Summary

  • RIPE’s 1989 design gave a coordinating forum no executive authority over participating networks. Blokzijl’s chairmanship worked through meetings, public records, recognised consensus and delegation rather than ownership or command.
  • The proposed RIPE NCC separated continuous registry, database and secretariat work from volunteer coordination while leaving network operation with member organisations. RIPE and the RIPE NCC were related but never interchangeable.
  • Blokzijl’s informal selection and twenty-five-year tenure also exposed a succession debt. After his 2014 handover, the community documented the Chair’s function and built a repeatable NomCom process with terms, continuity and recall.

A title designed around a missing power

Fourteen people attended the first RIPE meeting in Amsterdam on 22 May 1989. They were not assembling a corporation, acquiring a backbone or conferring a licence. They were trying to coordinate IP networks that already belonged to different research institutions and operators. Europe’s network environment was fragmented not only by geography but by protocol choices, incumbent carriers and institutional mandates. A common packet format did not erase those boundaries.

The RIPE Terms of Reference, published as RIPE-001 on 29 November, gave the gathering a deliberately narrow centre. RIPE would exchange technical information, develop expertise, promote interconnection, establish agreement on common network-management practice and make its documents public. Participation was encouraged for all parties operating wide-area IP networks. The objective was ambitious: the operation and expansion of a pan-European IP network.

Then came the brake. RIPE was not a network service provider. The cooperating IP networks remained under the executive authority of their respective organisations. That clause is the constitutional fact behind Blokzijl’s title. A Chair could help networks agree on routing practice; the Chair did not become the employer of their engineers. A meeting could document what responsible operation required; it could not turn a participant’s fibre, router or budget into RIPE property.

An obsolete 1990 requirements document makes the original distribution of responsibility even clearer. Network services were provided by individual organisations, coordination was voluntary and best-effort, and interconnections ran over member-owned links under bilateral or multilateral agreements. The document should not be treated as current policy. Its value is historical: early coordination was expected to sit above an infrastructure it did not possess.

Four tasks, not one founder

The twenty-fifth-anniversary timeline says Blokzijl was appointed RIPE Chair in 1989. A later RIPE account says colleagues “volunteered” him; the eventual selection procedure says the first Chair was selected informally. Those descriptions are compatible at the level that matters. The evidence does not show a modern election, a fixed term or a pre-existing appointing authority. It shows a small operational community recognising someone who could keep its work moving.

Even then, the work was divided. The 1989 timeline names coordinators for connectivity and routing, network management and operations, the Domain Name System, and formal coordination. Blokzijl coordinated the last of these. Daniel Karrenberg, Thomas Lenggenhager, Francis Dupont and others occupied adjacent roles. A biography that puts every cable, route and database under one man would destroy the organisational design that made the work possible.

Blokzijl brought experience from high-energy physics networking at Amsterdam institutions, CERN and Nikhef. That background mattered because scientific networks had already taught their operators that a useful connection crosses budgets, institutions and technical cultures. It did not make him the designer of every European IP network. His comparative advantage was the ability to keep specialists working inside a shared frame.

The Chair’s instruments were consequently indirect. Meetings placed incompatible assumptions in the same room. Task forces turned a diffuse problem into bounded work. Minutes and RIPE documents made an agreement available to people who had not attended. A recognised Chair could tell when conversation had become a usable common position, communicate it and delegate its consequences. None of those instruments emitted a packet. Together they changed which packets independent organisations were prepared to carry.

When volunteering stopped being continuity

The limitation of a volunteer forum appeared quickly. RIPE-019, dated 16 September 1990 and co-authored by Blokzijl, Y. Devillers, Daniel Karrenberg and R. Volk, proposed a Network Coordination Centre. Its diagnosis was not that volunteers lacked knowledge. RIPE’s existing structure was well suited to development work but less suited to recurring administration. Day-to-day tasks needed dedicated technical staff.

The proposal did not simply centralise the network. It drew another boundary. RIPE would define the NCC’s tasks, the NCC would report to RIPE, participating organisations would join its work, and RIPE would review its operation. Some tasks would be centralised; others would remain distributed. The centre would not provide end-user support or operate member networks. Those duties stayed with the operations teams that possessed the infrastructure and responsibility. RIPE-019 grouped the prospective work into a delegated registry, network-information management and general operational support.

RIPE-035, the 1991 activity plan, showed why paid continuity mattered: maintaining a database, checking entries, exchanging data with other centres, mapping connectivity, keeping operational contacts, testing name-server relationships, collecting statistics, maintaining tools and supporting meetings all had different cadences. A quarterly gathering could decide that a record should exist. It could not update that record every week without someone accountable for the routine.

The plan also treated reporting as part of the service. Weekly, monthly, quarterly and annual outputs would make recurring work inspectable. That is a subtle shift from charisma to institution. A trusted convener can ask whether a task is being done. A staffed centre can leave dated evidence that it was done, identify an outstanding problem and be reviewed against an activity plan.

