Updated: 2026-06-17
Pages
- Home
- Contact
- Blog — recent posts and archive
- Research
- Computer Science
- Climate Science & AI
- Medical Science
- Methodology
- National summary — cross-regional pattern rollup
- Solution space — interventions sorted by mechanism
- Auckland
- Wellington
- Northland
- Waikato
- Bay of Plenty
- Gisborne
- Hawke’s Bay
- Taranaki
- Manawatū-Whanganui
- Nelson
- Tasman
- Marlborough
- West Coast
- Canterbury
- Otago
- Southland
- Labs — playable static demos
- David Simmons biography
- Sitemap — this page
- XML sitemap — machine-readable, for search engines
At a glance
- Blog posts: 24
- Research branches: 4 (Computer Science, Climate Science & AI, Medical Science, Auckland)
- Auckland — Problems published: 3 (of an 11-section scope; the rest are in progress)
- Auckland — evidence graph: 15 drivers, 15 camps, 17 evidence claims, 19 cited sources
- Auckland — rendered pages: 61
- David Simmons biography: 5 pages, driven from structured data
How to cite
The Auckland research pages are the part of this site most likely to be cited. A recommended citation form:
Simmons, L. (2026). ‹Page title›. Luke Simmons — Research. Retrieved 2026-06-17 from https://lukesimmonsnz.kiwi/research/auckland/‹section›/‹subpage›/
For a specific Auckland subpage, prefer the version with the most recent Updated date on the page itself. The entity data behind each Auckland page lives under content/auckland/data/ in the site’s source repository; every factual claim on a page carries its Source through that data layer.
For blog posts, the canonical URL is the post page (/blog/‹slug›/) and the date in the slug is the publication date.
Copyright and license
© Luke Simmons. All rights reserved.
There is no open-source license on the site’s source code or its written content. The prose, biographical material on David Roy Simmons, and the Auckland research content are all under default copyright — no permission to copy, redistribute, modify, or republish is granted by default. Quotation for the purposes of commentary, criticism, or research is usually permitted under the ordinary fair-dealing rules of New Zealand copyright law, but that is a reader judgement, not a license. If you would like to reuse any part of this work beyond fair dealing, please ask first via the contact form.
A note on AI assistance
Some content on this site is produced with assistance from large language models. Two channels are in use.
- Blog — a two-stage local-agent pipeline running on the site owner’s hardware via Ollama (
qwen2.5:14b). A daily agent writes private working notes from Hacker News and arXiv cs.AI; those notes are not published and stay on disk. Once a week, usually Sundays, a weekly agent synthesises the week of notes plus that week’s top HN and arXiv stories into a single public blog post with inline citations to the original sources. Posts signedagentare agent-authored; posts signedlukeare hand-written. - Longer-form prose — including research pages under
/research/and documentation such as the README and architecture notes — is drafted with Claude AI via Anthropic’s hosted API. Prompts and drafts transit Anthropic’s infrastructure under their terms.
What “edited” means varies by surface, and is disclosed on the page itself. The three research branches under Computer Science, Climate Science & AI, and Medical Science are AI-drafted reference notes that the site owner has not edited; they are published as-is for orientation, not as authoritative references. The Auckland research branch is AI-drafted in iterative dialogue with the site owner, and its typed entity graph enforces structural source citations on every factual claim — but independent per-claim fact-checking by the owner is a step that should sit between drafting and publication and is not yet formally in place.
Privacy and tracking
The site does not add its own tracking. The ingress layer does some of its own, and that is disclosed here honestly rather than glossed over.
- No site-owned analytics. No Google Analytics, Plausible, Fathom, or any other analytics service is loaded by the site.
- No site-owned third-party scripts beyond Google Fonts. No ad networks, no tag managers, no pixels, no beacons.
- No accounts, no logins, no newsletter. The site has no user database.
- Cloudflare is in front of the site as the public ingress while the owner transitions to dedicated self-hosting hardware. That means every request you make is visible to Cloudflare at the edge (they terminate TLS), and Cloudflare sets its own
__cf_bmbot-management cookie on responses by default. Cloudflare also retains request logs (IP, user-agent, URL, status) according to their own retention policy. This is Cloudflare’s processing, not the site’s, and is governed by Cloudflare’s privacy policy. - Fonts are loaded from Google Fonts via a
<link>tag. Your browser will make a request tofonts.googleapis.comandfonts.gstatic.comfor stylesheet and font files. Google may log that request in accordance with their own policies. - The contact form writes submissions to a flat file at the origin (
data/messages.jsonl). Once the site is publicly live, submissions will also be emailed to an address on this site’s own domain (@lukesimmonsnz.kiwi), which the site owner reads via ordinary mail forwarding. The submissions are not sent to any third-party form-handling service (Formspree, Basin, etc.). The only person who reads contact-form submissions is the site owner. - Origin server logs (URL, status code, timestamp) are retained on the origin machine for debugging. They are not shared. They will continue to exist once the origin moves off Cloudflare’s proxy.
Accessibility
The site aims to be readable and navigable without assistive technology, and usable with it:
- Keyboard navigation. Every link and form control is reachable by tab order. The contact form validates with visible text, not colour alone.
- Landmarks. Each page uses a single
<main>region, a top<nav>, a section-level subnav where relevant, and a<footer>. - Breadcrumbs on every content page indicate location within the site hierarchy.
- Headings follow a strict
h1→h2→h3order; no heading levels are skipped for visual effect. - Typography. Body text and headings are set in Inter (the operator format, site-wide since 2026-06-11). It loads from Google Fonts; if it fails to load, the site falls back to the system’s default sans-serif faces without loss of content.
- Contrast. Text and background meet WCAG AA contrast at normal body size; this is a goal, not an audit claim.
- Images on biography pages carry descriptive
alttext. Placeholder SVGs for figures under construction are labelled as such. - No JavaScript is required to read any content on this site. A single script adds
§anchor links to prose headings as a convenience; disabling it does not hide content.
If anything on this site is inaccessible to you, please say so via the contact form.
About this document
This page is regenerated from the site’s source whenever a new blog post is published. It reflects the site as of the Updated date at the top. Historical versions are not retained.