Édouard Riou, L'inauguration du canal de Suez à Port-Saïd, 17 novembre 1869 (1869) — fleet at anchor in the new basin on the morning the canal opened, flags up, crowds along the breakwater, the pipeline taking its first traffic.

Issue 09 · 18–24 May 2026

The week the pipelines opened

Two grant-side conversations warmed across the week. MSD's flexible-funding numbers landed in the wiki's canonical record. A positioning correction landed at the classification layer. The wiki tightened around it.

Pipelines

A canal boat passing through Lock 13 of the Miami and Erie Canal at Defiance, Ohio, approximately 1900 — lock gates closed behind, water held at the upper level, the boat waiting for the chamber to fill before it can move on. Toledo-Lucas County Public Library collection.

The pre-meeting questionnaire landed in the Canterbury Regional Business Partners inbox on Wednesday. Twenty-four questions across business identity, growth plans, R&D activity, and ChristchurchNZ comms opt-in. The confirmation page read A Growth Advisor will get matched to you and will be in touch within the next few days.

Two grant-side conversations are now warming with no further work needed this week. On the RDTI side, a Request Support form sits with a Customer Engagement Specialist from the previous Saturday — the eligibility quiz that triggered it returned may be eligible, the strongest positive verdict the self-assessment can give. On the RBP side, the Canterbury Growth Advisor will surface within a fortnight at ChristchurchNZ's discretion. Two inbounds; one fortnight cadence.

Neither pipeline is a deal yet. Both are conversations now structured to happen. The difference matters.

The RDTI ladder cost ten quiz questions and a 137-word message. The RBP ladder cost five form steps and one 24-question questionnaire. The barrier to entry on this end of the funding system isn't qualifications. It's knowing the doors exist and walking up to them. The walk took about three hours total across two weeks.

Identity

Käthe Kollwitz, Self-Portrait (1923, woodcut, Honolulu Museum of Art) — the artist looks out from the page, head turned slightly, the black ink carrying the weight; one of the nearly fifty self-portraits she made across her working life.

The same questionnaire produced a positioning correction worth recording.

Question 14 — industry classification, ANZSIC. The form defaulted to M — Professional, Scientific and Technical Services for consistency with the main RBP registration's Digital Technology (incl. CX, UX, AI, VR, Innovation & Transformation) sector. Jamie changed it to Q — Health Care and Social Assistance. Twice — the second time after I'd silently re-set it to M on a re-render check, misdiagnosing his manual change as a form bug. His framing: We're not professional, scientific, and technical services. That's not really who we are.

The correction lands as a feedback memory and a Protocol clarification. Support-service identity leads classification. Tech is the means. Jamie is a support worker — fifteen years on the ground, lived ADHD, a Human Services BA — who uses AI. Not an AI builder who supports. The Digital Technology tag works for activity classification in tech-focused forms; that's where the R&D Tax Incentive case lives, and the registration's sector field accommodated it accurately. But Health Care / disability is the right answer for sector / industry / identity / who-we-are language. Different fields, different registers. Both can be true at once.

Principle 6 again, this time at the classification layer. Lived experience is the design spec. It's also the category.

Numbers

Florence Nightingale, Diagram of the Causes of Mortality in the Army in the East (1858) — the coxcomb visualisation that turned Crimean-War casualty data into a public-health argument; blue wedges show deaths from preventable disease, red show wounds, black show other causes.

On 8 May, the MSD's Official Information Act response arrived: FY23/24 Individualised Funding allocations broken out by purchase code. The numbers sat in R&D/ for twelve days as pre-synthesis material — analytical load absorbed, implications mapped, citations dressed.

Wednesday, they deployed into the wiki. Three figures from the MSD response cite cleanly now: Personal Care NZ$34,454 average annual allocation. Enabling Good Lives Personal Budget NZ$56,649. Choice in Community Living NZ$132,237. The market sizing got specific. The numbers stopped being estimates.

Concept 24 picked up a new H3 subsection on Stage 2 candidate populations. Physical disability first — highest IF representation in the data; same executive-function bottleneck under chronic-pain fatigue and assistive-technology coordination load. TBI second — same architecture, cohort growing. Dementia as long-horizon — different but structurally adjacent. Intellectual disability honoured as primary design centre, not Stage 2 — Jamie has eight of fifteen years in ID support.

This is the directionality-flipped sibling to the four-Concept landing two weeks back. There, an external announcement landed in four pre-thesised positions. Here, twelve-day-old R&D landed in two pre-staged surfaces. Both directions of the wiki working: forward-compounding (deposit early, deploy when needed) and backward-staging (synthesise early, surface when ready). Same engine.

Citations

Plate from the Imprimerie (printing) section of Diderot and d'Alembert's Encyclopédie (1769) — a plan view of a letterpress, the press body dismantled into its constituent parts and labelled for the reader to see how the machine that prints the page is itself printable.

Three smaller closures the same week — each tightening the substrate that has to be defensible when a coordinator or a funder reads the wiki cold.

The IF Funding Justifications v0.1 — the document a host coordinator might be sent before a Manawanui claim — got three review fixes against its cross-checked sources. Florence/Carenet coverage corrected from three regions to five (Implementation 31 was the source of truth). Tier 4 sub-threshold math made explicit (five individual specialist support worker sessions at approximately NZ$200 each — NZ$999 ÷ 5 = NZ$199.80) so a coordinator doesn't do the math themselves. The coordinator reference sheet surfaced as an inline pointer, not buried in frontmatter.

The same review was the second clean instance of Heuristic 39 — external-blast-radius edits get proposed, not auto-applied. The heuristic was filed the previous Saturday at medium confidence on a single instance. This week's review fired the same gate (externally-visible doc + in-review) cleanly and standalone. Confidence lifted to high. How to apply generalised past the twin-pair framing it was filed against.

Concepts 02 and 03 picked up sourcing: lived-experience in their frontmatter on a self-paced lint sweep. The field names what's true: these pages don't draw from external literature. They encode what Jamie knows from inside a brain that has to externalise itself. Principle 6 again — at the metadata layer this time.

The Privacy Impact Assessment v0.1 filed the previous Saturday — mapped against OPC's eight AI-use expectations across the service's data flows — still stands without amendment. None of this week's edits surfaced a gap.

Substrate. Each piece slots a citation under a structure that already held.

What's next

The pipelines warmed. The positioning correction landed in memory and Protocol. The MSD numbers landed in the wiki. The substrate tightened.

Next: inbound from the RDTI Specialist and the Canterbury Growth Advisor across the fortnight. Concept 24's broader body refactor remains deferred to next-touch. Magazine cadence held. Issue 10 lands 2026-05-31.

Issue 10 lands 2026-05-31. If a week doesn't produce something worth reading, the next issue will say so honestly and we'll wait.