to 6Mandela caused the request. Policy or fee changes, independent operator plans, vendor readiness, other training, government initiatives and global technical developments could also affect demand.

The report does not define the exact comparison dates in the quoted summary or provide an applicant-level pathway from programme participation to allocation.

The attendance figures require the same discipline. Fifty trainees measure seats filled at a hands-on event. The 120 delegates from 35 countries measure the wider AFRINIC-5 meeting, not necessarily the 6Mandela workshop. Dividing delegates mechanically by countries would produce an average of roughly 3.43, but it would reveal nothing useful about representation because the country distribution and relevant operator denominators are unknown. More than 100 attendees at AFRINIC-3 similarly describes the whole meeting, not the launch session, a vote or support for the programme.

Measurement errors accumulate when unlike categories are collapsed. Allocation, assignment, route announcement, traffic share, service availability, staff competence, meeting attendance and awareness all describe different states. A programme can increase awareness without producing deployment. It can improve competence without changing a firm’s business case. It can lead to an allocation that remains unrouted. It can help launch a service while also increasing the cost and complexity of maintaining compatibility. No single count captures the whole chain.

A credible evaluation would begin with a baseline and define the object of change. If the objective were awareness, it would measure knowledge before and after participation. If it were operational competence, it would use demonstrated tasks and later retention. If it were deployment, it would define production service, route visibility, traffic or another observable outcome. If it were economic value, it would count the costs of equipment, staff, security, support and continued IPv4 coexistence alongside any benefit.

It would also use a comparison or other method capable of separating programme influence from selection: organisations already inclined to deploy are also more likely to attend voluntary training.

No project-level evaluation of that kind was located. That absence does not prove 6Mandela failed. It means the evidence cannot sustain either a causal triumph narrative or a causal failure allegation.

Operator value is the proper centre of gravity

For participating networks, the strongest case for 6Mandela was the prospect of reducing information and experimentation costs. Shared instruction could make specialised knowledge available to operators that could not build a curriculum alone. Meetings could connect African engineers with peers who had encountered similar compatibility and operational problems. Research could identify regionally relevant constraints. A test bed could allow controlled experimentation away from customer traffic. These are plausible public-service benefits, and the programme’s voluntary form respected the operator’s ultimate choice.

The cost side is equally material. During a transition in which IPv4 compatibility remains necessary, an operator may have to maintain two operational environments. That can mean added routing policy, security controls, monitoring, logging, incident response, staff training, procurement and support. It can expand the number of configurations and failure surfaces before any older system can be retired. The evidence does not supply a universal cost or prove that every IPv6 investment is uneconomic.

It does establish why allocation and attendance counts cannot serve as a substitute for asking who paid, what burden persisted and what measurable benefit followed.

This is the central agency problem in a registry-led promotion programme. AFRINIC, technical promoters, vendors, trainers and public hosts could influence the direction while operators bore much of the implementation downside. Their incentives were not necessarily opposed. A vendor’s expertise might be essential; a registry might genuinely seek regional capability; an operator might welcome help. But aligned rhetoric is not proof of aligned economics. Transparent sponsorship, procurement, cost incidence and outcome measurement are the institutional tools that convert shared enthusiasm into accountable cooperation.

Funding is therefore not a minor administrative detail. If voluntary sponsorship paid for an event, that should be visible along with any conditions. If partners contributed instructors or equipment, the in-kind value and role should be stated. If registry-service fees supported discretionary technology promotion, members should be able to see the amount, authorisation and expected return. The checked record does not establish that member fees funded 6Mandela, much less that any such funding violated corporate rules. It leaves the incidence unknown.

Resource-access neutrality provides the necessary firewall. AFRINIC could accurately explain how to request IPv6 and could offer voluntary support to organisations choosing it. It could not use its position in the resource-request path to imply that access, standing, fees or continuity depended on adopting a promoted architecture. No 6Mandela record examined here shows such a penalty or condition. Stating the firewall explicitly would nevertheless have strengthened the programme because it would have made the boundary durable rather than implicit.

The strongest case for 6Mandela

The best argument in the programme’s favour is substantial. AFRINIC was a newly recognised regional registry responding to an information and coordination gap identified in its own community. Its recorded corporate objectives included education and Internet development. Operators participated voluntarily. The programme proposed practical tools—training, operational research and testing—rather than coercion. It attracted experience from African operators, international technical actors, trainers, vendors and a host government. Follow-on reports show real activity.

Allocations increased, training occurred, workshops were held, a testbed appeared later and AFRINIC continued supporting IPv6 in multiple formats.