A centre embedded in a community

The RIPE NCC was formally established in April 1992 under the initial legal framework provided by RARE. The official history notes an important sequence: distribution of IP address space was added in 1992; it was not among the centre’s original activities. The institution that would be remembered as the first Regional Internet Registry began with a wider coordination and information problem rather than appearing fully formed as an address allocator.

By January 1998, the RIPE NCC had become a legal entity in its own right. RIPE did not follow it into incorporation. The later accountability report describes RIPE as an open community of participants, without legal personality or formal membership, while the RIPE NCC is a membership-based not-for-profit association under Dutch law. One develops community consensus; the other employs staff, holds contracts and implements services and policies.

The distinction protects both sides. If the community and centre were synonyms, employees could be mistaken for the source of policy and community speakers could be mistaken for corporate officers. If they were unrelated, the centre could lose the operational mandate that justifies its work. “Embedded” captured the intended tension: close enough for community control and reporting, separate enough for employment, budgets and continuous execution.

Blokzijl stood near that seam but did not erase it. He helped create RIPE and the RIPE NCC; he was not thereby the executive controller of every organisation, the owner of the number registry or the operator of its constituent networks. RIPE-508’s later account of the Registry keeps responsibility distributed between the RIPE NCC and resource holders. Registry authority is a maintained evidence system, not a personal asset carried by a founder.

The cheap efficiency of unwritten rules

For twenty-five years, the Chair role rested heavily on recognised practice. This can be highly efficient. A small community does not need a paragraph for every behaviour when participants know who listens, who records the room accurately and who will be challenged if they overreach. The lack of an election calendar is not proof that the community lacked consent. RIPE’s own sources describe the 2014 transition as well supported.

But personal trust does not reproduce itself automatically. A newcomer cannot audit a shared memory. An outsider cannot distinguish an accepted custom from a convenient claim. A successor cannot know which actions belong to the role and which belonged only to the predecessor’s temperament. The longer a founder performs an uncodified function well, the easier it becomes to postpone explaining how that function can survive the founder.

Blokzijl announced in October 2013 that he would step down after RIPE 68. In May 2014 he handed the Chair to Hans Petter Holen and asked that the community establish a selection process. The important point is not that the handover failed. It is that an accepted transfer still did not constitute a reusable rule for the next transfer.

Turning a person into a procedure without erasing the person

The accountability review later treated documentation as a form of public legibility. Informal norms could work internally while remaining hard to demonstrate to newcomers. It recommended progress on a Chair replacement process and questioned how external-facing activity should be reported. The concern was not a finding of personal misconduct. It was that institutional confidence should not require access to twenty-five years of communal memory.

RIPE-714 first described the role. Its one-sentence core is modest: the Chair ensures that the RIPE community functions well. The practices beneath it include meeting agendas, creation and closure of useful working groups and task forces, consensus about formal procedures, communication of results, reporting and delegation. It also supplies a sharp external limit: RIPE is not formally represented in other organisations, so its Chair does not formally represent RIPE anywhere.

RIPE-727 then described selection. A Chair and Vice-Chair team would serve five-year terms, subject to a two-term limit. A nominating committee randomly drawn from community volunteers would consult, select candidates and submit them for a procedural confirmation by the RIPE NCC Executive Board. Continuity and recall received explicit paths, while a pragmatism clause allowed documented deviations for unforeseen circumstances.

This was not an administrative rejection of Blokzijl. It was the completion of his final delegation. The community extracted functions that had been carried in one person, placed them in a public role description and made succession possible without having to reproduce his biography.

The authority was in the receipt

RIPE’s early system worked because agreement produced artefacts. A terms-of-reference document stated the boundary of power. An activity plan told staff what recurrent work the community expected. Reports allowed participants to inspect execution. Database entries and contact records gave operators a common operational reality. Meeting records let absent organisations see how a conclusion had been reached.

The Chair’s authority was therefore less like a command key than a receipt chain. It connected an open discussion to a recognisable consensus, that consensus to a working group or service task, and the result back to a public record. The Chair could not compel a network to implement the result. Yet an operator could act because the process made other operators’ likely actions legible.

That is why the separation between RIPE and the RIPE NCC is more than an organisational footnote. Consensus without execution becomes a meeting habit. Execution without a community mandate becomes administrative discretion. The institution needed both, plus a visible interface between them. Blokzijl’s work occupied that interface for unusually long; the later succession procedure made the interface less dependent on him.

What the record cannot prove

The surviving documents do not reconstruct every discussion at RIPE 1 or identify a formal vote that selected Blokzijl. “Appointed,” “volunteered” and “selected informally” should not be inflated into procedural precision. Nor does a twenty-five-year tenure by itself prove capture, absence of dissent or universal approval.

The documents also do not measure how often Blokzijl intervened, how many outcomes changed because of him, or whether another Chair would have produced the same European network development. Official memorials describe influence and service, not a controlled causal experiment. The article can identify institutional functions and documented boundaries; it cannot allocate every result to one individual.

Finally, formal procedure does not eliminate reliance on judgment. A NomCom still interprets community support, a Chair still reads consensus, and the pragmatism clause still permits deviations. The improvement is not automation. It is that authority, continuity and exception now leave a clearer trail.

Sources