Summary

  • RIPE NCC's Jelena Cosic reported on 2 September 2026 that a flip-chart experiment at RIPE 92 helped several researchers connect with operators. The account does not quantify sustained collaborations or production deployments.
  • Reusable data-sharing arrangements, supporting tools, a contextualised dataset repository and recurring matchmaking are discussion ideas, not a launched service. A follow-up conversation is being organised for RIPE 93.
  • The useful next unit is a mutually accepted question. Finding a counterpart, permitting an evidence exchange and accepting an operational dependency remain separate decisions.

A sheet of paper can make a problem visible without first making it a project. That was the modest attraction of a flip chart placed in the meeting area at RIPE 92. Researchers could advertise questions; other participants could browse, add ideas and look for somebody worth talking to. No central repository had to be filled before an introduction could happen.

In a 2 September retrospective, Jelena Cosic, identified as RIPE NCC staff, describes several useful connections. The invitations were primarily directed at researchers. The account calls the exercise a first step, not a finished solution. Its result is evidence of contact, not a measured conversion rate from conference conversation to maintained infrastructure.

Two sides of the question

Operators were not merely invited to be data donors. The RIPE 92 opening-plenary transcript also records an invitation for them to put problems or datasets they wanted investigated on the boards, followed by a Thursday matching session. That matters when interpreting a retrospective which chiefly describes researchers finding help. It would be wrong to infer that operators had no invitation to set questions themselves.

The two-way exchange is where the next experiment could become more useful. An operator knows which decisions are actually available, what cannot be changed and how much engineering time a test can consume. A researcher can explain which uncertainty a method might resolve. Neither contribution is reducible to supplying files or supplying cleverness.

The first shared task could therefore be smaller than a project plan: agree what decision an answer might inform, name somebody on each side who can explain the relevant constraints, and arrange the next conversation. This is an editorial proposal for the continuation, not a description of a RIPE requirement. It would test whether two parties have accepted the same question before either assumes the other has accepted a larger job.

Discovery need not mean custody

The retrospective sets out possible ways to reduce the friction of sharing data, including reusable terms, extraction and anonymisation tools, and a repository carrying context, limitations, contacts and usage conditions. Those ideas could save repeated work. They cannot determine whether one particular exchange is useful or permitted.

A public question can lead to a suitable contact without exposing that person's raw records. A catalogue could describe material while leaving access to a separate agreement. An initial exchange might establish whether a limited aggregate is enough, or whether analysis within the provider's environment is feasible. These are design options, not verified features of a RIPE platform or assurances of legal compliance.

Cosic explicitly acknowledges that addressing legal requirements does not necessarily resolve commercial and security concerns. Starting with the question can help keep the first exchange proportionate. Moving a large collection before the parties know what they want to learn may increase both work and exposure without improving the match.

A meeting ticket is not a support agreement

The RIPE 92 RACI page describes tickets, travel, accommodation and presentation opportunities for selected researchers. These lower the cost of attending and being heard. They do not, by themselves, establish who will support a piece of software afterwards.

The retrospective is not blind to that distinction. It specifically contrasts research prototypes, simulations and proofs of concept with operators' need for maintained products, implementation guidance and continuing support. Matchmaking can make that discussion possible. It must not make it seem settled merely because an engineer liked a presentation.

Nor should an old funding programme be silently attached to the new discussion. RIPE NCC announced in April 2025 that its Community Projects Fund would end, while existing 2024 recipients' work would be unaffected. That does not establish the end of all research support. It does mean this retrospective is not evidence of a new call under that fund.

What the next meeting can learn

Cosic says a research–operator collaboration BoF is being organised at RIPE 93. The meeting website confirms Sofia, Bulgaria, on 26–30 October 2026. The agenda is still labelled a draft; the consulted material does not establish a precise time for this discussion.

It would be premature to demand a production-adoption score. Some useful research explains an observation, rules out an approach or establishes that a question is not worth pursuing further. Treating every academic connection as a product-development contract would narrow the very exchange the experiment seeks to encourage.

The more revealing question is which problems both parties accepted, what allowed work to continue and what caused it to stop. The retrospective supplies no such outcome count. That is a limit on what can be concluded, not evidence that the trial failed. A paper board was useful because it promised little. Its successor could retain that virtue while making the work after an introduction less likely to disappear between two inboxes.