On that account, demanding a complete evaluation system on launch day risks judging a young 2005 initiative by later administrative expectations. A programme can create value through relationships, confidence and knowledge that are difficult to quantify. It may be reasonable to begin with interested parties, learn by doing and formalise as the work matures. The record’s lack of coercion also matters: networks could accept useful help without surrendering the right to decline deployment.

That case establishes permission to educate and evidence of plausible value. It does not establish measured effectiveness or broad authority. Indeed, accepting the case makes the missing records more consequential, not less. If the project developed through learning, a later charter, budget, working-group record and evaluation should reveal what was learned. If the allocation increase reflected programme influence, an exposure and causation analysis should support that claim. If the testbed fulfilled the launch promise, a traceable project record should connect them.

If partners supplied value, a contribution ledger should distinguish help from governance.

The balanced conclusion is therefore not that 6Mandela was illegal, useless, corrupt or captured by vendors. No located evidence supports those allegations. Nor is the conclusion that protocol deployment has no technical benefit or that AFRINIC lacked the capacity to offer voluntary education. The evidence supports a narrower assessment: a real initiative generated visible activity and may have built capability, while its publicly auditable authority, funding, goals and results remained materially incomplete.

The record that an accountable programme needed

A proportionate charter would first have defined the programme as voluntary education, research and coordination. It would have stated that participation and deployment had no effect on unrelated number-resource eligibility, standing, fees or continuity. That single sentence would separate the registry service from the promotional role and prevent later rhetoric from laundering coordination into mandate.

Second, it would have identified a responsible owner and governance structure. Adiel Akplogan was the public presenter and institutional voice, but the record does not name a project director, budget holder, steering group or evaluator. A charter should have listed advisory participants, decision rights, conflicts and the reporting chain, while confirming that partners and meeting attendees could not bind non-participating operators.

Third, it would have published a funding ledger. Cash, sponsorship, staff time, equipment and donated instruction should be distinguishable. Mandatory registry-service revenues, if used, should be separated from discretionary contributions. A partner’s commercial interest would not disqualify its help; disclosure would allow operators to judge the advice and compare the programme with alternative uses of funds.

Fourth, it would have separated outputs from outcomes. Meetings held, people trained and experiments conducted are outputs. Skills retained, production services launched, routing observed, traffic carried, failures avoided and costs reduced are outcomes. IPv6 allocations are registry outputs with an uncertain relationship to either operational deployment or economic value. Each category needs its own definition.

Fifth, it would have specified baseline, target, calendar and geography. The Forum’s external proposal of an initial two- or three-country core demonstrates what a concrete rollout concept might look like, but it is not proven to have been AFRINIC’s adopted plan. A public programme document should have named any target countries, explained their selection, identified operator denominators and set milestones for training, research and test-bed use.

Finally, it would have closed the record. A completion or succession report should explain whether 6Mandela ended, became another programme, transferred assets or simply ceased using the name. It should connect—or explicitly decline to connect—the 2008 testbed and later support formats to the original project. It should report negative and null results as well as activity. Without that end point, institutional memory can absorb every later IPv6 service into an origin story that the contemporaneous evidence does not prove.

A bounded verdict

6Mandela marked a genuine change in posture. AFRINIC did more than stand ready to record IPv6 requests: it used its convening power, brand and relationships to encourage knowledge, experimentation and deployment. That could be a legitimate corporate service. It could also shape investment choices whose costs and risks fell principally on independent networks. The difference between responsible service and mandate laundering lies in voluntary choice, a transparent control and funding structure, and measurements centred on operator value.

The public record shows the launch, declared purpose, four work modes, intended collaboration, open discussion channel and later activity. It shows support, instruction, meeting participation and an increase in allocations. It does not show a formal approval instrument, a project-level budget, binding partner commitments, a quantified deployment target, a causation method, an evaluation or a closure record. Those gaps bar any confident claim that 6Mandela produced the allocation increase, delivered the later testbed or drove Africa’s subsequent IPv6 activity.

Most importantly, nothing in recognition by IANA, support by a ministry, discussion among members or participation by technical organisations elevated AFRINIC into a sovereign. The registry had no police, prosecutor, judge or punishment power over network architecture. Its legitimate tools were invitation, education, research, testing and coordination. Operators retained the right and responsibility to decide.

The evidence supports neither celebration without accounts nor condemnation without proof. It supports a more useful finding: 6Mandela was a plausible voluntary intervention whose public ambition is easier to document than its programme architecture or incremental effect. The lesson is not that registries must remain silent. It is that the further a ledger institution moves into technology advocacy, the more clearly it must show who authorised the work, who controlled it, who paid, who benefited, what changed and what remained the operator’s choice.